Systems and methods for patient-specific therapy recommendations for cardiovascular disease - Patents.com

Abstract

Provided herein are methods and systems for making patient-specific therapy recommendations for patients with known or suspected cardiovascular disease, such as atherosclerosis.

Description

[Technical field]

[0001] Claiming priority This application claims the benefit of U.S. Provisional Patent Application No. 63 / 209,164, filed June 10, 2021, and U.S. Patent Application No. 17 / 693,229, filed March 11, 2022, the entire contents of which are incorporated herein by reference.

[0002] Federally funded research or development This invention was made in part with Government support under the National Heart, Lung and Blood Institute of the National Institutes of Health (HL126224). The Government has certain rights in this invention.

[0003] The present disclosure relates to methods and systems for making patient-specific therapy recommendations for patients with known or suspected cardiovascular disease, such as atherosclerosis. [Background technology]

[0004] Myocardial infarction (MI) and ischemic stroke (IS), the main consequences of unstable atherosclerotic lesions, are the most common causes of death worldwide (World Health Organization (WHO). Cardiovascular diseases (CVDs) Fact Sheet, 2017, available online at who.int / en / news-room / factsheets / detail / cardiovascular-diseases-(cvds), 23 April 2020). Guidance for the prevention of MI and IS is currently based on the effectiveness of treatments at the population level.

[0005] According to the World Health Organization (WHO), cardiovascular diseases (CVDs), including coronary artery disease and lower limb artery disease, are the leading cause of death and disability worldwide (The Atlas of Heart Disease and Stroke, WHOrganization, Editor, 2014), which is mainly due to unstable atherosclerosis resulting in myocardial infarction and ischemic stroke worldwide (World Health Organization (WHO). Cardiovascular diseases (CVDs) Fact Sheet, 2017, 23 April 2020, available at who.int / en / news-room / fact sheets / detail / cardiovascular-diseases-(cvds)). Despite the revolutionary advances in new treatments over the past 30 years, CVD still imposes a disproportionate economic cost (Bloom et al., The Global Economic Burden of Noncommunicable Diseases, WE Forum, Editor. 2011: Geneva), costing the US economy alone $320 billion annually (Mozaffarian et al., Heart Disease and Stroke Statistics-2015 Update: A Report from the American Heart Association. Circulation, 2015. 131(4): p. e29).This is exacerbated by aging and changing racial demographics (Gierada et al., Projected outcomes using different nodule sizes to define a positive CT lung cancer screening examination. Journal of the National Cancer Institute, 2014. 106(11): p.dju284; Warner, J. Stroke Costs Reaching Trillions: Without Action, Financial Costs of Strokes to Reach $2.2 Trillion by 2050. Stroke Health Center 2006 (cited 14 November 2014); available at https: / / www.webmd.com / stroke / news / 20060816 / stroke-costs-reaching-trillions), affecting a greater proportion of people globally as economic development continues to narrow the gap between people in developed and developing countries.

[0006] In the United States, the American Heart Association (AHA) predicts that more than 9% of adults have a significant (greater than 20%) risk of an adverse event within 10 years, and more than 25% have a moderate risk (Association, AH, AHA STATISTICAL UPDATE Heart Disease and Stroke Statistics-2018 Update. Circulation Journal, 2018. 137). This creates 23 million high-risk patients and 57 million people at moderate risk.Of these, approximately 30 million people in the United States are currently taking statin therapy to prevent new or recurrent CV events, and almost all of the 16.5 million people currently diagnosed with CVD are taking maintenance pharmacotherapy (Ross, G., Too Few Americans Take Statins, CDC Study Reveals. American Council on Science and Health, 2015; Vishwanath, R. and LC Hemphill, Familial hypercholesterolemia and estimation of US patients eligible for low-density lipoprotein apheresis after maximally tolerated lipid-lowering therapy. Journal of Clinical Lipidology, 2014. 8: p. 18-28; Herper, M. How Many People Take Cholesterol Drugs? Forbes, 2008; Pearson et al., Markers of Inflammation and Cardiovascular Disease: Application to Clinical and Public Health Practice: A Statement for Healthcare Professionals From the Centers for Disease Control and Prevention and the American Heart Association. Circulation, 2003. 107(3): pp. 499-511).

[0007] According to the WHO, stroke accounts for 10% of all causes of death worldwide, causing at least 5.5 million deaths per year (The Atlas of Heart Disease and Stroke, WH Organization, Editor. 2014). Of the approximately 800,000 strokes per year in the United States, 87% are ischemic, and approximately 15% of all strokes are preceded by a transient ischemic attack (TIA) (Writing Group, M., D. Mozaffarian et al., Heart Disease and Stroke Statistics-2016 Update: A Report From the American Heart Association. Circulation, 2016. 133(4): p. e38-360; Bruce Ovbiagele, Stroke Epidemiology: Advancing Our Understanding of Disease Mechanism and Therapy. Neurotherapeutics, 2011. 2011(8): p. 319-329). Many ischemic stroke events are caused by atherosclerosis (Barrett et al., Stroke Caused by Extracranial Disease. Circ Res, 2017. 120(3): p. 496-501). In the United States, 2.3 million patients are thought to have clinically significant stenosis (>50%), of which 19% have stenosis of 70% or more (de Weerd et al., Prevalence of Asymptomatic Carotid Artery Stenosis in the General Population: An Individual Participant Data Meta-Analysis. Stroke, 2010. 41(6): p. 1294-1297).Stroke also incurs huge costs to society, which amount to between $36.5 billion (Go et al., Heart Disease and Stroke Statistics-2014 Update: A Report From the American Heart Association. Circulation, 2014. 129(3): p. e28-e292) and $74 billion (D.L. Brown et a., Projected costs of ischemic stroke in the United States. Neurology, 2006) annually, estimated to reach $2.2 trillion by 2050 (PTINR.com-Staff $2.2 trillion stroke cost projected. 2006; Brown et al., Projected costs of ischemic stroke in the United States. Neurology, 2006. 67(8): p. 1390-1395).

[0008] According to the WHO, "coronary heart disease is now the leading cause of death worldwide. It is increasing and has become a veritable pandemic that does not respect national borders" (The Atlas of Heart Disease and Stroke, WH Organization, Editor. 2014). Of the approximately 1.2 million coronary attacks per year in the United States, approximately 66,000 are new, approximately 305,000 are recurrent, and approximately 160,000 are silent myocardial infarctions (MI) (Writing Group, Mozaffarian et al., Heart Disease and Stroke Statistics-2016 Update: A Report From the American Heart Association. Circulation, 2016. 133(4): p. e38-360; Bruce Ovbiagele, Stroke Epidemiology: Advancing Our Understanding of Disease Mechanism and Therapy. Neurotherapeutics, 2011. 2011(8): p. 319-329). Coronary heart disease, caused by atherosclerosis, is the most common type of heart disease, causing 365,914 deaths in 2017 (Benjamin et al., Heart Disease and Stroke Statistics-2019 Update: A Report From the American Heart Association. Circulation, 2019. 139(10): p. e56-e528).

[0009] The relative risk levels for various degrees of obstruction remain unclear, with some reports seeming to support the idea that clinically nonobstructive coronary artery disease (CAD) actually poses a higher risk of plaque than more obstructive plaques, and others suggesting that stenotic plaques have a higher event rate (Chang et al., Coronary Atherosclerotic Precursors of Acute Coronary Syndromes. JOURNAL OF THE AMERICAN COLLEGE OF CARDIOLOGY (JACC), 2018. 71(22); Gaston A. Rodriguez-Granillo et al., Defining the non-vulnerable and vulnerable patients with computed tomography coronary angiography: evaluation of atherosclerotic plaque burden and composition. European Heart Journal-Cardiovascular Imaging, 2016. 2016(17): p. 481-491; Ahmadi et al., Do plaques rapidly progress prior to myocardial infarction? The interplay between plaque vulnerability and progression. Circulation research, 2015. 117(1): p. 99-104; Bittencourt et al., Prognostic Value of Nonobstructive and Obstructive Coronary Artery Disease Detected by Coronary Computed Tomography Angiography to Identify Cardiovascular Events. Circulation: Cardiovascular Imaging, 2014. 7(2): p.282-291; Virmani et al., Pathology of the Vulnerable Plaque. JACC, 2006. 47(8): p. C13-8; FD Kolodgie et al., Pathologic assessment of the vulnerable human coronary plaque. Heart, 2004. 90; Virmani et al., Lessons from sudden coronary death: a comprehensive morphological classification scheme for atherosclerotic lesions. Arterioscler Thromb Vasc Biol, 2000. 20(5): p. 1262-75). .

Prior technical literature

[0010]

Patent Document 1

Patent document 2

Patent Document 3

Patent document 4

Non-licensed literature

[0011]

Non-licensed literature 1

Non-Patent Document 27

Non-Patent Document 28

Non-Patent Document 29

Non-Patent Document 30

Direct Entries 54

Direct Entries 55

Direct Entries 56

Direct Entries 57

Direct Entries 64

Direct Entries 65

Direct Entries 66

Direct Entries 67

[0012] Regarding available and future therapies for cardiovascular disease, there is a great need to help health care providers tailor therapy recommendations to specific patients rather than adopting a "one size fits all" approach. [Means for solving the problem]

[0013] The present disclosure provides a method and system for selecting and recommending suitable therapeutic treatment plans for patients with cardiovascular disease, such as atherosclerosis. For example, physicians and other healthcare providers can use the new method and system to analyze and process non-invasively obtained data, such as arterial imaging data, e.g., computed tomography angiography (CTA) data, from patients with atherosclerosis to obtain predictive proteomic and genomic information. Based on this information, various candidate therapies, e.g., drug therapies and / or surgical interventions, can be simulated based on their mechanisms of action in an in silico system biological model as described herein, enabling the healthcare provider to provide the patient with a report recommending one or more specific drug therapies and / or surgical interventions to be used for the patient's treatment.

[0014] The present disclosure also provides methods for obtaining proteomic and / or genetic information, as well as methods for constructing in silico systems biological models.

[0015] The in silico system organism model is initially generated or trained using two types of data. First, data is used that is empirically determined from biological samples from development subjects, people for whom actual proteomic data is available showing differentially expressed protein levels linked to the specific characteristics and morphology of plaque in each of those subjects. Second, results from searches of published literature, experimental results, and / or other databases are used to find papers etc. and obtain detailed information about the proteins in the model. These two sources of data are used to create the initial in silico system organism model.

[0016] The initial in silico system biological model is then updated with calibration data, such as omics data, from subjects to validate and refine the initial model. The calibration data is again based on actual specimens that exhibit differentially expressed proteins and / or transcript levels linked to the specific properties and morphology of the plaques of each of those subjects. This update of the initial model results in a calibrated in silico system biological model. Given calibration data from many subjects, this step ensures that the model works as intended and also augments the model to make it more robust.

[0017] During operation, the calibrated in silico system biological model is then updated with patient-specific personalized data based on imaging of the patient's plaque, again, but without the need to perform invasive blood tests or biopsies. The calibrated in silico system biological model is also updated with the predicted effects of two or more different therapies. The methods and systems described herein use the patient's non-invasively obtained data, e.g., imaging data, to make therapy recommendations based on an automatic comparison of two or more different therapies whose predicted effects have been programmed into the model.

[0018] Provided herein is a method of making a therapy recommendation for a patient with known or suspected atherosclerotic cardiovascular disease, the method comprising the steps of receiving non-invasively obtained data of plaque from the patient; accessing a systems biological model of atherosclerotic cardiovascular disease, where (i) the systems biological model represents a plurality of pathways associated with atherosclerotic cardiovascular disease and (ii) the systems biological model includes disease-related molecular levels for each molecule in the systems biological model; updating the systems biological model using personalized molecular levels derived from the non-invasively obtained data from the patient to generate a patient-specific systems biological model; obtaining information regarding one or more therapy candidates for the patient; updating the patient-specific systems biological model with information regarding an intended effect of each therapy candidate; simulating a therapy response to each therapy candidate in the systems biological model to obtain a simulated therapy effect for each therapy candidate; comparing the simulated therapy effects in the systems biological model before and after the therapy response simulation for each therapy candidate; selecting one or more therapy candidates as preferred therapies based on the comparison; and providing a report recommending the preferred therapy for the patient.

[0019] In some embodiments, simulating the therapy response comprises setting, in at least one network, a reduced molecular level for plaque instability and setting an increased molecular level for plaque stability.

[0020] In some embodiments, the molecules are genes, proteins, or metabolites, and updating the system biological model using the personalized molecular levels comprises using disease gene transcription levels, disease protein levels, or a combination of both derived from non-invasively obtained data.

[0021] In some embodiments, the non-invasively obtained data is imaging data.

[0022] In some embodiments the imaging data is radiology imaging data.

[0023] In some embodiments, the radiology imaging data is obtained by computed tomography (CT), dual-energy computed tomography (DECT), spectral computed tomography (spectral CT), computed tomography angiography (CTA), cardiovascular computed tomography angiography (CCTA), magnetic resonance imaging (MRI), multi-contrast magnetic resonance imaging (multi-contrast MRI), ultrasound (US), positron emission tomography (PET), intravascular ultrasound (IVUS), optical coherence tomography (OCT), near-infrared spectroscopy (NIRS), or single-photon emission computed tomography (SPECT) diagnostic images, or any combination thereof.

[0024] In some embodiments, the methods described above further comprise processing the non-invasively acquired imaging data to obtain quantitative plaque morphology data, including anatomical structure data, tissue composition data, or both.

[0025] In some embodiments, the anatomical data comprises data regarding any one or more levels of remodeling, wall thickening, ulceration, stenosis, dilation, or plaque burden.

[0026] In some embodiments, the tissue composition data comprises data regarding any one or more levels of calcification, necrotic lipid core (LRNC), intraplaque hemorrhage (IPH), matrix, fibrous cap, and perivascular adipose tissue (PVAT).

[0027] In some embodiments, pathways are compartmentalized into cell-specific networks.

[0028] In some embodiments, the cell-specific network includes at least an endothelial cell network, a macrophage network, and a vascular smooth muscle cell network.

[0029] In some embodiments, the candidate therapy is a dyslipidemia management drug.

[0030] In some embodiments, the dyslipidemia management medication is a high-dose statin.

[0031] In some embodiments, the high-dose statin is atorvastatin.

[0032] In some embodiments, the dyslipidemia management drug is an intensive lipid-lowering drug.

[0033] In some embodiments, the enhanced lipid-lowering agent is a proprotein convertase subtilisin kexin type 9 (PCSK9) inhibitor or a cholesteryl ester transfer protein (CETP).

[0034] In some embodiments, the dyslipidemia management drug is a hypertriglyceridemia-lowering drug or a hypercholesterolemia-lowering drug.

[0035] In some embodiments, the candidate therapeutics are drugs that affect the inflammatory cascade.

[0036] In some embodiments, the agent that affects the inflammatory cascade is an anti-inflammatory agent.

[0037] In some embodiments, the anti-inflammatory agent is an inhibitor of IL-1.

[0038] In some embodiments, the inhibitor of IL-1 is canakinumab.

[0039] In some embodiments, the anti-inflammatory agent inhibits the activity of TNF.

[0040] In some embodiments, the anti-inflammatory agent inhibits IL12 / 23.

[0041] In some embodiments, the anti-inflammatory agent inhibits IL17.

[0042] In some embodiments, the agent that affects the inflammatory cascade is an inhibitor of danger signal-induced inflammatory cytokines.

[0043] In some embodiments, the agent that affects the inflammatory cascade is a resolvin precursor.

[0044] In some embodiments, the resolvin precursor is an omega-3 fatty acid.

[0045] In some embodiments, the omega-3 fatty acid is eicosapentaenoic acid (EPA), docosahexaenoic acid (DHA), or docosapentaenoic acid (DPA).

[0046] In some embodiments, the candidate therapy is an immunomodulatory agent.

[0047] In some embodiments, the immunomodulatory agent triggers innate immunity.

[0048] In some embodiments, the immunomodulatory agent is an immune tolerance stimulating agent.

[0049] In some embodiments, the immune tolerance stimulant increases the activity of Tregs.

[0050] In some embodiments, the candidate therapy is an antihypertensive agent.

[0051] In some embodiments, the antihypertensive agent is an ACE inhibitor.

[0052] In some embodiments, the candidate therapy is an anticoagulant.

[0053] In some embodiments, anticoagulants reduce thrombin generation and / or limit the action of thrombin.

[0054] In some embodiments, the candidate therapeutics are modulators of intracellular signaling.

[0055] In some embodiments, the candidate therapy is an anti-diabetic drug.

[0056] In some embodiments, the antidiabetic agent is metformin.

[0057] In some embodiments, the candidate therapy is a drug-eluting stent.

[0058] In some embodiments, the drug-eluting stent is coated with a drug that inhibits cell cycle progression by inhibiting DNA synthesis.

[0059] In some embodiments, the therapeutic candidate is a drug coated balloon.

[0060] In some embodiments, the drug coated balloon is coated with a drug that inhibits neointimal growth by delivering an anti-proliferative substance to the vessel wall.

[0061] In some embodiments, the candidate therapies are a combination of one or more of lipid-lowering agents, anti-inflammatory agents, and anti-diabetic agents.

[0062] In some embodiments, the method further comprises quantifying the patient's actual response to each potential therapy.

[0063] In some embodiments, the method further comprises detecting one or more candidate contraindications associated with each candidate therapy.

[0064] In some embodiments, the method further comprises identifying likely side effects for each potential therapy.

[0065] In some embodiments, the method further comprises identifying potential toxicities for each potential therapy.

[0066] In some embodiments, the method further comprises identifying potential future side effects in response to each potential therapy.

[0067] In some embodiments, the therapeutic response to each candidate therapy is simulated in the system biological model by determining a known set of molecules affected by the candidate therapy, defining a therapeutic effect molecular level for each molecule in the known set of molecules based on one or more known mechanisms of action of the candidate therapy on the known set of molecules, and estimating a therapeutic effect molecular level for other molecules represented in the system biological model other than the known set of molecules based on a simulated effect of the defined therapeutic effect molecular level of the known set of molecules on one or more of the other molecules represented in the network.

[0068] In some embodiments, the method comprises comparing the determined and estimated therapeutic effect molecule levels in the system biological model before and after a therapeutic response simulation for each potential therapy.

[0069] In some embodiments, the system biology model includes one or more pathways depicted in Table 5 or Table 6.

[0070] Also provided herein is a method of screening therapeutic candidate agents for treating atherosclerotic cardiovascular disease, the method comprising the steps of receiving non-invasively obtained data relating to plaque from each of a plurality of subjects diagnosed with atherosclerotic cardiovascular disease; accessing a systems biological model of atherosclerotic cardiovascular disease, where (i) the systems biological model represents a plurality of pathways associated with atherosclerotic cardiovascular disease and (ii) the systems biological model includes disease-related molecular levels for each molecule in the systems biological model; updating the systems biological model using the disease-related molecular levels derived from the non-invasively obtained data from the subjects to generate a validated systems biological model; updating the validated systems biological model with information about the therapeutic candidate agents based on a known mechanism of action of the therapeutic candidate agents; simulating a therapeutic response to the therapeutic candidate agents in the updated and validated systems biological model to obtain a simulated therapeutic effect; comparing the therapeutic effect in the updated and validated systems biological model before and after simulating the therapeutic response with the therapeutic candidate agents; and determining whether the therapeutic candidate agents have a therapeutic effect based on the comparison. In some embodiments, the method further comprises quantifying the actual response at a cohort level. In some embodiments, the selection method allows for selection of cases that increase the power of the clinical trial. In some embodiments, the selection method allows for selection of cases that decrease the power of the clinical trial.

[0071] Also provided herein is a method of selecting patient candidates to participate in a clinical trial testing the safety, or efficacy, or both, of a therapeutic candidate for patients with known or suspected atherosclerotic cardiovascular disease, the method comprising the steps of receiving non-invasively obtained data regarding plaque from the candidate subject; accessing a systems biology model of atherosclerotic cardiovascular disease; updating the systems biology model using personalized molecular levels derived from the non-invasively obtained data from the candidate subject to generate a subject-specific systems biology model; and updating the systems biology model with information regarding the therapeutic candidate based on a known mechanism of action of the therapeutic candidate. the candidate subject's response to the candidate therapeutic agent in the updated subject-specific systems biological model to obtain a simulated therapeutic effect for the candidate therapeutic agent; comparing the updated subject-specific systems biological model with the simulated therapeutic effect for each of the two or more combinations to the updated subject-specific systems biological model without the simulated therapeutic effect; and providing a report indicating whether the candidate subject's atherosclerotic cardiovascular disease is likely to be ameliorated or unaffected by the candidate therapeutic agent for the subject and / or whether the candidate subject will experience side effects from the candidate therapeutic agent.

[0072] A computer-implemented method is also provided herein, comprising the steps of: receiving a first input indicative of biological pathways associated with atherosclerotic cardiovascular disease; generating a first network based on the first input, the first network including nodes representing reference molecular levels and edges representing molecular interactions in one or more cell types; receiving a second input indicative of calibration data from a plurality of subjects diagnosed with the disease; determining disease-associated molecular levels for the molecules in the first network from the second input; and generating a second network based on the first network and the disease-associated molecular levels, wherein the second network calibrated using the second input represents an in silico systems biological model of the disease and includes disease-associated molecular levels for each molecule in the second network.

[0073] In some embodiments of the computer-implemented method, receiving a plurality of first inputs comprises querying a pathway database to identify biological pathways associated with atherosclerotic cardiovascular disease.

[0074] In some embodiments of the computer-implemented method, the one or more cell types comprise endothelial cells, vascular smooth muscle cells, macrophages, and lymphocytes.

[0075] In some embodiments of the computer-implemented method, the first network comprises (i) a core network representing molecular interactions specific to each respective cell type, (ii) a mid network representing molecular interactions across a subset of cell types, and (iii) a full network representing molecular interactions found in all cell types.

[0076] In some embodiments of the computer-implemented method, the edge representing an intermolecular interaction represents any one of translation, activation, inhibition, indirect effect, state change, binding, dissociation, phosphorylation, dephosphorylation, glycosylation, ubiquitination, and methylation.

[0077] In some embodiments of the computer-implemented method, receiving the second input comprises, for each subject, obtaining at least plaque computed tomography angiography imaging data from the subject, plaque morphology data, and proteomic data corresponding to the subject.

[0078] In some embodiments of the computer-implemented method, the method further comprises receiving transcriptomic data for at least a portion of the subjects.

[0079] In some embodiments of the computer-implemented method, the molecule is a protein, a gene, or a metabolite.

[0080] In some embodiments of the computer-implemented method, the first network includes nodes representing baseline levels of proteins and genes and edges representing protein-protein interactions, gene-gene interactions, and protein-gene interactions in one or more cell types.

[0081] In some embodiments of the computer-implemented method, the disease molecular levels are either measured molecular levels from the subject, or estimated molecular levels based on a virtual tissue model, or non-invasively obtained imaging data from the subject, or both.

[0082] In some embodiments of the computer-implemented method, determining the disease molecular level for the molecules in the first network comprises identifying disease molecular levels for the set of molecules from a second input, where the disease molecular levels of the set of molecules are provided by the second input from the subject, and estimating the disease molecular level for molecules in the first network other than the set of molecules based on the disease molecular levels of a subset of the set of molecules, where the subset of the set of molecules are represented by adjacent nodes in the first network.

[0083] In some embodiments of the computer-implemented method, generating the second network comprises indicating a disease molecule level for each node in the first network whose disease molecule level is obtained from calibration data from the subject, and indicating a disease molecule level for each node in the first network whose disease molecule level is estimated.

[0084] Also provided is a computer-implemented method of making a therapy recommendation for a patient with known or suspected atherosclerotic cardiovascular disease, the method comprising the steps of receiving non-invasively obtained imaging data of atherosclerotic plaques from the patient; accessing a trained in silico system biological model of atherosclerotic cardiovascular disease, the trained in silico system biological model comprising a network with a disease molecular level for each of a plurality of nodes, each node representing a different molecule; updating the system biological model for the patient using the disease molecular levels derived from the imaging data; determining a known set of molecules affected by a therapy candidate; determining a therapy effect molecular level for each molecule in the known set of molecules based on one or more effects of the therapy candidate on the known set of molecules; estimating therapy effect molecular levels for other molecules represented in the in silico system biological model other than the known set of molecules based on a simulated effect of the determined therapy effect molecular level of the known set of molecules on one or more of the other molecules represented in the network; The method includes simulating a therapeutic response to each of a set of therapeutic candidate therapies in the updated and trained in silico system biological model by comparing a determined therapeutic effect molecular level with an estimated therapeutic effect molecular level in the silico system biological model, determining a preferred therapy based on the comparison, and optionally providing a report indicating the preferred therapy for the patient.

[0085] In some embodiments of the computer-implemented method, updating the network using disease molecule levels derived from the imaging data comprises comparing the computed tomography angiography imaging data of the patient with a plurality of computed tomography angiography imaging data of a plurality of subjects, where the plurality of computed tomography angiography imaging data of the plurality of subjects were inputs for training the system biological model, and predicting disease molecule levels for the molecules in the network based on the comparison.

[0086] In some embodiments of the computer-implemented method, the candidate therapy is a dyslipidemia management drug.

[0087] In some embodiments of the computer-implemented method, the dyslipidemia management medication is a high-dose statin.

[0088] In some embodiments of the computer-implemented method, the high-dose statin is atorvastatin.

[0089] In some embodiments of the computer-implemented method, the dyslipidemia management medication is an intensive lipid-lowering medication.

[0090] In some embodiments of the computer-implemented method, the enhanced lipid-lowering agent is a proprotein convertase subtilisin kexin type 9 (PCSK9) inhibitor or a cholesteryl ester transfer protein (CETP).

[0091] In some embodiments of the computer-implemented method, the dyslipidemia management medication is an antihypertriglyceridemia medication or an antihypercholesterolemic medication.

[0092] In some embodiments of the computer-implemented method, the candidate therapies are drugs that affect the inflammatory cascade.

[0093] In some embodiments of the computer-implemented method, the drug that affects the inflammatory cascade is an anti-inflammatory drug.

[0094] In some embodiments of the computer-implemented method, the anti-inflammatory agent is an inhibitor of IL-1.

[0095] In some embodiments of the computer-implemented method, the inhibitor of IL-1 is canakinumab.

[0096] In some embodiments of the computer-implemented method, the anti-inflammatory drug inhibits the activity of TNF.

[0097] In some embodiments of the computer-implemented method, the anti-inflammatory drug inhibits IL12 / 23.

[0098] In some embodiments of the computer-implemented method, the anti-inflammatory drug inhibits IL17.

[0099] In some embodiments of the computer-implemented method, the drug that affects the inflammatory cascade is an inhibitor of inflammatory cytokines induced by danger signals.

[0100] In some embodiments of the computer-implemented method, the drug that affects the inflammatory cascade is a resolvin precursor.

[0101] In some embodiments of the computer-implemented method, the resolvin precursor is an omega-3 fatty acid.

[0102] In some embodiments of the computer-implemented method, the omega-3 fatty acid is eicosapentaenoic acid (EPA), docosahexaenoic acid (DHA), or docosapentaenoic acid (DPA).

[0103] In some embodiments of the computer-implemented method, the candidate therapy is an immunomodulatory agent.

[0104] In some embodiments of the computer-implemented method, the immunomodulatory agent triggers innate immunity.

[0105] In some embodiments of the computer-implemented method, the immunomodulatory agent is an immune tolerance stimulating agent.

[0106] In some embodiments of the computer-implemented method, the immune tolerance stimulating agent increases the activity of Tregs.

[0107] In some embodiments of the computer-implemented method, the candidate therapy is an antihypertensive drug.

[0108] In some embodiments of the computer-implemented method, the antihypertensive drug is an ACE inhibitor.

[0109] In some embodiments of the computer-implemented method, the candidate therapy is an anticoagulant.

[0110] In some embodiments of the computer-implemented method, the anticoagulant reduces thrombin generation and / or limits the activity of thrombin.

[0111] In some embodiments of the computer-implemented method, the candidate therapies are modulators of intracellular signaling.

[0112] In some embodiments of the computer-implemented method, the candidate therapy is an anti-diabetic drug.

[0113] In some embodiments of the computer-implemented method, the antidiabetic drug is metformin.

[0114] In some embodiments of the computer-implemented method, the candidate therapy is a drug-eluting stent.

[0115] In some embodiments of the computer-implemented method, the drug-eluting stent is coated with a drug that inhibits cell cycle progression by inhibiting DNA synthesis.

[0116] In some embodiments of the computer-implemented method, the candidate therapy is a drug-coated balloon.

[0117] In some embodiments of the computer-implemented method, the drug-coated balloon is coated with a drug that inhibits neointimal growth by delivering an anti-proliferative substance to the vessel wall.

[0118] In some embodiments of the computer-implemented method, the candidate therapies are a combination of one or more of lipid-lowering drugs, anti-inflammatory drugs, and anti-diabetic drugs.

[0119] In some embodiments of the computer-implemented method, determining the therapeutic effective molecule level comprises setting the therapeutic effective molecule level of the set of molecules to a reference level.

[0120] Also provided is a system comprising a memory configured to store instructions and a processor to execute the instructions to perform operations, the operations comprising: receiving a first input indicative of biological pathways associated with atherosclerotic cardiovascular disease; generating a first network based on the first input, the first network including nodes representing reference molecular levels and edges representing molecular interactions in one or more cell types; receiving a second input indicative of calibration data from a plurality of subjects diagnosed with the disease; determining disease molecular levels for the molecules in the first network from the second input; and generating a second network based on the first network and the disease molecular levels, the second network calibrated using the second input representing an in silico system biological model of the disease and including disease molecular levels for each molecule in the second network.

[0121] Also provided is one or more computer-readable media storing instructions that are executed by a processing device and that, when executed, cause the processing device to perform operations comprising: receiving a first input indicative of biological pathways associated with atherosclerotic cardiovascular disease; generating a first network based on the first input, the first network including nodes representing reference molecular levels and edges representing molecular interactions in one or more cell types; receiving a second input indicative of calibration data from a plurality of subjects diagnosed with the disease; determining disease molecular levels for the molecules in the first network from the second input; and generating a second network based on the first network and the disease molecular levels, the second network calibrated using the second input representing an in silico systems biological model of the disease and including disease molecular levels for each molecule in the second network.

[0122] Also provided is a system comprising a memory configured to store instructions and a processor to execute the instructions to perform operations, the operations including receiving non-invasively obtained imaging data of atherosclerotic plaque from a patient; accessing a trained in silico system biological model of atherosclerotic cardiovascular disease, the trained in silico system biological model comprising a network comprising a disease molecular level for each of a plurality of nodes, each node representing a different molecule; updating the system biological model for the patient using the disease molecular levels derived from the imaging data; determining a known set of molecules affected by a therapy candidate; determining a therapy effect molecular level for each molecule in the known set of molecules based on one or more effects of the therapy candidate on the known set of molecules; estimating therapy effect molecular levels for other molecules represented in the in silico system biological model other than the known set of molecules based on a simulated effect of the determined therapy effect molecular level of the known set of molecules on one or more of the other molecules represented in the network; The method includes simulating a therapeutic response to each of a set of therapeutic candidate therapies in the updated and trained in silico system biological model by comparing the determined therapeutic effect molecular levels with the estimated therapeutic effect molecular levels in the silico system biological model, determining a preferred therapy based on the comparison, and providing a report indicating the preferred therapy for the patient.

[0123] Also provided is one or more computer readable media storing instructions that are executable by a processing device and that, when executed, cause the processing device to perform operations including receiving non-invasively obtained imaging data of atherosclerotic plaques from a patient; accessing a trained in silico system biological model of atherosclerotic cardiovascular disease, the trained in silico system biological model comprising a network with a disease molecular level for each of a plurality of nodes, each node representing a different molecule; updating the system biological model for the patient using the disease molecular levels derived from the imaging data; determining a known set of molecules affected by a therapy candidate; determining a therapy effect molecular level for each molecule in the known set of molecules based on one or more effects of the therapy candidate on the known set of molecules; estimating therapy effect molecular levels for other molecules represented in the in silico system biological model other than the known set of molecules based on a simulated effect of the determined therapy effect molecular level of the known set of molecules on one or more of the other molecules represented in the network; The method includes simulating a therapeutic response to each of a set of therapeutic candidate therapies in the updated and trained in silico system biological model by comparing the determined therapeutic effect molecular levels with the estimated therapeutic effect molecular levels in the silico system biological model, determining a preferred therapy based on the comparison, and providing a report indicating the preferred therapy for the patient.

[0124] definition A "computational model" uses computer programs to simulate and study complex systems using algorithmic or mechanistic techniques.

[0125] A "predictive model" is a mathematical description, often described as artificial intelligence, machine learning, or deep learning, that calculates one or more outputs ("response variables") from one or more inputs ("predictors"). In this application, predictive models can be used to characterize tissues (as "virtual tissue models"), to predict molecular levels from characterized tissues, or to predict outcomes from either tissue properties and / or virtual omics.

[0126] "Systems Biological Model" refers to a model used to represent a set of interconnected biological pathways that may be used to simulate changes across the interconnected biological pathways under defined conditions.

[0127] An "in silico system biological model" refers to a computer representation of a biological system, for example, the biological system is atherosclerotic cardiovascular disease.

[0128] "Initial in silico system organism model" refers to an in silico system organism model that is generated or trained using actual proteomic data obtained from a development subject and information obtained from a literature search.

[0129] "Calibrated in an in silico system biological model" refers to an initial in silico system biological model that is updated using measured calibration data, such as omics data from a given subject (e.g., a human subject) diagnosed with cardiovascular disease or from a patient with known or suspected cardiovascular disease.

[0130] "Calibration data" refers to subject-derived or patient-specific data that can be used to update an in silico system biological model. Examples include measured omics data, such as transcriptomics, proteomics, and / or metabolomics data, e.g., non-invasively obtained. Calibration data can also be obtained from molecular or tissue assays, e.g., biopsies.

[0131] "Omics data" refers to biologically significant quantities of gene expression, transcriptomics, proteomics, or metabolomics based on directly measured molecular expression levels, for example, by blood tests, molecular assays, or tissue biopsies.

[0132] "Virtual omics data" refers to computationally predicted levels of biologically significant quantities of gene expression, transcriptomics, proteomics, or metabolomics (e.g., based on patient-derived imaging data) rather than directly measured molecular expression levels, e.g., by blood tests, molecular assays, or tissue biopsies.

[0133] "Network" refers to a graphical representation (edges) of interactions between various molecules (nodes).

[0134] An "artificial neural network" refers to a type of computational model that is mathematically structured to resemble the human brain as a series of interconnected "neurons", or weighted summations, thereby providing a means to represent complex relationships that are highly nonlinear.

[0135] "Direction" (of an edge) refers to the direction of interaction between a pair of molecules (e.g., when molecule A activates molecule B, the direction is from A to B).

[0136] A "biological pathway" refers to a series of actions between molecules that lead to a product or change.

[0137] "Baseline level (of a molecule)" refers to the biological state (e.g., expression level) of a molecule prior to a perturbation in a system biological model (e.g., in a healthy person or subject, before the subject or patient acquires a disease, or before the patient begins a new treatment for a diagnosed disease).

[0138] "Molecule" refers to a gene (also called a transcript or gene transcript), a protein, or a metabolite.

[0139] "Disease-associated level" (of a molecule) refers to the quantitative amount of a molecule (gene transcript, protein, or metabolite) from an individual subject diagnosed with a particular disease. In some cases, the disease-associated level of a molecule can be determined based on virtual omics data, which can include data obtained from plaque tissue, and may also include data from minimally diseased tissue, so long as the data is taken from a subject diagnosed with a disease, e.g., cardiovascular disease. Note that during model generation, disease-associated levels from the subject are utilized, while during operation in the clinic, personalized levels are used, and the term "calibration" applies in both contexts.

[0140] "Personalized level" (of a molecule) refers to the quantitative amount of a molecule (transcript, protein, or metabolite) from an individual patient. In some cases, the personalized level of a molecule may be determined based on virtual omics data. Note that during model generation, disease-associated levels from subjects are utilized, while during operation in the clinic, personalized levels are used, and the term "calibration" applies in both contexts.

[0141] "Phenotype" refers to a set of observable characteristics of an individual that results from the interaction of that individual's genotype with the environment. As used herein, it may be understood to refer to an "endotype" (a subtype of a disease state defined by a distinct pathophysiological mechanism), or a "theratype" (a means of grouping according to response to a particular alternative therapy), terms that may be used in the field of precision medicine for classification or typing performed without loss of generality by the methods and systems described herein.

[0142] "Biochemical reactions" refer to interactions between a plethora of molecules, such as molecules (e.g., transcripts, RNA, proteins, metabolites, inorganic compounds, etc.). Specifically, it refers to the transformation of one molecule into a different molecule inside a cell, usually (but not always) annotated with quantitative coefficients or terms that allow effects to propagate through a network.

[0143] A "biochemical relationship" is a semi-quantitative approximation to a biochemical reaction. "Reaction" and "relationship" are used interchangeably (i.e., interchangeably) in this disclosure without loss of generality.

[0144] The new methods and systems described herein provide numerous advantages and benefits, as well as improved ability to make patient-specific therapy recommendations for atherosclerotic cardiovascular disease.

[0145] The number of people affected by atherosclerosis is large. Most patients are unaware of the progression of the disease until symptoms appear. Patient risk management relies to a large extent on population-based scoring methods such as the Framingham risk score (Newby et al., Coronary CT Angiography and 5-Year Risk of Myocardial Infarction. N Engl J Med, 2018. 379(10): p. 924-933; Bergstrom et al., The Swedish CArdioPulmonary BioImage Study: objectives and design. J Intern Med, 2015. 278(6): p. 645-59), justifying the development of diagnostic methods for more accurate patient stratification. As multiple treatment options become available for patients with CVD, patient satisfaction must increasingly be based on the individual patient rather than on population-based risk factors / scoring or simple imaging methods. For example, the degree of stenosis, calcium scoring, or even fractional flow reserve (FFR) are not specific enough to determine the disease category of an individual patient at the level required to identify which treatment is best for the patient, i.e., to choose between observation, drug therapy, surgical intervention, surgery, or a specific treatment within one of these categories. This is important both economically and clinically because recent advances in drug therapy that target more effective specific mechanisms are generally much more expensive than earlier generation drugs such as statins, and are too expensive to be widely used by the population. These new drugs are also not the best therapy for all patients, and the methods and systems of the present invention can be used to match appropriate patients with the best therapy.

[0146] One current challenge is that the ability to measure responses to specific drug therapies remains poorly understood, and both undertreatment and overtreatment remain common problems, which can lead to many patients being unnecessarily treated, while at the same time wasting financial resources and patients undergoing unnecessary invasive procedures for the results obtained. Similarly, as long as methods are only proposed to assess vulnerable plaques, the problem remains that vulnerable plaques can only be found, and therefore their cause is systemic rather than localized. This makes localized treatment incompatible with the actual cause of the plaque, which may instead justify systemic treatment. The concept of "vulnerable patients" has been discussed, but if we are to obtain demonstrable improvement in outcomes for a given societal cost, for example through personalized therapy, we need markers to identify such individuals and be able to categorize the specific mechanisms causing vulnerability at the individual level. Each of these needs and opportunities represents a challenge to methods developed thus far, but is addressed by the methods and systems described herein.

[0147] The present disclosure aids in understanding the extent and speed of atherosclerosis progression under different alternative potential treatments. Advanced software-based techniques for extracting data embedded in images that would otherwise not be easily visually or quantitatively recognized provide biomarkers to identify patients with unstable atherosclerosis, and imaging to localize unstable atherosclerotic plaques, providing more accurate characterization from clinical care to developing more effective drugs for patients at risk for ischemic events.

[0148] The new methods and systems described herein provide improved outcomes and costs, including improved non-invasive diagnostics to identify which patients have ongoing disease, and the ability to make automated recommendations of the best therapy or combination of therapies for each particular patient based on a simulation of how a particular therapy is likely to affect a particular patient and how the patient will respond when receiving a particular therapy. The methods and systems can also be used to select or adjust doses of particular drugs based on the simulated patient response, as well as to simulate the effects of new drug candidates, i.e., virtual clinical trials.

[0149] Beyond indicating that there is a problem, the virtual biomarkers described here can also specifically classify patients as to the most effective way to treat that problem. Moreover, the onset of symptoms is considered in terms of both kinetic failure (e.g., pressure-induced ischemia of perfused tissues) and destructive events such as thrombosis and rupture (i.e., causing infarction). While plasma biomarkers play an important role as selection tools, they themselves are not sensitive or specific enough to recognize what is happening in tissues, for example in the plaque and in the tissues surrounding it (i.e., tissue and blood transcriptomics and proteomics).

[0150] Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this invention belongs.Methods and materials are described herein for use in the present invention, but other suitable methods and materials known in the art can also be used.Materials, methods, and examples are illustrative only and are not intended to be limiting.All publications, patent applications, patents, sequences, database entries, and other references mentioned herein are incorporated by reference in their entirety.In case of discrepancy, the present specification, including definitions, will take precedence.

[0151] Other features and advantages of the invention will become apparent from the following detailed description and drawings, and from the claims. [Brief description of the drawings]

[0152] [Figure 1] FIG. 1 is a high-level schematic flowchart illustrating how computational modeling can be used to represent relationships between clinical, physiological, and molecular entities or concepts to delineate the pathogenesis of diseases such as atherosclerosis across multiple time and space axes. [Figure 2A] A series of arterial non-invasive computed tomography angiography (CTA) images (left-most column labeled column A), 3D images generated from the CTA (column B), 2D / axial views of the CTA images (column C, although the white lines in the images in column B indicate the location of the sections), and histological images (column D) of tissues with characteristics of necrotic lipid core (LRNC), calcification (CALC), intraplaque hemorrhage (IPH), matrix / fibrous tissue (MATX), and fibrous cap / perivascular adipose tissue (FC / PVAT). Without loss of generality, the specific tissues shown are examples. [Figure 2B] 1 is a schematic image showing multiple objectively validated measurements for characterizing plaque morphology by analysis software, among which tissues can be differently colored elements to define type as one of the following categories, for example LRNC, CALC, IPH, matrix, fibrous cap, and PVAT, or other relevant tissue types as appropriate. Without loss of generality, the specific tissues shown are examples. [Figure 3A-3F]Serial histological images (left column) and images of non-invasive computed tomography analysis (middle column) of two subjects in the study cohort for unstable (A-C) and stable (D-F) atherosclerosis. The middle column (Figure 3B and Figure 3E) shows the 3D view provided by the imaging software. The right column (Figure 3C and Figure 3F) shows the classifier output for stability phenotypes aligned with the 3D images, where red means unstable plaque, yellow means stable plaque, and green means minimal disease. [Figure 4] FIG. 1 is a diagram of a workflow outlining the steps for determining a function f using a training dataset for optimizing various types of models. The results can be further applied to supervised or unsupervised clustering for use in virtual omics. [Diagram 5] FIG. 1 is a diagram of a workflow showing an example of how reactions or relationships (i.e., interactions on a biological pathway) are identified and how concentration / rate constants, or other quantitative relationships, are enumerated. Note that, without loss of generality, number 116 is an example and more or fewer references may be used, and that while many resources are in the public domain, proprietary or non-public resources may also be used without loss of generality. [Figures 6A-6C] 1 is an image collectively illustrating the workflow steps taken to create an in silico system biomodel of atherosclerosis for simulation of an individual subject's response to different therapies, e.g., drug therapies and / or surgical interventions. [Figure 6A]FIG. 1 is a schematic diagram showing how a particular system biomodel as described herein is created from molecular data and literature-based sources, and how the model can be updated based on subject data to calibrate the initial model, and how the calibrated model is updated with patient imaging data and with specific drug mechanism of action (MOA) data that may be useful for a given patient to perturb the system to provide a simulated treatment response for that patient and resulting therapy recommendation for that patient. Without loss of generality, the specific figures shown are examples. [Figure 6B] FIG. 1 is a schematic diagram showing how three types of biodata can be derived from non-invasive radiology data using machine learning according to different reference truth bases. Input 1 from the diagram represents patient data (CTA) that is not used in the modeling but for validating the model. Result 2 from the diagram is a set of structural properties of anatomical tissues (quantitative plaque morphology) defined by histopathology. Results 3 and 4 represent virtual transcriptomics and proteomics data defined by and validated from inputs "B" and "C", respectively. Without loss of generality, the input in "B" may be microarray or RNAseq data or other means for assessing coding or non-coding RNA, and the input in "C" may be liquid chromatography mass spectrometry or other means for assessing protein levels. [Figure 6C] FIG. 6B is a schematic showing how the results from FIG. 6B can be used to calibrate the quantities of reactions or relationships in a systems organism model. Now, looking at the molecular level, item 2 (quantitative plaque morphology) is retained as a continuation from FIG. 6B. Expression data 3 can be used to calibrate the rate constants, or the relative magnitudes or weights of relationships regarding how one molecule affects another. Level data 4 can be used to calibrate the molecular levels. These reactions / relationships are interconnected together to make up the systems organism model 5. [Figure 7A] FIG. 1 is a block diagram of an example system for generating an in silico biological model of atherosclerotic cardiovascular disease. [Figure 7B] FIG. 1 is a block diagram of an example system for making therapy recommendations based on an in silico system biological model. [Figure 8A] 1 is a flow chart of an example process for generating an in silico systemic biological model of atherosclerotic cardiovascular disease. [Figure 8B] 1 is a flowchart of an example process for making therapy recommendations based on an in silico system biological model. [Figure 8C] 1 is a flowchart of an example process for making therapy recommendations based on an in silico system biological model. [Figure 9] FIG. 1 is a schematic diagram of example system components that may be used to implement the systems and methods. [Figure 10] FIG. 1 is a schematic showing examples of how pathways can be compartmentalized into cell-specific networks, here an endothelial cell network, a macrophage network, and a vascular smooth muscle cell (VSMC) network. Without loss of generality, the specific cell types shown are examples. [Figure 11] FIG. 1 is a schematic showing first level targets for a reference unstable subject (subject P491) depicted in a layout that highlights compartmentalization with plasma with serum LDL (pink) shown to reflect relationships with proteins at the cell membrane and in the extracellular domain of endothelial cells (green), macrophages (orange), VSMCs (aquamarine), lymphocytes (blue). Without loss of generality, the specific compartments and cell types shown are examples. [Figure 12]1 shows an image of the "full" extent integrated intimal network of an unstable subject (subject P491) in the untreated or baseline state. Note that, without loss of generality, other integrated networks for adventitial, medial, or perivascular spaces may also be used. [Figure 13A] 13 is an image showing the calibration of an individual subject, and a map showing the molecules that had direct measurements on the EC core network. [Figure 13B] 13 is an image showing individual subject calibration, depicting interpolated values ​​showing the propagation of levels from non-interpolated proteins according to the type and weight of the relationships drawn from the pathway designations. [Figure 14] A heatmap identifying the top 25 proteins in terms of variance among signatures in an experimental cohort as described herein, in this case for the mid-range network of endothelial cells. This heatmap is provided as an example and other cell types, network ranges, or protein levels should be understood without loss of generality. [Figure 15] A heatmap identifying the top 25 proteins in terms of variance among signatures in our experimental cohort, in this case for the mid-range network of VSMCs. This heatmap is shown as an example and other cell types, network ranges, or protein levels should be understood without loss of generality. [Figure 16] A heatmap identifying the top 25 proteins in terms of variance between signatures in our experimental cohort, in this case for the macrophage mid-range network. This heatmap is shown as an example and other cell types, network ranges, or protein levels should be understood without loss of generality. [Figure 17]A heatmap identifying the top 25 proteins in terms of variance among signatures in our experimental cohort, in this case for the lymphocyte mid-range network. This heatmap is shown as an example and other cell types, network ranges, or protein levels should be understood without loss of generality. [Figure 18] A heatmap identifying the top 25 proteins in terms of variance between signatures in our experimental cohort, in this case for the inner membrane mid-range network. This heatmap is shown as an example and other cell types, network ranges, or protein levels should be understood without loss of generality. [Figure 19A] 1 shows the intima model in the "core" region before simulation of enhanced lipid-lowering drug treatment. This heatmap is shown as an example, and other cell types, network regions, or therapeutic candidates should be understood without loss of generality. [Figure 19B] 1 shows the intima model in the "core" region after simulation of enhanced lipid-lowering drug treatment. This heatmap is shown as an example, and other cell types, network regions, or therapeutic candidates should be understood without loss of generality. [Figure 20] 1 is a "caterpillar" chart showing how different subjects can differ with respect to subject-specific plaque instability. [Figure 21A] 13 is a plot showing mean absolute cohort-level instability from multilevel analysis across multiple cell types and ranges. [Figure 21B] 13 is a plot showing mean absolute cohort-level instability from multilevel analysis across multiple cell types and ranges. [Figure 21C] 13 is a plot showing mean absolute cohort-level instability from multilevel analysis across multiple cell types and ranges. [Figure 21D]13 is a plot showing mean absolute cohort-level instability from multilevel analysis across multiple cell types and ranges. [Figure 21E] 13 is a plot showing mean absolute cohort-level instability from multilevel analysis across multiple cell types and ranges. [Figure 21F] 13 is a plot showing mean absolute cohort-level instability from multilevel analysis across multiple cell types and ranges. [Figure 21G] 13 is a plot showing mean absolute cohort-level instability from multilevel analysis across multiple cell types and ranges. [Figure 22A] 13 is a plot showing the mean relative treatment effect (positive means less instability). [Figure 22B] 13 is a plot showing the mean relative treatment effect (positive means less instability). [Figure 22C] 13 is a plot showing the mean relative treatment effect (positive means less instability). [Figure 22D] 13 is a plot showing the mean relative treatment effect (positive means less instability). [Figure 22E] 13 is a plot showing the mean relative treatment effect (positive means less instability). [Figure 22F] 13 is a plot showing the mean relative treatment effect (positive means less instability). [Diagram 23] 1 is a radar chart showing the degree of absolute atherosclerotic plaque stability for an example set of patients, with the outer line being better for the patient, green meaning minimal disease, yellow meaning stable plaque, and red meaning unstable plaque. [Figure 24] A radar chart showing the relative improvement after treatment simulation for an example set of patients. There are different ways of presenting the data, which are also shown in absolute charts, to better visualize the change rather than the overall effect of the treatment, respectively. Here, the outer lines represent more pronounced effects, green means improvement, and red means worsening of the disease. [Figure 25A] FIG. 1 is a diagram of personalized subject treatment recommendations for a patient based on actual data. [Figure 25B] FIG. 1 is a diagram of personalized subject treatment recommendations for a patient based on actual data. [Figure 25C] FIG. 1 is a diagram of personalized subject treatment recommendations for a patient based on actual data. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0153] The methods and systems described herein not only characterize atherosclerosis in terms of morphology and stability based on non-invasively obtained data, e.g., non-invasive imaging data, of a patient's arteries (e.g., using CT angiography), but also provide therapy recommendations for individual patients based on the nature and stability of plaque, all using only non-invasively obtained data, e.g., imaging data, from the patient, such as arterial imaging data. For example, by obtaining genotypic and / or phenotypic information for a given patient (i.e., through virtual omics modeling or based on actual measurements), the novel methods and systems described herein can be used to model the patient's expected response to various therapies, including drug / pharmaceutical therapies and interventional or surgical therapies, in order to recommend therapies predicted to provide superior outcomes for that particular patient.

[0154] Since morphological and biological characteristics of atherosclerotic plaques can be determined by non-invasive imaging, diagnostic accuracy is improved. To do this, quantitative connections between scales were established. Specifically, as shown in FIG. 1, as time progresses, atherosclerosis progresses at spatial scales starting at the molecular level on time scales of seconds to minutes, and progressing to the systemic level on time scales of months, years, and decades. As described herein, computational modeling techniques were used to represent relationships across multiple temporal and spatial scales.

[0155] The need for new methods and systems is clear. Myocardial infarction (MI) and ischemic stroke (IS), the main consequences of unstable atherosclerotic lesions, are the most common causes of death worldwide. However, any recommendations available for the prevention of MI and IS are currently based solely on the efficacy of treatment at the population level, and no realistic means are currently available to tailor treatment to individual patients. Until now, personalized treatment strategies for atherosclerotic cardiovascular disease (CVD) have not been possible. Without loss of generality, other adverse consequences of atherosclerosis include various forms of aortic disease, such as claudication, amputation, and aneurysms.

[0156] In the setting of CVD, an existing biobank containing detailed disease-specific information at various morphological and molecular scales was used to create a dedicated in silico system biological model. Applications of the model include evaluation of drug side effects, exploring drug combinations, and modeling the effect of drugs and surgical interventions on specific patients. There is great value in being able to identify in advance whether an individual patient is likely to respond to a drug or not. The inclusion of extensive molecular pathway analysis offers the advantage of addressing the baseline complexity required in many clinical scenarios when measurement of molecular species in plasma or tissue biopsies is not possible.

[0157] However, incorporating molecular pathway analysis into an in silico environment requires recognition of the numerous structural and biological features that characterize unstable atheroma, with numerous different pathways intertwined with a complex set of interactions. For example, collagen fibrils provide structural stability (World Health Organization (WHO). Cardiovascular diseases (CVDs) Fact Sheet (2017), see who.int / en / newsroom / fact-sheets / detail / cardiovascular-diseases-(cvds)). Collagen degradation does the opposite (Lambin et al., Radiomics: the bridge between medical imaging and personalized medicine. Nature Reviews Clinical Oncology 14, 749-762, doi:10.1038 / nrclinonc.2017.141 (2017)). Reduction of atherogenic lipoproteins due to efflux of phospholipids and cholesterol improves stability (Lee et al., Radiomics and imaging genomics in precision medicine. Precision and Future Medicine 1, 10-31 (2017)).The endothelial to mesenchymal transition can affect tissue architecture with both stabilizing and destabilizing effects (Buckler et al., Virtual Transcriptomics: Non Invasive Phenotyping of Atherosclerosis by Decoding Plaque Biology From Computed Tomography Angiography Imaging. Arteriosclerosis, thrombosis, and vascular biology, Atvbaha121315969, doi:10.1161 / atvbaha.121.315969 (2021); Peyvandipour et al., Novel computational approach for drug repurposing using systems biology. Bioinformatics 34, 2817-2825 (2018)).Perivascular adipose tissue has been suggested to increase plaque inflammation (Nguyen et al., Identifying significantly impacted pathways: a comprehensive review and assessment. Genome biology 20, 1-15 (2019); Reda et al., Machine learning applications in drug development. Computational and structural biotechnology journal 18, 241-252 (2020); Pai et al., netDx: interpretable patient classification using integrated patient similarity networks. Molecular systems biology 15, e8497 (2019)), leading to atherosclerosis, MI, or IS (Adam et al., Machine learning approaches to drug response prediction: challenges and recent progress. NPJ precision oncology 4, 1-10 (2020)).

[0158] According to the present disclosure, given the complexity and multifactorial biological behavior of atherosclerosis, comprehensive disease modeling as presented herein required consideration of a more complete biological network than previously reported, including biological processes represented by pathway networks of molecular interactions essential for disease progression to capture sufficiently granular information, including prediction of important biological responses of the disease to different drugs.

[0159] In this disclosure, we describe a comprehensive in silico systems-in-biological model of atherosclerosis using a curated network of molecular pathways to effectively delineate and predict unstable disease. Using molecular data from subject plaque samples, we developed an integrated in silico systems-in-biological model incorporating disease-specific pathways across multiple cell types. This calibrated in silico systems-in-biological model can then be used to make therapy recommendations for individual patients. By simulating the effects of various drug treatments on molecular processes related to the stabilization of atherosclerotic lesions, we assessed the potential of the model to effectively predict personalized pharmacological effects, highlighting the potential of clinical practice and individualized therapy for the prevention or suppression of adverse events such as MI and IS.

[0160] The present disclosure also provides systems and methods for using these models to make patient-specific therapy recommendations for individual patients based solely on non-invasive arterial imaging data.

[0161] I. How to obtain phenotype / endotype / seratatype data based on virtual omics modeling Information about common biological processes related to plaque characterization and stability may be obtained non-invasively through virtual omics methods. Briefly, the method includes receiving a non-invasively obtained imaging dataset of atherosclerotic plaque from a subject, processing the non-invasively obtained imaging dataset to obtain quantitative plaque morphology data, processing the quantitative plaque morphology data with a virtual phenotype model to obtain predicted protein and / or gene expression data for the plaque from the subject, and generating phenotypic data for the atherosclerotic plaque from the subject based on the molecular data.

[0162] Phenotypic data refers to a particular set of observables of individual patients, subjects, or development subjects that result from the interaction of their genotype with the environment. In particular, phenotypic data can include endotypic data, which pertains to subtypes of disease states defined by distinct pathophysiological mechanisms, and / or theratypic data, which is used to group patients or subjects according to their response to specific alternative therapies.

[0163] Non-invasive data The first step in obtaining patient or subject data for the methods and systems described herein is to obtain the data non-invasively. For example, the data may be imaging data, i.e., images of plaque in an artery, and may be obtained by various methods well known in the art. In some embodiments, the imaging data set is obtained by a radiological method. For example, any of computed tomography (CT), dual-energy computed tomography (DECT), spectral computed tomography (spectral CT), computed tomography angiography (CTA), cardiovascular computed tomography angiography (CCTA), magnetic resonance imaging (MRI), multi-contrast magnetic resonance imaging (multi-contrast MRI), ultrasound (US), positron emission tomography (PET), intravascular ultrasound (IVUS), optical coherence tomography (OCT), near-infrared spectroscopy (NIRS), or single-photon emission computed tomography (SPECT) may be utilized. In a particular embodiment, CTA is utilized.

[0164] For example, in one embodiment, CTA can be performed as a pre-operative routine procedure in a hospital using a site-specific image acquisition protocol. CTA can be performed at 100 or 120 kVp, with CTDIvol16cm varying between 13.9 mGy and 36.9 mGy, or CTDIvol32cm varying between 7.9 mGy and 28.3 mGy. Contrast injection rates and volumes with saline boost can be used as needed. In general, a caudo-cephalic scan direction from the aortic arch to the apex can be selected using intravenous contrast. Axial image reconstructions of about 0.5 mm to about 1.0 mm, for example 0.65 mm, 0.9 mm, or 1.0 mm, can be used and transferred to a digital workstation for vascular CTA image analysis.

[0165] Variations of these examples of non-invasive imaging are contemplated and may be used by those of skill in the art.

[0166] Organizational Model Data such as imaging data obtained from the non-invasive imaging methods described herein are loaded into image processing software, e.g., ElucidVivo® (Elucid Bioimaging Inc, Boston, Massachusetts) software, which contours (segments) the lumen and outer wall of the common carotid, internal thoracic, and external carotid arteries to provide quantitative plaque morphology data. See also U.S. Pat. Nos. 10,176,408, 10,740,880, 11,094,058, and 11,087,460, each of which is incorporated herein by reference. Specifically, the software produces a fully three-dimensional segmentation of the lumen, wall, and each tissue type with an effective resolution approximately three times higher than the reconstructed voxel size, improving differentiation of soft tissue plaque components over manual inspection. The common carotid and internal thoracic arteries are defined as targets, and the lumen and wall are automatically assessed and manually edited when necessary.

[0167] The software provides vascular structural measurements, including the degree of stenosis (calculated by both area or diameter), wall thickness (the distance between the luminal border and the border of the vessel outer wall), and remodeling index (the ratio of vessel area with plaque to vessel area without plaque used as a reference). Studies in animal models and histological analysis of human plaque lesions have characterized prominent, yet common, structural and biotissue features, such as increased inflammation, accumulation of large necrotic lipid central core (LRNC), intraplaque hemorrhage (IPH), thin and rupture-prone fibrous cap due to extracellular matrix (ECM) degradation, smooth muscle cell (SMC) apoptosis, levels of calcification (CALC), matrix / fibrous tissue (MATX), and fibrous cap / perivascular adipose tissue (FC / PVAT).

[0168] The software includes algorithms to reduce blur caused by imaging in the scanner. A patient-specific 3D point spread function is adaptively determined so that image intensity is restored to more closely represent the original material imaged, which reduces artifacts such as calcium blooming and allows for the differentiation of less obvious tissue types. Specifically, image restoration is performed in conjunction with tissue characterization based on expert-annotated tissue structures (including both proteomic and transcriptomic information), as described, for example, in U.S. Pat. Nos. 10,176,408, 10,740,880, 11,094,058, and 11,087,460, each of which is incorporated herein by reference.

[0169] As shown in FIG. 2A, CTA can be processed to obtain 3D images. FIG. 2A includes four columns of images (from left to right) showing a CTA image (column A), processed images (columns B and C), and corresponding histopathological annotations (column D). Specifically, as explained above, the images in column A were processed using ElucidVivo® software to produce high-resolution, fully three-dimensional sections of the lumen, wall, and each tissue type, as shown in columns B and C of FIG. 2A. Finally, column D of FIG. 2A shows the corresponding histological sections stained with hematoxylin (LRNC, CALC), Prussian blue (IPH; arrows), and Masson's trichrome to visualize fibrous tissue (MATX).

[0170] Processing of CTA images allowed multiple objectively validated measurements to be made, thereby enabling characterization of plaque morphology by CTA analysis software. These assessments included anatomical structure ("Structure") and tissue characterization ("Composition") as shown in Figure 2B. Both Figure 2A and Figure 2B show tissues with necrotic lipid core (LRNC), calcification (CALC), intraplaque hemorrhage (IPH), matrix / fibrous tissue (MATX), and fibrous cap / perivascular adipose tissue (FC / PVAT). Without loss of generality, these specific tissue types are given as examples.

[0171] The overlapping density of tissues such as LRNC and IPH, for example, necessitates methods for accurate classification. To avoid the constraints of traditional analysis of CTA utilizing fixed thresholds, the accuracy required to reveal molecular pathways was achieved by an algorithm that takes into account the distribution of tissue components rather than assuming a constant material density range. In this way, the software maximizes criteria to mimic expert annotation using a microscope while at the same time making a mathematical judgement to interpret the Hounsfield Units (HU) of adjacent voxels by mitigating the variability of scanners, reconstruction kernels, and contrast levels. In this way, the software fundamentally addresses the subjectivity inherent in other analytical techniques.

[0172] The non-invasively obtained image data is processed by software to provide quantitative plaque morphology output information, such as anatomical data and tissue composition data, for example, anatomical data includes measurements of lumen and wall remodeling, wall thickening, ulceration, stenosis, dilation, plaque burden, or any one or more of the measurements listed in Table 1.

[0173] As outlined in Table 1, vascular structural measurements included the degree of stenosis (calculated by both area or diameter), wall thickness (the distance between the luminal border and the border of the outer vessel wall), and remodeling index (the ratio of the vessel area with plaque to the vessel area without plaque, used as a reference).

[0174] [Table 1]

[0175] Histological composition data included calcification (CALC), necrotic lipid core plaque (LRNC), intraplaque hemorrhage (IPH), and matrix / fibrous tissue (MATX), see Table 2 below.

[0176] [Table 2]

[0177] Volumetric measurements may also be utilized, either instead of or in addition to area measurements. Similarly, various forms of spatially labeled data representing these may also be used. Without loss of generality, these particular tissue types are given as examples.

[0178] 3A-3F show an exemplary embodiment of histology and non-invasive computed tomography analysis of two patients in the study cohort described in the following examples for unstable (FIGS. 3A-3C) and stable (FIGS. 3D-3F) atherosclerosis. Histology with Masson's trichrome staining of CEA samples showed extensive necrotic lipid cores with rupture of the fibrous cap in unstable lesions (FIG. 3A), whereas fibrosis and abundant collagen predominated in stable cases (FIG. 3D). The histological appearance of these two phenotypes was consistent with the results of non-invasive CTA analysis by ElucidVivo software visualized in a 3D view (FIGS. 3B and 3E) and by the classifier output for the stable phenotype (FIGS. 3C and 3F; original images are in color, red=unstable plaque features, yellow=stable plaque features, green=minimal disease). Other stains such as H&E, Movat, etc. can also be used without loss of generality.

[0179] Virtual Omics Models As described in more detail below, the virtual omics model is constructed from various machine learning models. Briefly, any of several methods, devices, and / or other features are used to perform a specific information task (such as classification or regression) using several examples of data of a given format, and then this same task can be performed on unknown data of the same type and format from a new patient or subject. The machine (e.g., computer or processor) "learns" by, for example, identifying patterns, categories, statistical relationships, etc. exhibited by the training data. The results of the learning are then used to predict whether new data will exhibit the same patterns, categories, and statistical relationships.

[0180] Examples of such models include neural networks, support vector machines (SVMs), decision trees, hidden Markov models, Bayesian networks, Gram-Schmidt models, reinforcement-based learning, genetic algorithms, and cluster-based learning. Several may be used to create a pool of trained machines from which to select. These may include methods of feature selection and pruning, feature ranking, random generation of feature sets, correlation between features, PCA (principal component analysis), ICA (individual component analysis), parameter variation, and any method known to one skilled in the art.

[0181] Supervised learning occurs when training data is labeled as reflecting the "correct" outcome, i.e., when the data belongs to a class or exhibits a pattern. Supervised learning techniques include neural networks, SVMs, decision trees, hidden Markov models, Bayesian networks, etc. A test data set encompassing known classes may be used to determine whether a trained learning machine can identify patterns in the data and / or classify the data. The test data set is preferably generated independently of the training data set. A training data set (of known or unknown classes) is used to train the learning machine. Data may be suitable for training the learning machine regardless of whether the class of the data is known or unknown. Unsupervised learning occurs when training data is not labeled as reflecting the "correct" outcome, i.e., when there is no indication in the data itself of whether the data belongs to a class or exhibits a pattern. Unsupervised learning techniques include Gram-Schmidt, reinforcement-based learning, cluster-based learning, etc.

[0182] Thus, while some embodiments of the present invention may utilize machine learning and / or deep learning methods, these methods are not always required in all embodiments.

[0183] In one embodiment, one or more neural networks may be generated and / or updated using virtual omics from vascular CT images processed as described in Figures 2A and 2B according to the virtual tissue model of Figure 6B and together with the quantitative plaque morphology data of Figure 6B and optionally additional covariates. The one or more neural networks take the 3D vascular images and provide calibration data according to the virtual representation and virtual proteomics model of Figure 6B that combines the spatially resolved signals across multiple layers with covariate information encoded as scalars (e.g., without loss of generality, vascular location, patient attributes, etc.) to produce individual patient calibration data with molecular level information utilized by the system biological model.

[0184] This method overcomes two problems. First, the amount of annotated data required for training is both low- and high-dimensional CT image volumes. The present disclosure leverages the dimensionality reduction provided by the virtual tissue model, which also brings the opportunity for objective validation. It also leverages a large amount of unlabeled blood vessels, made possible from using a validated image processing step, from which the virtual omics network can learn a rich representation of vascular structures in a semi-supervised or self-supervised manner. Second, the output is high-dimensional. It addresses this by utilizing a neural architecture that builds a common representation of the input, which is shared across components that predict individual omics levels.

[0185] In another embodiment, one or more deep learning networks can be used for prediction of adverse events and / or drug interaction effects. The general representation described herein can be imported into a new model, which uses the features it provides to predict adverse events directly or after further refinement with labeled data. These features can also be fused with multiple predictions from the system biology model to estimate drug interaction effects.

[0186] In another embodiment, a neural network can be used to perform part or all of the simulation of the therapeutic effect, and it should be noted that multiple parts of the system biological model itself may be distinguishable. A reaction rate network essentially consists of a system of coupled ODEs and PDEs that can be implemented in a neural network to enable both model training and model inference to be accelerated. A neural network can be utilized to find the preferred initial state of such a reaction network to efficiently enable an optimal solution.

[0187] Generating Phenotype, Endotype, and / or Theratype Data for Atherosclerotic Plaques Quantitative plaque morphology data received from processing of the CTA images (which relates, for example, to plaque contour, characterization, type), as described above in the section "Tissue Model," is processed against one or more hypothetical proteomic / transcriptomic models, as described above, to obtain inferred / predicted gene expression and / or protein level data for plaque from the subject. In other words, the tissue model is further processed against known gene expression and / or known protein level patterns to generate a predicted omics model (i.e., the tissue model based on the imaging data is correlated to gene expression and / or protein level patterns).

[0188] The predicted omics model then allows the clinician to predict 1) which gene transcript levels are likely to be elevated and which gene levels are likely to be decreased in plaque, and / or 2) which protein levels are likely to be elevated and which protein levels are likely to be decreased in plaque. The omics levels (elevated / decreased / unchanged) are based on patients without atherosclerosis. As a result, this data provides information about the mechanisms related to plaque pathophysiology, plaque instability, or other related biological findings, thereby generating phenotype data, endotype data, and / or theratype data for atherosclerotic plaque from subjects.

[0189] II. How to Generate an in Silico System Biological Model Creation and Training of In Silico System Biological Models In silico system biological models are initially generated or trained using two types of data. First, data is used that is empirically determined from biological samples from development subjects, people for whom actual proteomic data is available showing differentially expressed protein levels linked to the specific characteristics and morphology of plaque in each of those subjects. Second, results from searches of published literature, experimental results, and / or other databases are used to find papers etc. and obtain detailed information about the proteins in the model. These two sources of data are used to create the initial model.

[0190] An example of a mathematical framework for multi-scale analysis is given below.

[0191]

number

[0192] The function y(t) refers to the phenotype y at time t. The function x is the cellular and molecular level at time t, and z represents the patient-level outcome or condition at time t. The present disclosure provides a system of equations, or nonlinear models f and g, where f is at decreasing scale and g is at increasing scale. An example of a function f is a predictive modeling diagram, and y can be expressed as scalar data, vector data, or multidimensional data as depicted to derive expression profiles, protein concentrations, or other low-level information. An example of a function g can also be a predictive model, but at increasing scale, a different model than f. Inverse functions of f and g can also be derived.

[0193] Further details are given in FIG. 4. Here, the steps for determining the function f are outlined. The training data set is used to optimize various types of models, and the results can be further applied to supervised or unsupervised clustering. The resulting correlations can be analyzed at the cohort or individual level using techniques such as Gene Set Enrichment Analysis (GSEA) to reveal biological processes and molecular pathways at the cohort level and / or in individual patients. GSEA can be performed, for example, using EnrichR (see amp.pharm.mssm.edu / Enrichr), further passing the results from Gene Ontology biological processes, and further passing the data to other systems such as Revigo (revigo.irb.hr), for example, to determine non-redundant processes. Individual patient level inference can be applied, where both the degree of dysregulation as well as the statistical significance of the model can be considered. This can be variously described as virtual omics.

[0194] Without loss of generality, the virtual omics model itself can be utilized, for example, as follows: All or selected probes from macroarrays, or samples from mass spectrometry, or other assay methods to obtain so-called omics data can be selected. Single and multiple variable regression models, covering linear and non-linear modeling techniques, are run on predictor sets built from development cohorts, including plaque morphology, attributes, clinical (research) values, and / or other variables, to recognize that some clinical factors may affect expression data or models, to investigate what the additional value of morphology data is over clinical and attribute data, and to identify that when morphology and other variables have independent information content, different predictor sets can be used, some of which use only plaque morphology, while others also use experimental values, attribute values, and other values ​​in the synthetic model. The results of each model can be output and tabulated to identify the best achieved performance for each type.

[0195] Predictive performance can be determined based on the accuracy of prediction relative to true or reference values. Models can be built with variation, e.g., with different sets of morphometric measurements according to the physiological basis of the hypothesis, with automated optimization using cross-validation, e.g., while simultaneously varying tuning parameter values, and / or with partitioning the data such that the training set on which cross-validation was performed is strictly separated from the isolated validation data set to test performance using a lockdown model. The use of histologically validated plaque features can, for example, produce interpretable models and, when combined with cross-validation, can mitigate overfitting.

[0196] Supervised model quality (MQ) can be determined as an example, but not only by this method, as the product of two measures for each model type. The MQ for continuous estimation models was calculated as the product of the concordance correlation coefficient (CCC) and the predicted vs. observed regression slope for continuous estimation (the former is to measure the tightness of the fit, but is augmented by the latter to ensure balanced predictions against the observed). The MQ for dichotomous categorical prediction models was calculated as the product of the area under the receiver characteristic curve (AUC) and kappa for dichotomous predictions (the former is to measure the overall classification performance, but is augmented by the latter to ensure performance in both high and low expression classes).

[0197] The above can be performed using deep learning networks of various network topologies, and using raw images or enriched images identified with tissue type annotations and / or obtained from spatial normalization such as, but not limited to, unwrapping.

[0198] Recognizing the existence of various virtual omics processing steps, the present disclosure builds upon them with further steps that provide further convenience. For example, models of complex biological behaviors, sometimes called pathways or cell signaling networks, are described with mathematical descriptions using differential equations or other mathematical descriptions that capture behaviors such as mass transfer, reaction dynamics due to enzymes, various inhibition processes, and other approximations to biochemical reactions / relationships.

[0199] In general, the variables identified describe the expected behavior in a population of patients or animals, i.e., in general, they are not applicable to a particular individual. However, they provide structure and a level of calibration for a patient population. One exemplary embodiment is illustrated in FIG. 5, where reference literature and / or in vitro studies are mined or performed, respectively, to reveal terms in the system bioequation, e.g., concentrations, levels, and / or rate constants. Specifically, as shown in FIG. 5, literature is mined to identify reactions between biomolecules (left part of the figure). It should be noted that there are several software programs that can visually and programmatically represent this information, including tools such as Cell Designer (https: / / www.celldesigner.org / ), cytoscape (see cytoscape.org), etc. Without loss of generality, the specific sources and reactions shown are examples.

[0200] On the right side of Figure 5, the responses are mapped. On the top right, the relationship between TGFβ and Tregs is shown. On the center right, the relationship between TNF and foam cells, Th1, mast cells, Th17, and TACE is shown. On the bottom right, the relationship between TL-6, foam cells, smooth muscle cells, and mast cells is shown. In this way, as will be explained in more detail below, these responses are modeled together and linked in a multi-compartment system biological model, typically with compartments for other organs, plasma, etc., to the extent that they affect the development of atherosclerosis (Parton et al., New models of atherosclerosis and multi-drug therapeutic interventions. Bioinformatics 35, 2449-2457, doi:10.1093 / bioinformatics / bty980 (2018)).

[0201] The present disclosure goes beyond patient populations so that facilities can obtain individual patient-level results. As shown in FIG. 6A, the present disclosure provides a method for using virtual omics result vectors and virtual omics data from individual patients with known or suspected CVD (whether the patient is a subject for validating the method or, of course, an intended patient seen in a clinical setting that the present invention is intended to support) to train and update individual-level rate constants and concentrations, respectively, in an in silico system organism model. This has the effect of using a system organism model developed with generalized data (e.g., updated or calibrated using data from a subject) that will be further updated or calibrated for an individual patient at a given time point. This can be simulated into the future with or without additional simulations that perturb the model according to the mechanism of a given treatment candidate to identify an "untreated or baseline" state for the patient, thereby simulating the effect as if untreated or baseline, and also as if treated with a particular drug or instrument intervention, thereby generating simulation results for the likely effects (responses) of various specific treatments. Additionally, the use of outcome data compiled by machine learning or other predictive models can be combined with the pre-treatment type simulations mentioned to generate personalized patient event-free survival curves.

[0202] Specifically, first, there is the development cohort, shown on the left side of the image, as shown in Figure 6B. For the development cohort, research CTA images and clinical CTAs were fed into the modeling software. Tissue measurements performed in the first level of processing included anatomical structure and tissue characterization using the tissue modeling software, trained using pathologist-annotated samples (annotated as "Training CTA" in Figure 6B). This generated quantitative plaque morphology data. These data were then fed forward as inputs into the model to reveal molecular profiles that determine plaque phenotypes. Once plaques were profiled and established, the experimental workflow utilizes a set of cases with paired transcriptomic and / or proteomic data from microarrays in the development cohort. These truth data were used to build virtual transcriptomic and / or proteomic models in the development cohort, which were then locked down for application to isolated subjects as a validation of model capabilities (annotated as "Validation CTA" in Figure 6B).

[0203] Update of the initial in silico biological model The initial model is then updated with calibration data from subjects, such as omics data, to validate and refine the initial model. The calibration data is again based on actual specimens that show differentially expressed proteins and / or transcript levels linked to the specific characteristics and morphology of the plaques of each of those subjects. This update of the initial model results in a calibrated model. Given data from many subjects, this step ensures that the model works as intended and also reinforces the model to make it more robust.

[0204] Test data (e.g., from subjects) are fed forward to obtain information about plaque morphology and to obtain estimated gene and / or protein measurements (see the right side of FIG. 6A). This information is then fed into an in silico model as described below, and the in silico model is calibrated based on the information obtained in FIG. 6B. Specifically, as shown in FIG. 6C, information about plaque morphology and the estimated gene and / or protein measurements obtained in FIG. 6B are fed into an in silico model (Parton et al., New models of atherosclerosis and multi-drug therapeutic interventions. Bioinformatics 35, 2449-2457, doi:10.1093 / bioinformatics / bty980 (2018)). The reactions (levels of various molecules) included in the in silico model are then calibrated. Based on the calibration, modeling allows the construction of biological pathways, which can predict the levels of various molecules in the biological pathways.

[0205] More specifically, information obtained from CTA imaging is input into an in silico system biological model, which is a (set of) networks characterizing atherosclerotic cardiovascular disease, where (each) network contains nodes (each node representing a different protein) and edges between pairs of nodes (each edge representing a protein-protein interaction in a given cell type, including "self-edges" as a means to represent transcription / translation processes). Each node in the network has information representing protein levels, which can be calibrated based on data from multiple subjects (e.g., computed tomography angiography imaging data and proteomic data of plaques).

[0206] Use of Calibrated In Silico System Biological Models Then, during operation, the calibrated model is updated again, this time with patient-specific personalized data based on imaging of the patient's plaque, without the need to perform invasive blood tests or biopsies. The calibrated model is also updated with the predicted effects of two or more different therapies. The methods and systems described herein use the patient's imaging data to make therapy recommendations based on an automated comparison of two or more different therapies whose predicted effects have been programmed into the model.

[0207] For example, once the initial in silico system biological model has been calibrated, the biological pathways contained in the in silico system biological model can then be manipulated based on the mechanism of action of various drugs to simulate the ultimate outcome of treating a patient with a particular drug. Finally, the patient's survival probability can also be estimated based on the drug simulation, and the system automatically makes therapy recommendations, as described in more detail below.

[0208] III. System for generating in silico biological models of atherosclerosis Given the above, we now disclose an example of a system for generating such a system-biological model. FIG. 7A is a block diagram of an example of a system 300a for generating an in silico system-biological model of atherosclerotic cardiovascular disease. The system 301a includes an input device 340, a network 320, and one or more computers 330 (e.g., one or more local processors or cloud-based processors). The computer 330 may include a virtual-omics engine 310, a network generation engine 304, and a network calibration engine 308. In some implementations, the computer 330 is a server. In this disclosure, an "engine" may include one or more software modules, one or more hardware modules, or a combination of one or more software modules and one or more hardware modules. In some implementations, one or more computers are dedicated to a particular engine. In some implementations, multiple engines may be installed and run on the same one or more computers.

[0209] The input device 340 is configured to acquire the pathway data 302a and the subject data 302b and provide the pathway data 302a and the subject data 302b to another device via the network 320. The pathway data 302a includes biological pathways (e.g., pathway names, identifiers) associated with atherosclerotic cardiovascular disease. The subject data 302b includes data (e.g., computed tomography angiography imaging of plaque, proteomics, transcriptomics) from a plurality of subjects diagnosed with atherosclerotic cardiovascular disease. For example, the input device 340 may include a server 340a configured to acquire the pathway data 302a from a pathway database. In some implementations, one or more other input devices can access the subject data 302b acquired by the server 340a and transmit the acquired subject data 302b to the computer 330 via the network 320. Network 320 represents a computer network (different from biological networks such as first network 306 and second network 314) and may include one or more of a wired Ethernet network, a wired optical network, a wireless WiFi network, a LAN, a WAN, a Bluetooth network, a cellular network, the Internet, or other suitable networks, or any combination thereof.

[0210] The computer 330 is configured to obtain the pathway data 302a and the subject data 302b from the input device 340 and generate an in silico system biological model of the disease represented by the network. In some implementations, the computer 330 stores the pathway data 302a and the subject data 302b in a database 332 and accesses the database 332 to retrieve a desired data set. The database 332, such as a local database or a cloud-based database, can store the pathway data 302a, the subject data 302b, the first network 306, the second network 314, or other suitable data.

[0211] In some implementations, the pathway data 302a is obtained from differential expression analysis. Each pathway in the pathway data 302a includes at least one differentially expressed molecule. For example, the computer 330 obtains first molecular expression data (e.g., gene expression data, protein expression data) of a first set of subjects diagnosed with atherosclerotic cardiovascular disease and second molecular expression data of a second set of subjects without atherosclerotic cardiovascular disease. The differential expression analysis identifies molecules, e.g., RNAs, genes, or proteins, whose expression is differential between these two sets of subjects. The gene expression data is obtained from microarrays, RNA sequencing, single-cell RNA sequencing, or reverse transcriptase PCR. Without loss of generality, the protein levels can be measured by liquid chromatography-mass spectrometry (e.g., LC-MS or LS-MS / MS).

[0212] The network generation engine 304 is configured to define / train a system biological model by receiving publicly available and / or experimentally determined data, such as pathway data 302a, and generating a first network 306. The first network 306 (also called an initial network or a reference network) characterizes a baseline of the disease, since the network has not yet been calibrated using subject data 302b. In some implementations, the first network 306 is a data structure representing nodes, edges between the nodes, and information contained in each node (e.g., protein levels). In some implementations, the pathway data 302a is obtained from findings from academic literature.

[0213] The network generation engine 304 can perform one or more tasks, such as protein segregation by cell type 304a, network pruning 304b, compartmentalization 304c, and intimal network creation 304d. Protein segregation by cell type 304a identifies the cell type in which each protein-protein interaction occurs. With reference to FIG. 10, for example, protein-protein interactions in endothelial cells, macrophages, and vascular smooth muscle cells (VSMCs) are identified. Network pruning 304b removes non-proteins and proteins with missing information.

[0214] Compartmentalization 304c aims to localize proteins by assigning a compartment to each protein, including compartments for each cell type intracellularly (intracellularly for VSMCs), cell membranes, extracellular space, and blood.

[0215] Create Intima Network 304d generates an intima network that represents a topologically accurate plasma interface because it takes into account the topological relationships between compartments. The resulting intima network is referred to as the first network 306. The first network 306 includes baseline levels of proteins. Note that, without loss of generality, other integrated networks for adventitial space, media space, or perivascular space, etc., may also be used.

[0216] The virtual omics engine 310 is configured to receive subject data 302b and generate virtual omics data 312. The subject data 302b includes computed tomography angiography (CTA) imaging data of plaque from the subject, plaque morphology data, and proteomics data corresponding to the subject. As shown in FIG. 6B, molecular measurements such as protein levels (proteomics) and gene expression (transcriptomics) can be estimated based on a comparison of CTA images used in training the virtual omics engine 310 with CTA images of patients not used for training. The subject data 302b corresponds to the data used to train the virtual omics engine 310. During training, the virtual omics engine 310 identifies features in the CTA imaging data (e.g., a particular plaque morphology) that are predictive of the molecular measurements. After training, the virtual omics engine 310 is validated, for example, through a cross-validation scheme or using isolated subjects. The virtual omics data 312 represents inferred molecular measurements, such as transcript or protein levels. When measured molecular measurements are available, the measured protein levels can be used as input to the network calibration engine 308.

[0217] The network calibration engine 308 is configured to receive the first network 306 and the virtual omics data 312 and generate a second network 314. The second network 314, updated from the first network 306 using the virtual omics data 312 derived from the subject data 302b, includes disease-associated protein levels for each protein in the second network. In some implementations, the measured omics data is used to update the first network in addition to or instead of the virtual omics data. To update the first network, the network calibration engine 308 first identifies disease-associated protein levels for a set of proteins whose disease-associated protein levels are known from the virtual omics data 312. For proteins whose disease-associated protein levels are not known, the network calibration engine 308 iteratively estimates the disease-associated protein levels for the protein based on the protein's neighboring nodes in the first network. After the disease-associated protein levels of all proteins in the first network are found (either from the virtual omics data 312 or by estimation), the network calibration engine 308 outputs the second network 314. The computer 330 can store the second network in a database 332.

[0218] The computer 330 can generate rendering data that, when rendered by a device having a display, such as a user device 350 (e.g., a computer with a monitor 350a, a mobile computing device 350b such as a smartphone, or another suitable user device), can cause the device to output data including the first network 306 and the second network 314. Such rendering data can be transmitted by the computer 330 to the user device 350 over the network 320 and processed by the user device 350 or an associated processor to generate output data for display on the user device 350. In some implementations, the user device 350 can be coupled to the computer 330. In such cases, the rendered data can be processed by the computer 330, causing the computer 330 to output data visualizing, for example, the second network 314 on a user interface.

[0219] FIG. 8A is a flow chart of an example of a process 400 for generating a calibrated in silico system biological model of atherosclerotic cardiovascular disease. The calibrated in silico system biological model is an updated model from a reference (intima) model, which is built based on publicly available data or otherwise known data, such as pathway data, by using omics data from a subject. The process is described as being performed by a system of one or more computers suitably programmed according to the present specification. For example, the computer 330 of FIG. 3A can perform at least a portion of the exemplary process. In some implementations, various steps of the process 400 can be performed in parallel, in combination, in a loop, or in any order.

[0220] The system obtains a plurality of first inputs indicating biological pathways associated with atherosclerotic cardiovascular disease (402). For example, the system queries a pathway database (e.g., the Kyoto Encyclopedia of Genes and Genomes (KEGG)) to identify biological pathways associated with atherosclerotic cardiovascular disease. In some implementations, each pathway in the biological pathways includes at least one differentially expressed molecule.

[0221] To identify differentially expressed molecules, the system obtains first molecular expression data of a first set of subjects diagnosed with atherosclerotic cardiovascular disease and second molecular expression data of a second set of subjects without atherosclerotic cardiovascular disease. The system performs a differential expression analysis on the first molecular expression data and the second molecular expression data to identify differentially expressed molecules. In some implementations, the first molecular expression data and the second molecular expression data are gene expression data. In some embodiments, the first molecular expression data and the second molecular expression data are protein expression data.

[0222] The system generates a first network based on the first input (404). The first network includes nodes representing baseline levels of proteins and edges representing protein-protein interactions in one or more cell types. The first network includes proteins, genes, mRNAs, nutrients, cellular events, external signals, or combinations thereof found in biological pathways. The system represents the proteins in the plurality of first inputs as nodes in a graph (also called a state graph), initializes baseline levels for each of the proteins, represents the protein-protein interactions as edges in the graph, and outputs the graph as the first network. The baseline levels indicate the state of the nodes. The one or more cell types are associated with atherosclerotic cardiovascular disease. In some implementations, the one or more cell types include a cell type that includes at least one protein whose level is altered by atherosclerotic cardiovascular disease. The one or more cell types may include, for example, endothelial cells, vascular smooth muscle cells, macrophages, and lymphocytes. Without loss of generality, other cell types may be included.

[0223] In some implementations, each edge in the first network is directed with a weight, and a directed edge indicates the direction of protein-protein interaction, e.g., molecule A activates molecule B. The weight may indicate the type of protein-protein interaction, e.g., activation, inhibition, dissociation, methylation, glycosylation, translation, repression, modification, etc. The weight is positive for activation and translation. The weight is negative for inhibition, repression, and modification. The edges in the first network can have information indicating a dependency state, i.e., molecule A interacts with molecule B under certain conditions, e.g., when a reference level of molecule B meets a threshold. The first network can be displayed in a graphical format in a user interface, e.g., using cytoscape.

[0224] The first network includes (i) a "core network" representing protein-protein interactions specific to each respective cell type, (ii) a "mid network" representing protein-protein interactions occurring in multiple, but not all, cell types, and (iii) a "full network" representing protein-protein interactions occurring in all cell types. The edges represent protein-protein interactions that represent any one of a variety of types of interactions including, for example, activation, inhibition, indirect effect, state change, binding, dissociation, phosphorylation, dephosphorylation, glycosylation, ubiquitination, and / or methylation.

[0225] The system can separately calibrate the core network, the mid network, and the full network by using a second input to generate a calibrated sub-network. After calibration, the system generates a second network including the calibrated sub-network. Specifically, the protein-protein interaction of the i-th molecule with the j-th molecule is expressed as Σ j w(j,i)*s j (td(j,i)), where w(j,i) is the weight of the edge between the i-th molecule and the j-th molecule, and s j is the reference level of the jth molecule, t is the time step, and d(j,i) is the delay of the edge between the ith and jth molecules. The edge delay indicates the time step required for a protein-protein interaction to occur.

[0226] The system obtains a second input indicative of calibration data from a plurality of subjects diagnosed with atherosclerotic cardiovascular disease (406). The second input includes non-invasively obtained data, such as imaging data of plaque from the subject, morphological data obtained from the plaque, and proteomic data corresponding to the plaque, for each subject.

[0227] The imaging data may be acquired by computed tomography (CT), dual-energy computed tomography (DECT), spectral computed tomography (spectral CT), computed tomography angiography (CTA), cardiovascular computed tomography angiography (CCTA), magnetic resonance imaging (MRI), multi-contrast magnetic resonance imaging (multi-contrast MRI), ultrasound (US), positron emission tomography (PET), intravascular ultrasound (IVUS), optical coherence tomography (OCT), near-infrared spectroscopy (NIRS), or single-photon emission computed tomography (SPECT) diagnostic images, or any combination thereof.

[0228] If proteomic data is not available or in addition to proteomic data, the system can acquire transcriptomic data. In some implementations, the system acquires transcriptomic data for at least a portion of the subjects. The transcriptomic data is acquired by microarray, RNA sequencing (RNA-seq), single-cell RNA sequencing (scRNA-seq), reverse transcriptase PCR (RT-PCR), or any combination thereof. In some implementations, the system acquires proteomic data, e.g., protein levels obtained from protein mass spectrometry, for at least a portion of the subjects. In some implementations, the system acquires liquid chromatography mass spectrometry data of various molecules for at least a portion of the subjects.

[0229] Where omics data is available, the first network includes nodes representing baseline levels of proteins and genes and edges representing protein-protein, gene-gene, and protein-gene interactions in one or more cell types.

[0230] The system determines disease-associated protein levels for proteins in the first network from the second input 408. The disease-associated protein levels for particular proteins correspond to one or more of measured protein levels from a tissue sample from the subject, estimated protein levels based on one or more virtual omics models of the subject, or protein levels corresponding to imaging data obtained non-invasively from the subject. In different embodiments, the specific protein may be one or more of lipopolysaccharide binding protein (LBP), integrin subunit alpha 2b (ITGA2B), toll-like receptor 4 (TLR4), lipocalin 2 (LCN2), S100 calcium binding protein A8 (S100A8), S100 calcium binding protein A9 (S100A9), cyclin-dependent kinase inhibitor 1A (CDKN1A), matrix metallopeptidase 1 (MMP1), receptor for advanced glycation end products (RAGE), heme oxygenase 1 (HMOX1), SMAD family member 2 (SMAD2), and coagulation factor VIII (F8). Without loss of generality, many other molecular species are utilized by the present invention. These are not to be considered as definitive or limiting, but are given as examples.

[0231] The system identifies disease associated protein levels for the set of proteins from a second input, the disease associated protein levels for the set of proteins being obtained from a second input from the subject. The system estimates disease associated protein levels for proteins in the first network other than the set of proteins based on the disease associated protein levels of a subset of the set of proteins, the subset of the set of proteins being represented by adjacent nodes in the first network.

[0232] The system generates 410 a second network based on the first network and the disease-associated protein levels. The second network, which is an updated network from the first network using the second input, represents a calibrated in silico system biological model of atherosclerotic cardiovascular disease and includes disease-associated protein levels for each protein in the second network. To generate the second network, the system identifies disease-associated protein levels for each node whose disease-associated protein levels are obtained from calibration data from the subject, and identifies disease-associated protein levels for each node whose disease-associated protein levels are estimated.

[0233] IV. METHODS AND SYSTEMS FOR PREDICTING APPROPRIATE THERAPY / TREATMENT REGIME FOR A PARTICULAR PATIENT In general, various therapies, e.g., drug therapies and / or surgical interventions, may be used to treat cardiovascular diseases such as atherosclerosis. The in silico system biological model described herein can simulate how a real patient responds to a particular therapy (i.e., whether and to what extent the therapy has a beneficial effect) based on the mechanism of action of the particular drug therapy. In the following, an example of an embodiment is given of how a personalized therapy treatment plan can be simulated in the in silico system biological model described herein by manipulating / physically altering the levels of some molecules, e.g., RNA, DNA, or all or part of genes or proteins, in a model based on the mechanism of action of the therapy, e.g., drug therapy. Thus, the present disclosure provides a method of simulating a therapy response in a real patient by adjusting the levels of certain gene transcripts and / or protein levels in the in silico system biological model described herein.

[0234] 7B is a block diagram of an example system 300b for providing therapy recommendations for patients with known or suspected atherosclerotic cardiovascular disease based on an in silico system biomodel. System 300b includes an input device 340, a network 320, and one or more computers 330. Computer 330 may include a virtual omics engine 310, a network calibration engine 308, and a therapy response simulation engine 316. Engines not described with reference to FIG. 7 are described here.

[0235] 7B , trained on the subject data 302b, is configured to receive patient data 302c and generate virtual omics data 312. The patient data 302c includes a computed tomography angiography (CTA) imaging dataset of atherosclerotic plaques from the patient. Based on comparing the patient data 302c to the subject data 302b, the virtual omics engine 310 predicts levels of several molecules (e.g., protein levels).

[0236] The network calibration engine is configured to receive the virtual omics data 312 (e.g., predicted protein levels of a patient based on a CTA imaging dataset) and the first network 306, and generate a second network 314. As described with reference to FIG. 7A, the first network 306 is a trained in silico system biological model of atherosclerotic cardiovascular disease. The molecular levels in the first network 306 are updated based on multiple subjects, not yet for a specific patient. The network calibration engine 308 aims to update the first network 306 to generate a second network 314, which is a patient-specific network, for a given patient. To generate the second network 314, the network calibration engine 308 updates the molecular levels in the first network 306 based on the virtual omics data 312, and the updated molecular levels are referred to as personalized molecular levels. For molecules that are missing virtual omics data (i.e., molecules whose levels are not predicted by the virtual omics engine 310), the network calibration engine 308 estimates a personalized molecule level based on the personalized molecule levels of neighboring nodes in the first network. In some implementations, the network calibration engine 308 removes molecules whose molecule levels cannot be estimated.

[0237] The therapy response simulation engine 316 simulates the therapy response for each therapy candidate in the second network, which is a trained in silico system biological model calibrated for a given patient. The therapy response simulation engine 316 determines a known set of molecules affected by the therapy candidate, for example, based on published scientific discoveries regarding the mechanism of action, and determines one or more therapy effect molecular levels for each molecule in the known set of molecules (e.g., proteins, genes), for example, based on the known mechanism of action of the therapy candidate. The therapy response simulation engine 316 estimates the therapy effect molecular levels based on the simulated effect of the determined therapy effect molecular levels, and compares the determined therapy effect molecular levels and the estimated therapy effect molecular levels in the second network before and after the therapy response simulation for each therapy candidate. As an output of the therapy response simulation engine 316, a therapy recommendation 318 is generated, which is a report indicating a preferred therapy for the patient. The therapy recommendation 318 is transmitted to a user device 350, for example, a monitor 350a and a smartphone 350b. The therapy recommendations 318 may be stored in a database 332 for access and retrieval by a computer 330.

[0238] 8B is a flow chart of an example process 450 for making therapy recommendations to patients with known or suspected atherosclerotic cardiovascular disease. The process is described as being performed by one or more computer systems suitably programmed in accordance with the present specification. For example, computer 330 of FIG. 7B can perform at least a portion of the example process. In some implementations, various steps of process 450 can be performed in parallel, in combination, in a loop, or in any order.

[0239] The system receives non-invasively obtained imaging data of plaque from the patient (452). The non-invasively obtained imaging data may be obtained by computed tomography (CT), dual-energy computed tomography (DECT), spectral computed tomography (spectral CT), computed tomography angiography (CTA), cardiovascular computed tomography angiography (CCTA), magnetic resonance imaging (MRI), multi-contrast magnetic resonance imaging (multi-contrast MRI), ultrasound (US), positron emission tomography (PET), intravascular ultrasound (IVUS), optical coherence tomography (OCT), near-infrared spectroscopy (NIRS), or single-photon emission computed tomography (SPECT) diagnostic imaging, or any combination thereof.

[0240] The system accesses a trained in silico system biological model of cardiovascular disease (454). The trained in silico system biological model includes a network that characterizes cardiovascular disease. The network includes disease-associated molecular levels for each of a plurality of nodes, each node representing a different molecule, e.g., a protein or a gene or a nucleic acid. In some implementations, the network includes proteins, and the disease molecular levels represent disease-associated protein levels for the proteins and disease-associated gene levels for the genes. The network includes protein-protein interactions in one or more cell types, including endothelial cells, vascular smooth muscle cells, macrophages, and lymphocytes. In some implementations, these cell types are cell types that include at least one molecule whose levels are altered by cardiovascular disease. In some implementations, the trained in silico system biological model is a reference model constructed using publicly available or otherwise known data. In some implementations, the trained in silico system biological model is a model updated from a reference model using calibration data from a subject, as described herein.

[0241] The system updates the in silico system biological model for the patient using the personalized molecular levels derived from the non-invasively obtained data, e.g., imaging data (456). The system compares the imaging data of the patient to the imaging data of multiple subjects, which were inputs for updating the in silico system biological model. Based on the comparison, the system predicts personalized molecular levels for the molecules in the network.

[0242] The system obtains information about two or more candidate therapies for the patient, or compares one candidate therapy to a reference level (458). The candidate therapies may include, for example, (i) a lipid-lowering drug, (ii) an antidiabetic drug, (iii) an anti-inflammatory treatment, and (iv) any combination of (i)-(iii). For example, the system receives an identifier for the candidate therapies.

[0243] For example, the lipid-lowering agent may be any one or more of a statin, a proprotein convertase subtilisin kexin type 9 (PCSK9) inhibitor, or a cholesteryl ester transfer protein (CETP). The anti-diabetic agent may include, for example, metformin. The anti-inflammatory treatment may include, for example, an anti-IL1β agent, an anti-TNF agent, an anti-IL12 / 23 agent, and an anti-IL-17 agent. Without loss of generality, these treatments are given as examples.

[0244] The system simulates the therapeutic response for each therapeutic candidate in the trained in silico system biological model (460) by the following sub-processes: The system determines a known set of molecules affected by the therapeutic candidate (460a). The system determines a therapeutic effect molecular level for each molecule in the known set of molecules based on one or more known mechanisms of action of the therapeutic candidate on the known set of molecules (460b). To determine the therapeutic effect level, the system sets the therapeutic effect molecular levels of the set of proteins to a reference level. In some implementations, the reference level is determined based on the levels of molecules observed from a subject or patient without the disease, or a reference may have already been developed for the subject or patient for some form of drug therapy, where the simulation is considered to be in addition to the reference therapy.

[0245] The system estimates (460c) a therapeutic effect level for other molecules represented in the in silico system biological model other than the set of known molecules based on the simulated effect of the determined therapeutic effect level of the set of known molecules, e.g., proteins, on one or more of the other molecules represented in the network. The system determines (460d) a simulated therapeutic effect level for each molecule represented in the in silico system biological model based on the determined therapeutic effect level and the estimated therapeutic effect level. When the molecule is a protein, the therapeutic effect level is a therapeutic effect protein level. When the molecule is a gene, the therapeutic effect molecule level is a therapeutic effect gene level.

[0246] The system compares the simulated therapeutic effect levels in the in silico system biological model before and after simulating the therapeutic response to each potential therapy (462).

[0247] The system selects one or more of the candidate therapies as a preferred therapy based on the comparison (464).

[0248] The system provides a report recommending a preferred therapy to the patient (466). The report includes the predicted efficacy of the candidate therapies and the change in the therapeutic effect molecule levels before and after the therapeutic response simulation to the preferred therapy. The report can be visualized on a user interface, as shown in FIG. 25A-C. In some embodiments, the system compares the therapeutic effect levels before and after the therapeutic response simulation to only one specific therapy to determine whether the therapy has a beneficial effect on the specific patient, and if so, to what extent. This process is calculated for each of the candidate therapies, and then their respective degrees of effect, if any, are compared to select the best therapy for that specific patient.

[0249] FIG. 8C presents another implementation for making a therapy recommendation. In particular, a flowchart of an example of a process 470 for clinical decision support is presented. The process is described as being performed by a system of one or more computing devices suitably programmed according to the present disclosure. For example, the computer 330 of FIG. 7B can perform at least a portion of the process. In some implementations, the various steps of the process 470 can be performed in parallel, in combination, in a loop, or in any order.

[0250] The system operations include receiving non-invasively obtained data relating to plaque from a patient (472). For example, imaging data may be received by the system. The operations also include updating the trained in silico system biological model using personalized calibration data derived from the received data to generate an in silico patient-specific system biological model (474). The trained in silico system biological model comprises a set of networks, each network comprising a plurality of nodes, each node representing a reference molecular level, and a plurality of edges between pairs of nodes, each edge representing an intermolecular interaction. At least two of the nodes represent molecules whose levels are affected by atherosclerotic cardiovascular disease. At least one of the networks includes disease-associated molecular levels for each of the nodes in the network. In some implementations, at least a set of the networks includes nodes that respectively correspond to one or more of, for example, glycosylated low density lipoprotein (glyLDL), oxidized LDL (oxLDL), minimally modified LDL (mmLDL), or very low density lipoprotein (VLDL). Operation of a system with such nodes may also include perturbing an in silico patient-specific system biomodel to simulate, for example, the therapeutic effect of a lipid-lowering drug on a patient (476). Operation of a system with such perturbations may also include providing an output indicating a level of improvement in atherosclerotic cardiovascular disease with an exemplary lipid-lowering drug for the patient, and a recommendation to support clinical decision-making as to whether the exemplary lipid-lowering drug would be beneficial for the patient (478).

[0251] FIG. 9 illustrates an example block diagram of system components that may be used to implement the systems and methods described herein. FIG. 9 illustrates a computing device 500 that represents any one or more of various types of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. Computing device 550 is intended to represent various types of mobile devices, such as personal digital assistants, mobile phones, smartphones, and other similar computing devices. In addition, computing device 500 or 550 may include a Universal Serial Bus (USB) flash drive. A USB flash drive may store an operating system and other applications. A USB flash drive may include input / output components, such as a wireless transmitter or a USB connector that may be inserted into a USB port of another computing device. The components illustrated herein, their connections and relationships, and their functions are intended to be exemplary only and are not intended to limit the implementation of the inventions described and / or claimed in this document.

[0252] The computing device 500 includes a processor 502, a memory 504, a storage device 506, a high-speed controller 508 that connects to the memory 504 and a high-speed expansion port 510, and a low-speed controller 512 that connects to a low-speed bus 514 and the storage device 506. Each of the components 502, 504, 506, 508, 510, and 512 may be interconnected using various buses and mounted on a common motherboard or in other manners as appropriate. The processor 502 may process instructions for execution within the computing device 500, including instructions stored in the memory 504 or the storage device 506 for displaying graphical information for a GUI on an external input / output device, such as a display 516 coupled to the high-speed controller 508. In other implementations, multiple processors and / or multiple buses may be used, along with multiple memories and types of memory, as appropriate. Also, multiple computing devices 500 may be connected, with each device providing a portion of the required operations, for example as a bank of servers, a group of blade servers, or a multiprocessor system.

[0253] The memory 504 stores information in the computing device 500. In one implementation, the memory 504 is one or more volatile memory units. In another implementation, the memory 504 is one or more non-volatile memory units. The memory 504 may also be another form of computer-readable medium, such as a magnetic disk or an optical disk.

[0254] The storage device 506 is capable of providing mass storage for the computing device 500. In one implementation, the storage device 506 may be or include a computer-readable medium, such as a floppy disk device, a hard disk device, an optical disk device, a tape device, a flash memory or other similar solid-state memory device, or an array of devices, including devices in a storage area network or other configuration. The computer program product may be tangibly embedded in an information carrier. The computer program product may also include instructions that, when executed, perform one or more methods, such as those described above. The information carrier is a computer-readable or machine-readable medium, such as the memory 504, the storage device 506, or a memory on the processor 502.

[0255] The high-speed controller 508 manages bandwidth-intensive operations for the computing device 500, while the low-speed controller 512 manages less bandwidth-intensive operations. Such allocation of functions is merely exemplary. In one implementation, the high-speed controller 508 is coupled to the memory 504, the display 516, e.g., through a graphics processor or accelerator, and is coupled to a high-speed expansion port 510 that can accept various expansion cards (not shown). In this implementation, the low-speed controller 512 is coupled to the storage device 506 and the low-speed bus 514. The low-speed expansion port, which may include various communication ports, e.g., USB, Bluetooth, Ethernet, wireless Ethernet, may be coupled to one or more input / output devices, such as a keyboard, a pointing device, a microphone / speaker pair, a scanner, or to a networking device, such as a switch or router, e.g., through a network adapter.

[0256] The computing device 500 may be implemented in a number of different forms, as shown in the figure. For example, it may be implemented as a standard server 520, or multiple times in a group of such servers. It may also be implemented as part of a rack server system 524. In addition, it may be implemented in a personal computer, such as a laptop computer 522. Alternatively, components from the computing device 500 may be combined with other components in a mobile device (not shown), such as device 550. Each such device may include one or more of the computing devices 500, 550, and the entire system may consist of multiple computing devices 500, 550 in communication with each other.

[0257] The computing device 500 may be implemented in a number of different forms, as shown in the figure. For example, it may be implemented as a standard server 520, or multiple times in a group of such servers. It may also be implemented as part of a rack server system 524. In addition, it may be implemented in a personal computer, such as a laptop computer 522. Alternatively, components from the computing device 500 may be combined with other components in a mobile device (not shown), such as device 550. Each such device may include one or more of the computing devices 500, 550, and the entire system may consist of multiple computing devices 500, 550 in communication with each other.

[0258] Computing device 550 includes, among other components, a processor 552, a memory 564, and input / output devices such as a display 554, a communication interface 566, and a transceiver 568. Device 550 may also be provided with a storage device, such as a microdrive or other device, to provide additional storage. Each of components 550, 552, 564, 554, 566, and 568 are interconnected using various buses, and some of the components may be mounted on a common motherboard or in other manners as appropriate.

[0259] The processor 552 can execute instructions in the computing device 550, including instructions stored in the memory 564. The processor can be implemented as a chipset of chips including separate analog and digital processors. In addition, the processor can be implemented using any of a number of architectures. For example, the processor can be a CISC (Complex Instruction Set Computer) processor, a RISC (Reduced Instruction Set Computer) processor, or a MISC (Minimum Instruction Set Computer) processor. The processor can coordinate other components of the device 550, such as, for example, control of a user interface, applications executed by the device 550, and wireless communication by the device 550.

[0260] The processor 552 can communicate with a user through a control interface 558 and a display interface 556 coupled to a display 554. The display 554 can be, for example, a TFT (thin film transistor liquid crystal display) display or an OLED (organic light emitting diode) display, or other suitable display technology. The display interface 556 can comprise appropriate circuitry for driving the display 554 to present graphical and other information to the user. The control interface 558 can receive commands from the user and translate them for issue to the processor 552. Additionally, an external interface 562 in communication with the processor 552 can be provided to enable short-range communication of the device 550 with other devices. The external interface 562 can enable, for example, wired communication in some implementations or wireless communication in other implementations, and multiple interfaces can also be used.

[0261] The memory 564 stores information in the computing device 550. The memory 564 may be implemented as one or more of one or more computer-readable media, one or more volatile memory units, or one or more non-volatile memory units. An expansion memory 574 may also be provided and connected to the device 550 through an expansion interface 572, which may include, for example, a SIMM (single in-line memory module) card interface. The expansion memory 574 may provide extra storage space to the device 550 or may store applications or other information for the device 550. In particular, the expansion memory 574 may include instructions for implementing or supplementing the processes described above, and may also include secure information. Thus, for example, the expansion memory 574 may be provided as a security module for the device 550 and may be programmed with instructions that enable secure use of the device 550. In addition, secure applications may be provided via a SIMM card along with additional information, such as by placing identifying information on the SIMM card in an unhackable manner.

[0262] The memory may include, for example, flash memory and / or NVRAM memory, as discussed below. In one implementation, a computer program product is tangibly embodied in an information carrier. The computer program product includes instructions that, when executed, perform one or more methods, such as those described above. The information carrier is a computer-readable or machine-readable medium, such as memory 564, expansion memory 574, or memory on processor 552, which may be received, for example, via transceiver 568 or external interface 562.

[0263] Device 550 can communicate wirelessly through a communication interface 566, which may include digital signal processing circuitry if necessary. Communication interface 566 may enable communication under various modes or protocols, such as GSM voice calls, SMS, EMS, or MMS messaging, CDMA, TDMA, PDC, WCDMA, CDMA2000, or GPRS, among others. Such communication may occur, for example, through a (radio frequency) transceiver 568. In addition, short-range communication may occur, such as using Bluetooth, Wi-Fi, or other transceivers (not shown). In addition, a GPS (Global Positioning System) receiver module 570 may provide additional navigation-related and location-related wireless data to device 550, which may be used as appropriate by applications executing on device 550.

[0264] Device 550 can also communicate audibly using audio codec 560, which can receive spoken information from a user and convert it into usable digital information. Audio codec 560 can also generate audible sounds for the user, such as through a speaker in a handset of device 550. Such sounds can include sounds from a voice telephone call, can include recorded sounds, such as voice messages, music files, and the like, and can also include sounds generated by applications running on device 550.

[0265] As shown in the figure, the computing device 550 may be implemented in a number of different forms. For example, it may be implemented as a mobile phone 780. It may also be implemented as part of a smartphone 782, a personal digital assistant, or other similar mobile device.

[0266] Various implementations of the systems and methods described herein may be realized in digital electronic circuitry, integrated circuits, specially designed ASICs (application specific integrated circuits), computer hardware, firmware, software, and / or combinations of such implementations. These various implementations may include implementations in one or more computer programs executable and / or interpretable on a programmable system including at least one programmable processor, which may be special purpose or general purpose, coupled to receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0267] These computer programs (also known as programs, software, software applications or codes) contain machine instructions for a programmable processor and may be implemented in high-level procedural and / or object-oriented programming languages ​​and / or assembly / machine languages. As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and / or device used to provide machine instructions and / or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal, e.g., magnetic disks, optical disks, memory, programmable logic devices (PLDs). The term "machine-readable signal" refers to any signal used to provide machine instructions and / or data to a programmable processor.

[0268] To enable interaction with a user, the systems and techniques described herein may be implemented on a computer having a display device, e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor for displaying information to the user, as well as a keyboard and pointing device, e.g., a mouse or trackball, by which the user can provide input to the computer. Other types of devices may be used to enable interaction with the user as well. For example, feedback provided to the user may be any form of sensory feedback, e.g., visual feedback, auditory feedback, or tactile feedback, and input from the user may be received in any form, including acoustic input, speech input, or tactile input.

[0269] The systems and techniques described herein may be implemented in a computing system that includes back-end components, e.g., as a data server, or includes middleware components, e.g., an application server, or includes front-end components, e.g., a client computer having a graphical user interface or web browser through which a user can interact with an implementation of the systems and techniques described herein, or includes any combination of such back-end, middleware, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication, e.g., a communications network. Examples of communications networks include a local area network ("LAN"), a wide area network ("WAN"), and the Internet.

[0270] A computing system may include clients and servers. Clients and servers are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.

[0271] V. Types of Therapy The in silico system biology model described herein can be used to model the effect of any therapy, e.g., drug therapy or surgical therapy, whose mechanism of action is known or discovered, e.g., the mechanism of action is described in the public record or otherwise known and converted into data that can be used to update the calibrated model.The system biology model can then be updated with data representing the plaque characteristics of a particular patient, and then a specific model of a candidate therapy can be added to the system biology model updated with the information of the particular patient.The results of applying drugs to the patient-specific system biology model can be compared, and the best performing therapy can be recommended to the patient, or the therapy may not be recommended to the patient.

[0272] First, it is important to note that the drug therapy / surgical intervention listed below are only examples. Those skilled in the art will conduct literature review of drug therapy and / or surgical intervention therapy before carrying out the method described herein, and determine the parameters required to model the effectiveness of that particular drug therapy and / or surgical intervention therapy. For example, based on literature review, determine which molecules represented in the trained in silico system biological model should be manipulated and how much their levels should be changed.

[0273] A review of the current literature indicates that there are many different endotypes of atherosclerosis. For example, the endotype of increased LDL is associated with the genetic factors LDLR, PCSK9, APOE, APOB-100, SORT1, ANGPTL3, CELSR2, PSRC1, and HMGCR, and the biomarkers total cholesterol, LDL-C, ApoB, ApoB-100, ox-LDL, modified LDL, sdLDL, and PCSK9. The endotype characterized by increased Lp(a) is primarily genetically determined by the LPA locus and is not significantly influenced by other genetic, dietary, or environmental factors.

[0274] Biomarkers associated with increased Lp(a) include Lp(a), apolipoprotein isoform(a), and antibodies to Lp(a).Endotypes associated with arterial damage (arterial hypertension) are associated with genetic factors ADAMTS7, THBS2, CFDP1, NOX4, EDNRA, PHACTR1, GUCY1A3, CNNM2, and CYP17A1, and biomarkers endothelin, angiotensin, adrenomedullin, natriuretic peptides, von Willebrand factor, cell adhesion molecules, endothelial progenitor cells, endothelial microparticles, nitric oxide, and asymmetric dimethylarginine.

[0275] Endotypes characterized by inflammation are associated with the following genes: CXCL12, MCP-1, TLR, SH2B3, HLA, IL-6R, IL-5, and PECAM1, as well as the following biomarkers: TNF, IL-1b, IL-6, IL-12, IL-18, IL-23, IFN-g, IL-17, IL-22, TH17 cells, hsCRP, pentraxin 3, sCD40L, VCAM, and ICAM.

[0276] Finally, endotypes characterized by metabolic risk factors are associated with the following genes: TCF7L2, HNF1A, CTRB1 / 2, MRAS, ZC3HC1, MIR17HG, and CCDC92, as well as the following biomarkers: blood glucose, blood insulin, C-peptide, glycated hemoglobin, glycated albumin, sRAGE, and fructosamine (Vadim V. Genkel, Igor I. Shaposhnik, "Conceptualization of Heterogeneity of Chronic Diseases and Atherosclerosis as a Pathway to Precision Medicine: Endophenotype, Endotype, and Residual Cardiovascular Risk", International Journal of Chronic Diseases, vol. 2020, Article ID 5950813, 9 pages, 2020).

[0277] Examples of drug therapy In general, any suitable drug therapy is contemplated by the present application. For example, any compound that targets (e.g., inhibits) a particular gene, protein, or metabolite. "Inhibit" refers to the ability of a compound to control, interfere with, restrict, block, or regulate the function of a molecule. Exemplary compounds include small molecules, nucleic acids (e.g., interference RNA (RNAi), short interfering RNA (siRNA), micro interfering RNA (miRNA), small temporal RNA (stRNA), or short hairpin RNA (shRNA), small RNA-induced gene activation (RNAa), small activating RNA (saRNA), messenger RNA (mRNA)), inhibitory antibodies.

[0278] Dyslipidemia Management Drugs High levels of low-density lipoprotein cholesterol (LDL) are a hallmark of cardiovascular diseases such as atherosclerosis. These diseases can therefore be treated with lipid-lowering medications (e.g., intensive lipid-lowering therapy, fibrates, niacin, fish oil, statins (such as atorvastatin), ezetimibe, bile acid sequestrants, proprotein convertase subtilisin kexin type 9 (PCSK9) inhibitors, cholesteryl ester transfer protein (CETP), adenosine triphosphate citrate synthase (ACL) inhibitors, omega-3 fatty acid ethyl esters, and marine-derived omega-3 polyunsaturated fatty acids (PUFAs)).

[0279] For example, the effect that enhanced lipid-lowering drugs have on subjects can be represented in an in silico system biological model, which allows clinicians to predict whether enhanced lipid-lowering drugs will benefit patients.For example, in some embodiments, the level of LDL, for example, gene level, protein level, or both levels, is physically lowered in an in silico system biological model by 75%, 50%, 40%, 30%, 25%, 20%, 10%, or 5%, depending on what is known about how drugs affect LDL levels.For example, if it is believed in the literature that a particular drug is effective in a patient when the patient's LDL level is lowered by 25%, the model is updated to show a substantial reduction of 25%. In some embodiments, gene levels, protein levels, or both of LDL products, such as glycosylated (glyLDL), oxidized (oxLDL), and minimally modified (mmLDL), as well as VLDL, are also manipulated (i.e., lowered) in an in silico system biological model, for example, by 75%, 50%, 40%, 30%, 25%, 20%, 10%, or 5%.

[0280] Reducing the levels of these molecules in the in silico system organism model indicates changes in the levels of one or more genes, proteins, or both, as well as other molecules directly or indirectly associated with the LDL mechanistic pathway. If the in silico system organism model indicates a reduction in the likelihood of stroke or myocardial infarction, the enhanced lipid-lowering drug is considered to be beneficial to the patient. If the in silico system organism model shows no change or a worsening of one or more conditions of the patient over time, the enhanced lipid-lowering drug is not considered to be beneficial to the patient and is not recommended.

[0281] Anti-inflammatory drugs Inflammation is highly associated with atherosclerosis. Thus, therapies that inhibit IL-1, IL1β, TNF, IL12 / 23, IL17, or other drugs that affect the inflammatory cascade may be beneficial in treating subjects with atherosclerosis. Examples of therapies include colchicine, canakinumab, inhibitors of inflammatory cytokines triggered by danger signals, and resolvin precursors (e.g., omega-3 fatty acids such as eicosapentaenoic acid (EPA), docosahexaenoic acid (DHA), or docosapentaenoic acid (DPA)). However, to date, it has been difficult to identify which patients will benefit and which will not, and the latter patients may suffer dangerous side effects until or unless a likely response can be demonstrated. As a result, these drugs are not yet widely used, despite their apparent promise.

[0282] Thus, the present disclosure provides, in some embodiments, a method for simulating the effect that an anti-inflammatory drug has on a subject or patient.For example, in some embodiments, the gene level, protein level, or both of inflammatory molecules (such as, but not limited to, IL-1, IL1β, TNF, IL12 / 23, or IL17) are physically manipulated (i.e., lowered) in an in silico system biological model, for example, by 75%, 50%, 40%, 30%, 25%, 20%, 10%, or 5%, depending on what is known in the literature about how a particular drug affects inflammation.For example, if it is believed in the literature that a particular drug is effective in some patients when the patient's IL-1, IL1β, TNF, IL12 / 23, or IL17 levels are lowered by 25%, the model is updated to show a substantial decrease of 25%. Reducing the levels of these molecules in an in silico system biological model simulates genetic changes, protein changes, or both changes in other molecules that are directly or indirectly associated with inflammatory molecular pathways. In some cases, without loss of generality, the molecular levels can be increased, for example, with resolvin precursor therapy or therapy that raises HDL, for example.

[0283] Lower plaque instability is a desired therapeutic outcome. That is, if the in silico system biological model shows improved stability after simulating the effect of the anti-inflammatory drug, the anti-inflammatory drug is deemed beneficial to the subject. Plaque stability is quantified based on molecular levels. If the molecular levels of the subject are similar to those of subjects with stable atherosclerosis, the patient is likely to have relatively high plaque stability. The relative change in the subject's plaque stability before and after the anti-inflammatory drug is quantified by the change in molecular levels in the in silico system biological model.

[0284] Antidiabetic drugs Metabolic disease is associated with a significantly increased risk of developing cardiovascular disease, such as atherosclerosis. In some subjects, a critical aspect of the development and progression of cardiovascular disease is the impaired reduction of blood glucose levels. Therefore, in some cases, treatment with antidiabetic drugs may be beneficial for subjects or patients with cardiovascular disease.

[0285] Thus, the present disclosure provides, in some embodiments, a method for simulating the effect that an antidiabetic drug has on a subject. For example, in some embodiments, the gene levels, protein levels, or both of glucose / metabolism-related molecules (such as, but not limited to, MTOR, NFκβ1, ICAM1, or VCAM1) are also physically manipulated (i.e., lowered) in an in silico system biological model by, for example, 75%, 50%, 40%, 30%, 25%, 20%, 10%, or 5%, depending on what is known in the literature about how a particular drug affects glucose levels and / or metabolism. For example, if it is believed in the literature that a particular drug is effective in some patients when the patient's MTOR, NFκβ1, ICAM1, or VCAM1 levels are lowered by 25%, the model is updated to show a substantial reduction of 25%. Lowering the levels of these molecules in an in silico system biological model indicates genetic changes, protein changes, or both changes in other molecules that are directly or indirectly related to the glucose / metabolism-related molecules. If the in silico system biological model shows that the patient's diabetes levels are reduced, the antidiabetic drug is considered to be beneficial to the subject. If the in silico system biological model shows no change in diabetic symptoms or shows a worsening, the antidiabetic drug is not considered to be beneficial to the patient and is not recommended.

[0286] Other Drug Classes Other drug classes are also contemplated, for example immunomodulatory agents, such as those that are immune tolerance stimulants or enhance the activity of Tregs, or that elicit innate immunity.

[0287] Hypertensive drugs (such as ACE inhibitors) and anticoagulants (which reduce thrombin generation and / or limit thrombin activity) may also be considered.

[0288] The elicitation of innate immunity and the modulation of intracellular signaling suggest new targets for therapeutic treatment, including the inhibition of inflammatory cytokines induced by danger signals. As an example, simulating immune tolerance with enhanced Treg activity is explored. As another example, removing chylomicron remnants (large triglyceride-rich lipoproteins) protects against the formation of atherosclerosis, since chylomicron particles and triglyceride-rich particles are directly and indirectly involved in the development of plaque.

[0289] Combination Therapy In some cases, a subject may benefit from one or more combinations of the above-referenced therapies.Thus, in some embodiments, a method is provided for simulating the effect that enhanced lipid-lowering drug and anti-inflammatory drug have on a subject, the effect that enhanced lipid-lowering drug and anti-diabetic drug have on a subject, the effect that anti-inflammatory drug and anti-diabetic drug have on a subject, or the effect that enhanced lipid-lowering drug, anti-inflammatory drug, and anti-diabetic drug have on a subject.

[0290] For combination therapies, in determining the known set of affected molecules, the therapy response simulation engine 316 considers a first set of molecules affected by a first therapy, a second set of molecules affected by a second therapy, and a third set of molecules affected by the interaction of the first and second therapies. After determining the known set of molecules, the therapy response simulation engine 316 determines a therapy effect molecular level for each molecule in the known set of molecules based on the known mechanism of action of the given combination therapy. Additional steps after determining a therapy effect molecular level are described above with reference to FIG. 8B.

[0291] Surgical intervention In some embodiments, drug therapy is not an appropriate treatment plan for a given patient, and surgical intervention is the only option. If simulations in the in silico system organism model for various possible drug candidates for a given patient do not show the predicted benefit to the patient, surgical intervention should be considered. In general, surgical intervention can result in larger-scale changes than drug therapy, such as complete tissue removal represented by a widespread reduction in protein levels, or changes in anatomical structures such as the placement of a stent, which can disrupt or interfere with connections in the system organism model. In either case, there may be the addition of a localized drug, such as a drug-eluting stent (DES), which does not address the current condition but rather addresses known effects by the organism resulting from a response to the surgical intervention, which may be compensatory but may have its own undesirable side effects. Perturbations or changes may be made in the system organism model to be trained to represent various aspects of such surgical interventions.

[0292] Surgical interventions include, but are not limited to, surgery, DES, atherectomy devices, intravascular endovascular laser (IVL), drug coated balloons, tunable temperature balloons, and / or prosthetic heart valves.

[0293] Drug-eluting stents Stents may be developed for specific patient populations depending on the atherosclerotic characteristics and patient comorbidities. Diabetic patients may respond better to different drugs. In addition, if the biological characteristics of the vessel wall and the patient's response are understood prior to the intervention, potential adverse or allergic reactions to specific drugs, polymers, or metals can be determined in advance. DES generally consist of three components: a metal stent, a polymer, and a drug. Any one of these variables may affect long-term patency.

[0294] For patients with elevated stent thrombosis MI, DES with BP is probably preferred. This is further supported by the recently reported BIOSTEMI trial showing superiority of the ultra-thin BP sirolimus-eluting stent ORSIRO® over the DP everolimus-eluting stent XIENCE® in terms of TLF at 1 year. For patients at high bleeding risk, BioFreedom™ or Resolute Onyx™ with 1 month of dual antiplatelet therapy (DAPT) have the most supportive data (Comparison of Contemporary Drug-eluting Coronary Stents - Is Any Stent Better than the Others?, available at www.touchcardio.com / interventional-cardiology / journal-articles / comparison-of-contemporary-drug-eluting-coronary-stents-is-any-stent-better-than-the-others, accessed May 7, 2021).

[0295] Patients with diabetes represent a challenging cohort. Most comparative trials of different DES have not shown a difference in the effect of stent type between patients with and without diabetes. In PLTATINUM PLUS, there was no difference in the risk of the primary endpoint between those stented with PROMUS™ and those stented with XIENCE™ (3.5% vs. 3.5%, RR 1.00, 95% CI 0.62-1.60). However, in the subgroup with diabetes, XIENCE was favored (7.8% vs. 3.0%, RR 2.50, 95% CI 1.16-5.38, interaction p=0.05). However, this relationship was not seen in the 5-year follow-up data of the earlier PLATINUM trial with a similar design. A comparison of BP and PP DES in patients with diabetes was recently investigated by Bavishi et al., who focused on the latest generation of stents and included 5190 patients from 11 RCTs in a meta-analysis. After a mean follow-up of 2.7 years, there were no differences in the range of outcomes between the two stent types, including target lesion revascularization (RR 1.02, 95% CI 0.85-1.24, p=0.80) and stent thrombosis (1.66% vs. 1.83%, RR 0.84, 95% CI 0.54-1.31, p=0.45). This relationship did not differ between diabetic patients treated with insulin and those without insulin (Comparison of Contemporary Drug eluting Coronary Stents - Is Any Stent Better than the Others?, available at www.touchcardio.com / interventional-cardiology / journal-articles / comparison-of-contemporary-drug-eluting-coronary-stents-is-any-stent-better-than-the-others, accessed May 7, 2021).

[0296] Atherectomy Devices Four different methods of atherectomy were utilized for the treatment of femoropopliteal or infrapopliteal small vessel disease: plaque resection (unidirectional) atherectomy, rotational atherectomy / aspiration, laser atherectomy ablation, and orbital atherectomy.

[0297] Atherosclerotic plaque molecular signatures, morphological size and volume can determine the ability of a stent to fully deploy and remain patent in the area of ​​interest, which can affect long-term and short-term outcomes.

[0298] Understanding lipid volume, matrix ratio, degree of calcium, arc, thickness, volume, area, and their impact on long-term outcomes can help determine whether patients will respond more aggressively and whether long-term outcomes / patency will be improved when selecting a different atherectomy device for lesion preparation.

[0299] Intravascular Lesion (IVL) Atherosclerotic plaque molecular signatures, morphological size and volume can determine the efficacy of IVL within the area of ​​interest, which may affect long-term and short-term outcomes. In some cases, the power and pulse of the lithotripsy may be determined by the plaque morphology.

[0300] Drug-Coated Balloons The revascularization rate of target lesions in coronary and peripheral arterial disease may be influenced by plaque morphology and / or atherosclerotic molecular signature. Depending on the atherosclerotic characteristics and patient comorbidities, different drug-coated balloons may be developed for specific patient populations. When diabetic patients are combined with different ratios of plaque biomaterials, it can be determined which drug balloon / excipient combination is most suitable for a specific patient. The type of drug (currently either paclitaxel or sirolimus), excipient, and release timing can be adapted depending on the plaque morphology to prolong the patency of the target lesion. Highly lipid-based lesions such as in-stent restenosis can affect long-term patency and justify the use of patient-specific drugs. Highly calcified lesions may require different types of drug-coated balloons. Atherectomy and specific drug-coated balloon combinations can be selected based on the molecular signature of the plaque.

[0301] Variable Temperature Balloon Atherosclerotic lesion molecular signatures can help determine if a patient is not a good candidate (will not respond well) to drug-coated balloons or drug-eluting stents and will require alternative interventional therapies. Patients may have comorbidities or allergies to certain drugs that require a different therapeutic approach. This can avoid devastating violent reactions and long-term effects of implants that may be permanent. For certain lesion morphology characteristics, the use of "hot balloons" or "cold balloons" may be justified.

[0302] Cryoplasty combines the expansive forces of angioplasty with the simultaneous application of cold thermal energy to the arterial wall. Both mechanisms are accomplished simultaneously by filling the angioplasty catheter with nitric oxide instead of the usual contrast and saline / solution mixture. Cryotherapy has been shown to biologically alter the behavior of arterial cellular components in the benign healing process (The Next Generation PolarCath System available at evtoday.com / articles / 2018-jan supplement / the-next-generation-polarcath-system). TM System, accessed May 10, 2021).

[0303] Several scientific studies have demonstrated that this cooling process within blood vessels results in weakening plaques, promoting uniform expansion to reduce vascular injury, altering elastin fibers to reduce recoil in the vessel wall while collagen fibers remain undispersed and are able to maintain structural integrity, and inducing smooth muscle apoptosis associated with reduced neointima formation and subsequent less restenosis (The Next Generation PolarCath System, available at evtoday.com / articles / 2018-jan-supplement / the-next-generation-polarcath-system). TM System, accessed May 10, 2021).

[0304] So-called hot balloons are currently under development and have the potential to alter the morphology and thickness of the fibrous cap while reducing the neointimal hyperplasia seen with standard angioplasty balloons.

[0305] Artificial Heart Valves Understanding the molecular signature of heart valve disease phenotypes can help determine which drugs can halt and potentially reverse the disease before it progresses to an irreparable state. In addition, a patient's specific valvular disease pathology can determine the long-term efficacy and patient response to specific prosthetic heart valves (TAVR: self-expanding, balloon-expandable, or surgically implanted different valves).

[0306] Heart valves are complex three-layered structures that ensure unidirectional flow of blood. Scientists are closely investigating how the properties of the two main cell types, valvular endothelial cells (VECs) and valvular interstitial cells (VICs), and their mechanical relationship with the valve extracellular matrix promote structural integrity and age-related remodeling. Abnormal changes in VECs, VICs, and the extracellular matrix at the molecular level lead to malformation and dysfunction of the tissue as a whole. Improving our understanding of the biological properties of heart valves, the effects of cardiovascular drugs, and remodeling changes are important for the development of new therapies for heart valve diseases (Xu, S. and KJ Grande-Allen (2010). "The role of cell biology and leaflet remodeling in the progression of heart valve disease." Methodist Debakey Cardiovasc J 6(1): 2-7).

[0307] Although the clinical and pathological features of the most frequent intrinsic structural diseases affecting the heart valves are well established, the mechanisms of heart valve disease are poorly understood, and effective therapeutic candidates are underdeveloped. Major advances in the structure, function, and biological properties of native valves, as well as the pathobiology, biomaterials and biomedical engineering, and clinical management of heart valve disease have occurred over the past few decades (Schoen, FJ (2018). "Morphology, Clinicopathologic Correlations, and Mechanisms in Heart Valve Health and Disease." Cardiovasc Eng Technol 9(2): 126-140).

[0308] Surgical interventions in CAD include coronary artery bypass grafting (CABG), percutaneous coronary intervention (PCI, e.g., balloon angioplasty with or without stent placement). Procedures for valve replacement or repair, including transcatheter aortic valve replacement (TAVR), are also important, as is the need for coronary artery evaluation in the preoperative work-up.

[0309] Optimal Medical Therapy (OMT) Most subjects taking statins are prescribed relatively low doses, but there are indications that plaque requires higher strength, so various approaches exist. One approach is to increase the dose; for example, high-dose atorvastatin is often prescribed for subjects with hypercholesterolemia. There is growing agreement that hypertriglyceridemia and hypercholesterolemia are distinct (Le, NA and MF Walter, The role of hypertriglyceridemia in atherosclerosis. Curr Atheroscler Rep, 2007. 9(2): p. 110-5), and at least one recent drug (Vascepa®) is currently attracting attention.For subjects with hypertriglyceridemia, the Reduction of Cardiovascular Events with EPA-Intervention Trial (REDUCE-IT) trial (Bhatt et al., REDUCE-IT USA: Results From the 3146 Patients Randomized in the United States. Circulation, 2020. 141(5): p. 367-375; Bhatt et al., Cardiovascular Risk Reduction with Icosapent Ethyl for Hypertriglyceridemia. N Engl J Med, 2019. 380(1): p. 11-22; Bhatt et al., Reduction in First and Total Ischemic Events With Icosapent Ethyl Across Baseline Triglyceride Tertiles. J Am Coll Cardiol, 2019. 74(8): p. 1159-1161; Bhatt, DL, Reduce It. Eur Heart J, 2019. 40(15): p. 1174-1175, Bhatt et al., Effects of Icosapent Ethyl on Total Ischemic Events: From REDUCE-IT. J Am Coll Cardiol, 2019. 73(22): p. 2791- 2802, Boden et al., Profound reductions in first and total cardiovascular events with icosapent ethyl in the REDUCE-IT trial: why these results usher in a new era in dyslipidaemia therapeutics. Eur Heart J, 2019) and other clinical trials have reported improved results.Detailed quantitative studies to determine how IPE affects the tissue of the vessel wall have not yet been performed, because it has not previously been possible to noninvasively and quantitatively assess changes in plaque morphology.

[0310] Other emerging drug classes The elicitation of innate immunity and the modulation of intracellular signaling suggest new targets for therapeutic treatment, including the inhibition of inflammatory cytokines induced by danger signals (Zimmer et al., Danger signaling in atherosclerosis. Circ Res, 2015. 116(2): p. 323-40). As an example, simulating immune tolerance with enhanced Treg activity is explored (Herbin et al., Regulatory T-cell 25 response to apolipoprotein B100-derived peptides reduces the development and progression of atherosclerosis in mice. Arterioscler Thromb Vasc Biol, 2012. 32(3): p. 605-12). As another example, removing chylomicron remnants (large triglyceride-rich lipoproteins) (Rahmany, S. and I. Jialal, Biochemistry, Chylomicron, in StatPearls. 2020: Treasure Island (FL)) protects against the formation of atherosclerosis, as chylomicron particles and triglyceride-rich particles are directly and indirectly involved in the development of plaque (Tomkin, GH and D. Owens, The chylomicron: relationship to atherosclerosis. Int J Vasc Med, 2012. 2012: p. 784536).

[0311] Drug candidates in other therapeutic areas, such as immunomodulators in cancer, may have side effects such as aggravating atherosclerosis by activation of T cells in plaques that can lead to plaque rupture, but there are no accurate methods to track these effects. It is widely recognized that effective markers are needed during the development of drugs for atherosclerosis, as well as unrelated diseases, as well as post-marketing companion diagnostics.

[0312] VI. Examples of Use The present invention can be used as a clinical decision support system. The present invention provides a tool to assist clinical decision making by informing clinicians of what the likely effects are for various potential therapies, and to help discuss these options with patients. The present invention makes recommendations based on the statistical significance of likely improvements, and can compare candidate recommendations to identify those considered that bring about a greater degree of improvement than others. The recommendations can be understood as informing decisions that determine or lead to clinical actions.

[0313] Such recommendations and actions resulting from the use of the present invention allow therapy to be tailored to the individual, rather than based solely on population statistics. Currently, clinical guidelines are not able to use such diagnostic specificity, because there was no means to do so. Individuals have different genetic dispositions, environmental exposures, and different lifestyles. Both modifiable and non-modifiable risk factors affect what is best for the patient. The in silico system organism model provides a description of the disease and a way to process and calibrate it for the individual patient. This then allows the actual expected effect of therapy to be evaluated more specifically than was previously possible. The advantage is that the actual molecular level effect can be considered, rather than referring to the population as a whole, or at most to a subpopulation.

[0314] This is widely understood and is becoming more and more common in the treatment of cancer. However, cancer is generally informed by molecular diagnostics performed on biopsied tumor tissue, whereas biopsying atherosclerotic plaque tissue is not possible as this may cause undesired destruction. As a result, computer-based systems utilizing advanced technology, including forms of artificial intelligence, can extend beyond what a clinician would otherwise be able to do themselves. While tissue features are generally too complex in nature to be easily of interest to a human observer, the present invention analyzes data at a much finer level. It is the mixture of mathematical descriptions, knowledge representations, and architectures for the user interface, reporting system, and computational backbone that makes such decision support systems practical.

[0315] The usefulness of any diagnostic system must address what can be done with the information. Currently, there are many powerful therapies, both surgical, pharmaceutical, or combinations thereof, such as drug-eluting stents. By assessing personalized responses to these therapies, the present invention makes the diagnosis actionable by identifying the degree of improvement, annotating the level of improvement with the statistical significance of the calculation. These recommendations can be presented on screen based on a user interface, or in a printable PDF format that can be used among groups of clinicians or in communication with patients.

[0316] Identifying likely responses at the individual patient level Provided herein are methods and systems for identifying likely responses to candidate therapeutics at an individual patient level. More specifically, as described herein, an in silico system biological model is generated, trained, and updated to create a calibrated model. The calibrated model is then updated with patient-specific information (e.g., from virtual omics or histological analysis obtained from actual tissue and / or blood samples) to create a reference state. The in silico system biological model representing the reference state is then further updated to simulate one or more candidate therapies based on the mechanism of action of each therapy to arrive at various in silico system biological model representations of various simulated states for each candidate therapy. Based on the results, the patient is given a recommendation for an appropriate therapy or treatment plan, for example in the form of a report. The absolute pathology obtained as well as the relative improvement in pathology can be quantified and expressed as a likely response to each simulated therapy.

[0317] Quantifying actual responses at the individual patient level Also provided herein are methods and systems for quantifying actual responses at an individual patient level to potential therapeutic agents. More specifically, as described herein, an in silico system biological model is generated, trained, and updated to create a calibrated model. The calibrated model is then updated with patient-specific information (e.g., from virtual omics or histological analysis obtained from actual tissue and / or blood samples) to create a reference state. The in silico system biological model representing the reference state is then further updated to simulate each potential therapy based on the mechanism of action of each therapy to arrive at various in silico system biological model representations of various simulated states for each potential therapy. Based on the results, the patient is given a recommendation for an appropriate therapy or treatment plan.

[0318] After the patient has received the recommended treatment regimen for a period of time sufficient to elicit a therapeutic response, the in silico system biological model (i.e., the calibrated model updated with new patient-specific information) is updated with new patient-specific information (e.g., new virtual omics data) to create a model that represents a simulation of the effect of the recommended therapy (post-treatment simulation).

[0319] The baseline state is compared to the post-treatment simulation. If there is an actual improvement in the condition, the result indicates that the patient has improved under treatment, even if the specific changes in protein levels are not exactly as originally simulated. Furthermore, if the specific changes in protein levels are generally as simulated, it can be further determined that the treatment has caused improvement, and this method can be considered to be a surrogate endpoint of the treatment effect. In other words, in some embodiments, the simulation only needs to be generally correct to provide the intended utility in clinical practice.

[0320] Quantifying actual responses at the cohort level Also provided herein are methods and systems for determining actual response to a particular treatment at the patient or subject cohort level.

[0321] For example, an in silico system model can be constructed. More specifically, an in silico system biological model can be generated, trained, and updated to create a calibrated system as described herein. Then, for each patient or subject in the cohort, information from each patient (e.g., from virtual omics or histological analysis obtained from actual tissue and / or blood samples) is used to update each patient / subject's model to form a reference state. For each patient / subject in the cohort and for each therapy to be simulated, the calibrated model is perturbed based on the mechanism of action of the therapy to arrive at the simulated state.

[0322] After a period during which each patient / subject in the cohort has received the (tailored) recommended treatment, e.g., after a sufficient time to elicit a therapeutic response, the in silico system biological model, i.e., the calibrated model that has not been updated with new patient-specific information, is updated with new patient-specific information (e.g., new virtual omics or new histological analysis obtained from real tissue and / or blood samples) to create a model that represents a post-treatment simulation. If there is an actual improvement in disease state across the cohort of patients, it can be concluded that the patient improved under treatment, even if the specific changes in protein levels were not exactly as simulated. Furthermore, if the specific changes in protein levels were generally as simulated, it can be further said that the treatment caused improvement, and the method can be considered to be a surrogate endpoint of treatment efficacy. This can be carried out in the context of an observational study, a randomized clinical trial, or other study design.

[0323] Detection of contraindications at the individual patient level Methods and systems are also provided herein whereby simulated conditions are generated for each potential therapy and then contraindications are detected at an individual patient level.

[0324] For example, a method is provided herein for identifying likely responses at an individual patient level to a candidate therapeutic agent. More specifically, as described above, an in silico system biological model is generated, trained, and updated to create a calibrated system. The calibrated model is then updated with patient-specific information (e.g., from virtual omics or histological analysis obtained from real tissue and / or blood samples) to create a reference state. As also described above, the in silico system biological model representing the reference state is then further updated to simulate each candidate therapeutic based on the mechanism of action of each treatment to arrive at various in silico system biological models representing various simulated states for each therapeutic candidate. The adverse side effects in the simulated states are determined by looking at how the molecules are perturbed in the model. That is, even if there is an apparent improvement in the condition with respect to the disease state, there may be other adverse side effects to the patient besides the intended improvement.

[0325] Those other effects, once determined, may also be provided to the patient, for example, in a report.

[0326] Identification of likely adverse events, current actual toxicity, or likely future adverse events at an individual patient level Methods and systems are also provided herein whereby simulated conditions are generated for each potential treatment and then likely adverse reactions, current actual toxicities, or likely future adverse reactions are identified at the individual patient level.

[0327] For example, methods and systems are provided herein for identifying likely responses to potential therapeutics at an individual patient level. More specifically, as provided herein, an in silico system biological model is generated, trained, and updated to create a calibrated model. The calibrated model is then updated with patient-specific information (e.g., from virtual omics or histological analysis obtained from actual tissue and / or blood samples) to create a reference condition. The in silico system biological model representing the reference condition, also as described herein, is then further updated to simulate each potential therapy based on the mechanism of action of each therapy to arrive at various in silico system biological model representations of various simulated conditions for each potential therapy.

[0328] Adverse side effects are determined in simulated conditions (adverse reactions), i.e., even if there is a clear improvement in the condition with respect to the disease state, there may be other side effects that are harmful to the patient besides the intended improvement. This information can be used to change the therapy recommendation, i.e., for example, to downgrade the recommendation of a treatment that improves the disease state but may also have one or more side effects.

[0329] After a period during which the patient has received the (tailored) recommended treatment, e.g., after a sufficient time to elicit a therapeutic response, the in silico system biological model (i.e., one that has not been updated with new patient-specific information) is updated with new patient-specific information, e.g., information obtained through collection of tissue and / or blood samples from either the patient or non-invasive predictions (virtual-omics) using transcriptomics and / or proteomics and / or metabolomics, to create a model representing a post-treatment simulation.

[0330] If there is an actual improvement in the condition, it can be concluded that the patient improved under the therapy, even if the specific changes in protein levels were not exactly as simulated. Moreover, if the specific changes in protein levels were generally as simulated, it can be further determined that the therapy caused improvement, and the method can be considered a surrogate endpoint for efficacy of the therapy.

[0331] If there are side effects, it can be determined that the patient did not improve under treatment, even if the specific changes in protein levels were not exactly as simulated.

[0332] In some cases, an in silico model can be built with additional information regarding adverse events (i.e., step 1). All subsequent steps can then be repeated to determine additional improvements, side effects, or both, for altering treatments or for performing dynamic, combinatorial, multi-phase, or adaptive clinical trial designs or for managing individual patients.

[0333] A selection tool for strengthening clinical trials to "select in" cases that will increase the statistical power of the trial Also provided herein are methods and systems for creating and using screening tools for clinical trials to determine "select-in" cases. More specifically, an in silico system biological model is generated, trained, and updated to create a calibrated system, as described herein. The calibrated model is then updated with patient-specific information (e.g., from virtual omics or histological analysis obtained from real tissue and / or blood samples) to create a reference state. The in silico system biological model representing the reference state, as also described herein, is then further updated to simulate each potential treatment based on the mechanism of action of each treatment, to arrive at various in silico system biological models representing various simulated states for each potential treatment. The resulting disease state, as well as the relative improvement of the disease state, are quantified and expressed as a likely response to each simulated treatment.

[0334] A patient is selected for the trial if their likely improvement exceeds the inclusion criteria threshold. In other words, a patient is not selected for the trial if there are no other exclusion or inclusion criteria issues.

[0335] A selection tool to strengthen clinical trials by "selecting out" cases that reduce the statistical power of the trial Also provided herein are methods and systems for creating and using screening tools for clinical trials to determine "select out" cases. More specifically, as described above, an in silico system biological model is generated, trained, and calibrated to create a calibrated system. The calibrated model is then updated with patient-specific information (e.g., from virtual omics or histological analysis obtained from real tissue and / or blood samples) to create a reference state. The in silico system biological model representing the reference state, as also described herein, is then further updated to simulate each potential treatment based on the mechanism of action of each treatment to arrive at various in silico system biological models representing various simulated states for each potential treatment. Any adverse side effects (side reactions) in the simulated state are flagged, i.e., even if there is a clear improvement in the condition with respect to the disease state, there may be other side effects that are more harmful to the patient than the intended improvement.

[0336] If a patient's adverse events exceed the exclusion criteria threshold, the patient will not be selected for the trial. Otherwise, if there are no other exclusion or inclusion criteria issues, the patient will be selected for the trial.

[0337] (Example) The present invention is further described in the following examples, which do not limit the scope of the invention described in the claims.

[0338] Example 1 Creation of in silico biological models method Cohort assembly and proteomic processing A total of 22 male patients on statin therapy who underwent carotid endarterectomy (CEA) to prevent stroke due to advanced (>50% NASCET (Golriz Khatami, S. et al. Using predictive machine learning models for drug response simulation by calibrating patient specific pathway signatures. npj Systems Biology and Applications 7, 1-9 (2021))) stenosis were prospectively enrolled to express protein level differences between unstable and stable atherosclerosis, resulting in 18 patients with data from CTA, histology, and plaque proteomics for complete characterization (with three spatial scales) (see Figure 3A-F).

[0339] Below, the attributes of the study cohort are summarized in Table 3. Briefly, CEA was collected at the time of surgery and stored in a biobank, and details of specimen collection and processing were previously described. All specimens were collected with informed consent from patients, and the study was approved by an Ethical Review Board. Continuous variables are presented as median (interquartile range). No variables were found to differ significantly between stable and unstable phenotypes.

[0340] Attribute variables were summarized to characterize the cohort and identify significantly different values ​​across plaque subgroups. Categorical variables with less than 25% missing data were tabulated for rates and significance analyzed using Fisher's exact test. Continuous variables were tabulated as medians with interquartile ranges and significance analyzed by Wilcoxon nonparametric tests (using a confidence level of p = 0.05).

[0341] [Table 3]

[0342] Excised plaques were divided transversely at the most stenotic segments. The proximal half was used for protein analysis, and the distal half was fixed in 4% formaldehyde and prepared for histological analysis. Histological analysis was performed on Masson's trichrome stained sections to evaluate the presence of features of instability, such as necrotic lipid core (LRNC), intraplaque hemorrhage (IPH), fibrous cap thickness and integrity, and other factors according to the Virmani classification (Barrett, TJ Macrophages in Atherosclerosis Regression. Arteriosclerosis, thrombosis, and vascular biology 40, 20-33, doi:10.1161 / ATVBAHA.119.312802 (2020)), and symptomatic and asymptomatic patients were graded based on plaque stability (minimal, stable, or unstable), resulting in 18 patients who were appropriately matched for symptoms and plaque morphology features. Patients were further classified using analysis from CTA by ElucidVivo (Boston, MA, USA) for plaque morphology with anatomical and tissue characteristics and non-invasive plaque stability classification (see Figures 3A-3F). These methods can reveal universal biological processes related to plaque instability, as previously described (Kalluri. & Weinberg, The basics of epithelial-mesenchymal transition. J Clin Invest 119, 1420-1428, doi:10.1172 / JCI39104 (2009); Kovacic et al., Epithelial-to-mesenchymal and endothelial to-mesenchymal transition: from cardiovascular development to disease. Circulation 125, 1795-1808, doi:10.1161 / CIRCULATIONAHA.111.040352 (2012)).

[0343] LC-MS / MS analysis and protein identification Using a method previously described (Evrard, SM et al. Corrigendum: Endothelial to mesenchymal transition is common in atherosclerotic lesions and is associated with plaque instability. Nat Commun 8, 14710, doi:10.1038 / ncomms14710 (2017)), plaques from selected patients were processed for proteomic analysis. Briefly, 4 mm thick sections were taken from the proximal half of the lesion, one from the periphery and one from the center. Proteomic processing was performed using high-resolution isoelectric focusing (HiRIEF (Newby, AC et al. Vulnerable atherosclerotic plaque metalloproteinases and foam cell phenotypes. Thrombosis and haemostasis 101, 1006-1011 (2009))) with median normalization of peptide spectral match (PSM) level rates. FTMS master scan was followed by data-dependent MS / MS.Spectra were searched using MSFG+ (v10072) (Bittner et al., P6164 High level of EPA is associated with lower perivascular coronary attenuation as measured by coronary CTA. European heart journal 40, ehz746. 0770 (2019)) and Percolator (v2.08) (Antonopoulos, AS et al. Detecting human coronary inflammation by imaging perivascular fat. Science translational medicine 9, doi:10.1126 / scitranslmed.aal2658 (2017)), and the search results were grouped for Percolator target / decoy analysis. PSMs discovered at a PSM level and peptide-level FDR (false discovery rate) of 1% were used to infer gene identity, and median normalization of PSM-level rates was performed. Protein level FDR was calculated using the picked-FDR method (Rajsheker, S. et al. Crosstalk between perivascular adipose tissue and blood vessels. Curr Opin Pharmacol 10, 191-196, doi:10.1016 / j.coph.2009.11.005 (2010)).

[0344] Cellular Network Pathway Selection A systems organism model was created from a combination of proteomic pathways based on differences in plaque stability that represented end-stage disease and augmented with literature-based and database searches, e.g., from the Kyoto Encyclopedia of Genes and Genomes (KEGG) database, to ensure inclusion of earlier stages of atherosclerosis. Keywords were used to search the KEGG database (see, e.g., Table 4 below).

[0345] KEGG is a database resource for understanding the high-level functions and functions of living systems such as cells, organisms, and ecosystems from genomic and molecular-level information. It is a computational representation of a living system, consisting of molecular building blocks of genes and proteins (genomic information) and chemicals (chemical information), integrated with knowledge of the molecular schematics of interactions, reactions, and relationship networks (system information). It also includes disease and drug information (health information) as perturbations to the living system. In KEGG, the reference pathway maps of molecular interaction / reaction network diagrams are expressed in terms of KEGG Orthology (KO) groups, so that experimental evidence in a particular organism can be generalized to other organisms through genomic information. In other words, maps (such as those mentioned in Tables 5 and 6 below) are reference maps and are annotated with the identification number "mapxxxxx". These maps may then be generalized to Homo sapiens (i.e., humans) and are annotated with the identification number "hsaxxxxx". For example, map05417 refers to the lipid and atherosclerosis canonical pathway, and HSA05417 refers to the lipid and atherosclerosis pathway in Homo sapiens.

[0346] [Table 4]

[0347] Selected pathways were assigned according to their applicability to four major cell types: endothelial cells (EC), vascular smooth muscle cells (VSMC), macrophages, and lymphocytes (Table 5). In Table 5, a "1" is placed to indicate that the pathway has more than minor involvement in a given cell type. In general, pathways were considered to be either completely included or completely excluded for a cell type (Table 6). In Table 6, the table provides pathways that are common across many types of mammalian cells, including those identified. Some pathways contained cell type-specific portions. In such cases, the pathways were split before inclusion.

[0348] [Table 5A] [Table 5B]

[0349] [Table 6A] [Table 6B]

[0350] Table 7 below lists pathways important for lipid lowering. A high number in the "Lipid Importance" column means that the pathway is highly important, and a low number means that it is less important.

[0351] [Table 7]

[0352] Table 8 below lists pathways important for anti-inflammation. A high number in the "Inflammation Importance" column means that the pathway is highly important, and a low number means that it is less important.

[0353] [Table 8A] [Table 8B]

[0354] Table 9 below lists pathways important for anti-diabetes. A high number in the "Diabetes Importance" column means that the pathway is highly important, and a low number means that it is less important.

[0355] [Table 9]

[0356] To illustrate, KEGG pathway HSA05417 contains pathways specific to three of the cell types modeled in this work (EC, VSMC, and macrophages) and to the plasma compartment, in other words, pathway HSA5417 is one of the pathways that is broken down into cell type-specific fragments. Specifically, the relationship of the products from low density lipoprotein (LDL) to oxidized LDL (oxLDL), oxidized LDL (glyLDL), and minimally modified LDL (mmLDL) was identified in terms of their relationship to proteins in tissues (see Kanehisa, M.; "Post-genome Informatics", Oxford University Press (2000); Otsuka et al., Pathology of coronary atherosclerosis and thrombosis. Cardiovasc Diagn Ther 6, 396-408, doi:10.21037 / cdt.2016.06.01 (2016)).

[0357] HSA04514 ("Cell Adhesion Molecules") similarly contains pathway information for the three modeled cell types (EC, lymphocytes, and macrophages) and the content is partitioned accordingly. HSA04514 is also a pathway that is broken down into cell type specific fragments.

[0358] HSA04640, the “Hematopoietic Cell Lineage”, was partitioned to remove content that is not relevant for the cell types modeled in our work.

[0359] HSA04670, "Leukocyte Transendothelial Migration," divides the EC portion from the leukocyte portion, and two of the cell types modeled in this study were leukocytes (macrophages and lymphocytes).

[0360] HSA04931, "Insulin Resistance," is contained in both the VSMCs required for our study and in the liver not used in our study.

[0361] Similarly, some pathways contained content relating to the plasma-tissue interface, as noted.

[0362] The resulting pathway sets were synthesized into three ranges of cell networks, "Core", "Mid" and "Full", using a program to split the .kgml files by cell type. The "Core" network contained pathways unique to each respective cell type. "Mid" contained pathways shared by one other cell type. "Full" contained pathways shared by these and other human cell types that are generally relevant to mammalian cell function. Selected pathways in each range for each cell type were integrated into a cytoscape representation using BioNsi (Biological Network Simulator) (Shalhoub, J. et al. Systems biology of human atherosclerosis. Vascular and endovascular surgery 48, 5-17 (2014); Fava, C. & Montagnana, M., Atherosclerosis is an inflammatory disease, which lacks a common anti-inflammatory therapy: how human genetics can help to this issue. A narrative review. Frontiers in pharmacology 9, 55 (2018)), with edge weights disabled to allow a richer set of relationships than would otherwise be supported by BioNsi. The generated node list was then compared with available plaque protein measurements from our cohort. Proteins for which no direct experimental results were available and no incoming edges were pruned.

[0363] BioNSi is a tool implemented by the Cytoscape app for modeling biological networks and simulating their discrete-time dynamics. BioNSi includes a visual representation of the network that allows researchers to construct and set parameters and observe the network behavior under various conditions. Specifically, details about the use of BioNSi nomenclature to represent LDL products in the methods described herein include the following (this is not the way BioNSi is typically intended, but is used here as a means to represent more detailed biochemical properties as needed to support the simulations described herein): 1. The correct term is glyLDL, which reflects the glycosylation of LDL. 2. oxLDL is reflected as binding / association, not because that is the correct name, but because it has a weight of 1, which indicates that fragments are transformed and that oxLDL is the smallest fragment. 3. mmLDL is reflected as a state change, not because it is the correct name, but because it has a weight of 3, which is justified because Levitan 2010 shows that fragments are transformed and that mmLDL is a higher fragment than ox. 4. VLDL is reflected as an indirect effect, not because that is the correct name, but because it has a weight of 2, which is justified because the 93 patients with both TG and LDL had a mean LDL level of 15% (as a best effort approximation) with VLDL estimated as TG / 5.

[0364] Figure 10 shows HSA05417, "Lipid and Atherosclerosis," which includes unique pathways for three of the cell types modeled in this work (EC, VSMC, and macrophages) and good detail in the plasma compartment. Specifically, relationships that summarize products from LDL into oxidized LDL (oxLDL), oxidized LDL (glyLDL), and minimally modified LDL are identified with respect to their relationship to proteins in tissues. Adapted from the KEGG database for pathway HSA05417 (Kanehisa, M. and Goto, S.; KEGG: Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Res. 28, 27-30 (2000); Kanehisa, M; Toward understanding the origin and evolution of cellular organisms. Protein Sci. 28, 1947-1951 (2019); Kanehisa, M., Furumichi, M., Sato, Y., Ishiguro-Watanabe, M., and Tanabe, M.; KEGG: integrating viruses and cellular organisms. Nucleic Acids Res. 49, D545-D551 (2021)).

[0365] Table 10 below shows the detailed BioNSi edge mapping on import.

[0366] [Table 10]

[0367] BioNSi import also adds autoinhibitory loops (-9), but they may be removed when used without transcriptomic data or can represent transcription / translation processes when both proteomic and transcriptomic data are used.

[0368] In addition, a network for the integrated intima was created by compartmentalizing proteins into the intracellular, cell membrane, and extracellular spaces of each cell type, with distinct components for blood (see FIG. 11). Specifically, in FIG. 11, the first level targets of the reference unstable patient (patient P491) are represented in a layout that highlights compartmentalization with the cell membrane of ECs (green), macrophages (orange), VSMCs (aquamarine), and plasma with serum LDL (pink), shown to reflect the relationship of proteins in the extracellular domain. The majority of proteins are well compartmentalized, with approximately 15% localized to the extracellular domain. The intima network contained 4411 proteins, and after compartmentalization, as many as 1446 proteins were observed to be localized to multiple cell compartments (FIGS. 11 and 12).

[0369] Specifically, FIG. 12 shows the integrated intimal network of “full” coverage for an unstable patient (patient P491) in the untreated or baseline state.

[0370] Example 2 Calibrated network for each patient With the network definition thus created, the proteomic data were used to update the network using the calibration data from each patient. Approximately 50% of the proteins in the network were actually measured in the proteomic dataset. Because pathways encompass all selected protein-protein interactions in the pathway, protein level estimation for protein interactions lacking measurements in the dataset required interpolation. Two personalized networks were calibrated for each of the 18 patients in each cell type, in the integrated intima, and in each of the three ranges, including the entire database of protein-level vectors, called "exemplars," for a total of 540 personalized networks. The exemplar database showed a large variation in proteomic signatures after calibration of individual subjects, which corresponded to a range of estimated plaque instability of 39-96% in the baseline state.

[0371] The pseudocode for this algorithm is outlined as follows: -Establish true plaque phenotype from histological characteristics (minimal, stable, unstable) Loading protein levels Interpolate missing protein levels by iterating until a high similarity is achieved (cosine similarity metric as a measure of convergence): For each unpinned node: For each edge, record the proposal (overriding the weight from the outgoing edge) If the weight is negative (e.g., inhibition), an unweighted suggestion is made to moderately lower the target if the source is below average, or to correspondingly lower the target if the source is above average. Otherwise (e.g. activation), an unweighted suggestion is made to raise the target moderately if the source is below average, or correspondingly higher if it is above average. Generate a weighted average (dealing with missing values) Record the results and iterate towards global convergence (addressing the lack of convergence) Preserves protein levels

[0372] Examples of visualization of individual patient calibrator molecules are shown in Figures 13A and 13B. Figure 13A is a map (originally in color) representing molecules that had direct measurements against the EC core network. Specifically, some molecules showed high expression (or red), some showed low expression (or blue), and for some molecules, direct measurements were not available (green). Figure 13B represents the interpolated values ​​demonstrating the propagation of levels from non-interpolated proteins according to the relationship types and weights drawn from the pathway definitions.

[0373] Clustering analysis performed on the calibrated network identified proteins with large variability for each cell type and range. As an example, in the core range of EC, the proteins with the highest variability between unstable plaque, stable plaque, and minimal disease are interstitial collagenase (MMP1), lipopolysaccharide-binding protein (LBP), receptor for advanced glycation end products (RAGE), and integrin alpha-IIb (ITGA2B). In the mid-range, proteins such as TLR4 and HMOX1 also showed magnitude differences. In the mid-range network, VSMCs showed strong separation in proteins such as tumor protein (p53), mothers against decapentaplegic homolog 2 (SMAD2), and coagulation factor VIII (F8), while macrophages showed strong separation in proteins such as lipocalin 2 (LCN2), S100 calcium-binding protein (S100A8 / 9), and cyclin-dependent kinase inhibitor 1A (CDKN1A). In lymphocytes, matrix metalloproteinases (MMP1 / 9), insulin-like growth factor binding protein acid-labile subunit (IGFALS), and solute carrier family 2 (SLC2A1) were segregated, whereas integral endomembranes showed strong segregation in proteins such as SMAD2 and S100A9 across cell types, and interleukin-23 receptor (IL23R) in lymphocytes.

[0374] Specifically, Figures 14-18 are heat maps identifying the top 25 proteins with respect to variability among signatures in experimental cohorts for different cells. In each of the heat maps, the expression levels of the different proteins are shown (red for high expression, blue for low expression).

[0375] Figure 14 is a heatmap identifying the top 25 proteins with respect to variability among signatures in our experimental cohort, in this case for the mid-range network of endothelial cells. Strong separation is evident in proteins such as MMP1, TLR4, HMOX1, and others.

[0376] Figure 15 is a heatmap identifying the top 25 proteins with respect to variability among signatures in our experimental cohort, in this case for the mid-range network of VSMCs. Strong separation is evident in proteins such as TP53, SMAD2, F8, and others. Note that classical markers for cell types are not the focus, but rather proteins with large variation in levels across levels of instability.

[0377] Figure 16 is a heatmap identifying the top 25 proteins with respect to variability among signatures in our experimental cohort, in this case for the macrophage mid-range network. Strong separation of proteins is evident, such as LCN2, S100A8 / 9, CDKN1A, and others. Note that classical markers for cell type are not the focus, but rather proteins with large variation in levels across levels of instability.

[0378] Figure 17 is a heatmap identifying the top 25 proteins with respect to variability among signatures in our experimental cohort, in this case for the lymphocyte mid-range network. Strong separation is evident in proteins such as MMP1 / 9, IGFALS, SLC2A1, and others. Note that classical markers for cell type are not the focus, but rather proteins with large variation in levels across levels of instability.

[0379] Figure 18 is a heatmap identifying the top 25 proteins with respect to variability between signatures in our experimental cohort, in this case for the inner membrane mid-range network. Strong separation is evident in proteins such as SMAD2 and S100A9 (across cell types), IL23R (in lymphocytes), and several in the extracellular region. Stability clusters between unstable and minimal. Note that classical markers for cell types are not the focus, but rather proteins with large variation in levels across levels of instability.

[0380] Example 3 Treatment-dependent network perturbations Based on the identified proteins from the clustering results, plaque instability in this cohort was found to be primarily driven by a network coupled with endothelial injury, a regulated immune system response, and to some degree inflammation.As a result, treatment with enhanced lipid-lowering drugs (Sawada et al., From unbiased transcriptomics to understanding the molecular basis of Attorney Docket No. 53705-0002WO1 102 atherosclerosis. Current Opinion in Lipidology 32, 328-329, doi:10.1097 / mol.0000000000000773 (2021)) and treatment with IL1β antagonists as exemplary anti-inflammatory drugs (Alimohammadi et al., Development of a Patient-Specific Multi-Scale Model to Understand Atherosclerosis and Calcification Locations: Comparison with In vivo Data in an Aortic Dissection. Front Physiol 7, 238, doi:10.3389 / fphys.2016.00238 (2016)), and treatment with an antidiabetic drug that is hypothesized to be effective in treating atherosclerosis (Corti, A. et al. Multiscale Computational Modeling of Vascular Adaptation: A Systems Biology Approach Using Agent-Based Models. Front Bioeng Biotechnol 9, 744560, doi:10.3389 / fbioe.2021.744560 (2021); Casarin et al., A Computational Model-Based Framework to Plan Clinical Experiments - an Application to Vascular Adaptation Biology. Comput Sci ICCS 10860, 352-362, doi:10.1007 / 978-3-319-93698-7_27 (2018)).

[0381] Intensive lipid-lowering drug treatment was modeled by lowering patients' LDL levels by 25%, which is constrained by the minimum value to represent the clinically reported effects of such therapy (Morgan et al., Mathematically modelling the dynamics of cholesterol metabolism and ageing. Biosystems 145, 19-32, doi:10.1016 / j.biosystems. 2016.05.001 (2016)). For plasma lipids, LDL products were modeled, including glycosylated (glyLDL), oxidized (oxLDL), and minimally modified (mmLDL), and VLDL (Otsuka et al., Pathology of coronary atherosclerosis and thrombosis. Cardiovasc Diagn Ther 6, 396-408, doi:10.21037 / cdt.2016.06.01 (2016)). Details of LDL production are outlined above.

[0382] 19A-19B are illustrations of the intimal model in the "core" region before and after simulation of treatment with enhanced lipid-lowering drugs. The "untreated or baseline" panel shown in FIG. 19A shows protein levels after calibration for the unstable patient in FIG. 3A and FIG. 3D. LDL is at the center of the layout, and both direct and indirect effects of lowering LDL levels with the simulated therapy can be identified. Simulations with enhanced lipid-lowering drugs demonstrated changes in protein levels resulting from lowering LDL levels and their end products (e.g., oxLDL), both for directly affected proteins and for effects propagating through the network. Enhanced lipid-lowering drugs were found to lower the levels of many proteins related to plaque instability, but increase several proteins presumed to confer stability (FIG. 19B).

[0383] Anti-inflammatory treatment was modeled by keeping IL1β levels at the lowest level observed across proteins in the dataset. Anti-diabetic treatment was modeled by keeping MTOR, NFKβ1, ICAM1, and VCAM1 (based on the proven effects of metformin) at the lowest level observed across proteins in the dataset (Ally et al., Role of neuronal nitric oxide synthase on cardiovascular functions in physiological and pathophysiological states. Nitric Oxide 102, 52-73 (2020); Parton et al., New models of atherosclerosis and multi-drug therapeutic interventions. Bioinformatics 35, 2449- 2457, doi:10.1093 / bioinformatics / bty980 (2018)). "Lowest level" refers to the lowest number of subjects' data across multiple molecules, as determined according to the process.

[0384] Results from this particular example showed that simulation with enhanced lipid-lowering therapy was generally the most effective in reducing plaque instability, with only slight improvements in the simulated combination therapy. Anti-inflammatory and anti-diabetic therapies produced variable results from patient to patient, with overall inferior performance compared to enhanced lipid-lowering therapy. Combination therapy that included enhanced lipid-lowering and anti-diabetic drugs was generally best for patients who started with a highly unstable proteomic signature. This example shows that the present invention may be an effective strategy for selected patients. Moreover, the fact that some initially unstable patients showed no clear response to the simulated drug therapy suggested that this modeling approach could identify individuals who would be best treated with surgery rather than drugs. Patients with initially stable signatures showed less improvement with simulated therapy, with standard drug therapy alone providing sufficient prophylactic benefit. In addition, some patients who started with an unstable signature likely did not benefit from the simulated medical therapy and should undergo prophylactic surgery, suggesting that this modeling approach may identify high-risk individuals to improve decision-making between surgical intervention and medical therapy. Personalized patient treatment recommendations varied widely from patient to patient, highlighting the importance of individual predictions and more refined patient stratification, as enabled by the defined systems organism model of our study. Given the dominant inflammatory proteomic signature of unstable plaques, the modest effect observed by simulation with anti-inflammatory therapy deserves consideration. This finding could be due to the fact that, although only a single dose of treatment was simulated, effective inhibition of inflammatory pathways requires not only the sustained presence of an antagonist, but also a reduction of the cause. In addition, the chosen treatment targets IL1β, as this strategy has been shown to be effective at a population level and even more so in subgroups with enhanced systemic inflammation that may have been under-represented in the obtained model since they were not included in our cohort.In different cohorts or settings, the response to anti-inflammatory treatment may exceed that of intensive lipid-lowering drugs for patients with CVD as a comorbidity rather than a primary indication. Nevertheless, to be clinically applicable, the model should ideally capture such phenotypes. Inclusion of patients from these subgroups would improve efficacy, and if necessary, the model could be modified using an indicator of enhanced systemic inflammation, such as CRP. In both cases, the demonstration that the efficacy of the combination therapy was superior to that of intensive lipid-lowering therapy alone suggests that the modeling approach of this study can adequately simulate the effects of drugs targeting different pathways in the pathophysiology of the disease.

[0385] Predicting subject-specific drug responses Responses to drugs were then simulated in silico. In our study, the first category of simulated treatments was an intensive lipid-lowering drug, an anti-inflammatory drug (i.e., canakinumab), an antidiabetic drug (i.e., metformin), and a combination of an intensive lipid-lowering drug and an antidiabetic drug.

[0386] Two control simulations for each subject were also calculated as checks against the mathematical description to prevent unintended design or coding flaws. The first control simulation did not represent a change in treatment, and the expected outcome was derived as if it were the treatment and had gone through the same simulation, the same as in the reference case. If the output had been found to be different from the reference case, a logical or mathematical error would have been detected. The second control simulation was named "multiple insult", which simulated a "perfect storm" state of atherosclerosis risk factors that cause known disease progression. In this control, the expected outcome is that there should be a decrease in stability roughly proportional to the original stability, i.e., the further the subject's initial state is from these deleterious states, the worse the relative effect should be. If this did not happen, a logical and / or mathematical error would have been detected.

[0387] Multilevel analysis of simulated treatment effects Simulated treated and baseline conditions were assessed using multilevel analysis. Mean absolute cohort-level instability showed consistent estimates across cell types and ranges. Inter-individual variability is shown in Figure 20, where the mean effect sets the intercept and individual variability is defined by patient-specific effects.

[0388] Moreover, the distribution of absolute baseline instability showed a wide range across the experimental cohorts (Figures 21A-G). Specifically, in Figures 21A-G, each line indicates a particular therapy, and the response is shown as a dot for each network range. Each panel represents either the baseline condition (Figure 21A), or the simulated outcome after the network has been perturbed to reflect the effect (Figures 21B-G). The use of multiple ranges is shown because each represents a different sensitivity or specificity to the simulated effect. Too high a sensitivity can result in false positive results, which are mitigated by networks with higher ranges, but higher range networks may miss effects given that the set of pathways is more inclusive. Higher numbers indicate "more" unstable, i.e., lower instability is desirable from the subject or patient's perspective. The plots are shown as examples, and other network ranges or treatment candidates should be understood without loss of generality.

[0389] Additionally, results showed average relative treatment effects across cohorts, cell types, and network coverage (positive indicates reduced instability improved by treatment) (Figures 22A-F). Specifically, Figures 22A-F are plots each showing a different way of expressing the data that is also shown in absolute charts, with one that better visualizes the changes and not just the overall effect of treatment. The plots are shown as examples and should be understood without loss of generality to represent network coverage or treatment candidates.

[0390] In Figures 21 and 22, panels represent the results for each simulated therapy as well as the computational control. Each curve plots the absolute instability (Figure 21) and relative improvement (Figure 22) in each cell type and range. In general, it can be seen that the core range networks tend to show a greater response to the therapy than the full range networks, which is expected based on the assignment of pathways such that the more comprehensive the network, the less sensitive it is to perturbations. Similarly, different cell types respond differently based on the nature of the mechanism of action of the therapy and its impact on different types. Multilevel statistical analysis uses the differences in response by cell type and range to determine the dominance or certainty of the results and calculates the effect size based on the values ​​of the various responses, in order to establish a robust response calculation that is less sensitive to individual molecular level errors or lack of biological knowledge about some pathways and their assignment to cell types.

[0391] Multilevel analyses across cell types and ranges also yielded consistent estimates of mean absolute cohort-level response to treatment in mathematical controls.

[0392] Treatment effects ranged from a 20% improvement to no improvement. Not only did improvements vary from patient to patient, but the range of improvement observed differed based on how instability was estimated. Improvements in clinically symptomatic patients ranged from -8% to +20% and for asymptomatic patients ranged from -22% to +13%, but these ranges narrowed to -2% to +20% in patients with relatively unstable protein levels and -22% to +7% in patients with more stable protein levels. This brings two important implications. First, it creates an incentive to be able to distinguish between given patients rather than populations, and second, it is a crucial improvement over the standard clinical practice of using symptoms to guide treatment.

[0393] Intensive lipid-lowering therapy had the strongest effect, especially in patients who started with unstable plaque signatures and morphology. The simulated treatment predictions showed clear variability between subjects. For example, patients P491 and P773, initially characterized by highly unstable proteomic signatures, are predicted to have the best effect of treatment simulation. Indeed, simulations with intensive lipid-lowering therapy exceeded other monotherapies, while anti-inflammatory and anti-diabetic therapies provided improvements, as well as simulations with combination therapies surpassed the benefit of intensive lipid-lowering therapy (Table 11, Figures 23 and 24).

[0394] Table 11 below shows the absolute and relative improvement for baseline and treated cases. Bold patient IDs were annotated as unstable using histological characteristics and clinical symptoms. Key: Bas=Baseline, ILL=Enhanced lipid lowering, -IL1B=Anti-IL1B (anti-inflammatory), Met=Metformin (anti-diabetic), Comb=Combination, Imp=Improved. p-values: ****<0.0001, ***<0.001, **<0.01. Each patient is represented as a row with a quantitative assessment of absolute instability for baseline and each simulated state, followed by a quantitative relative improvement. The relative improvement cells are based on the significance of the improvement as judged by the overall decrease in instability. +7% or more indicates a statistically significant improvement over baseline, -5% to +6% indicates no statistically significant effect, and -7% or less indicates a statistically significant worsening over baseline state. Rows are ordered by baseline instability. Patients P834, P821, P298, P187, and P491 (all classified as unstable by histological criteria) were deemed to benefit most from treatment, with all of these patients going from very unstable to stable after treatment. Patients P853, P450, and P737 represent highly unstable plaques indicative of no therapeutic benefit, and these cases may be considered to benefit most from surgical intervention, given their highly unstable phenotype and lack of improvement with medical therapy. Patients P472, P265, and P682 represent patients for whom medical therapy may not be helpful or even necessary, given the stability of their plaques.

[0395] [Table 11]

[0396] The radar chart shown in FIG. 23 represents the degree of absolute atherosclerotic plaque stability. Four treatment candidates and two mathematical controls were simulated for each patient. The outer section (or green) represents the protein level signature with minimal disease, light gray (or yellow) represents stable plaque, and dark gray (or red) represents unstable plaque. Treatments included enhanced lipid-lowering, anti-inflammatory, and antidiabetic drugs, as well as a combination of enhanced lipid-lowering and antidiabetic drugs. The two controls were performed to remove mathematical errors in the model, as well as to simulate the expected effect of multiple insults representing maximum disease progression. Each of these conditions was plotted as an absolute effect on stability.

[0397] The radar chart shown in FIG. 24 represents the relative improvement after treatment simulation. For each patient, four treatment candidates and two mathematical controls were simulated. The outer light grey area (or green) indicates protein level signatures that result in improved stability, while the inner dark grey area (or red) indicates decreased stability. Treatments included enhanced lipid-lowering, anti-inflammatory, and antidiabetic drugs, as well as a combination of enhanced lipid-lowering and antidiabetic drugs. The two controls were performed to exclude mathematical errors in the model, as well as to simulate the expected effect of multiple insults representing maximum disease progression. Each of these conditions was plotted as a relative improvement compared to the untreated or baseline condition.

[0398] Second, a relatively clear threshold of approximately 76% absolute instability level was observed, at which subjects with higher instability at baseline showed benefit from intensive lipid-lowering drugs and further improved with combination therapy, but did not benefit when anti-inflammatory or anti-diabetic drugs alone were simulated.

[0399] One set of patients initially characterized as highly unstable did not show any response to simulated medical therapy. Importantly, this finding suggested that the modeling approach of this study could identify high-risk individuals who are more suitable for treatment with surgery rather than medical therapy. Furthermore, we found that patients who initially had more stable plaque signatures did not improve regardless of the simulated treatment category. In this proof-of-concept setting, only combination or intensive lipid-lowering therapy had an overall beneficial effect on stability, with combination therapy providing additional benefit for many patients.

[0400] Complete results, including a summary of the mean effects, confidence intervals, and estimates of attributable variance, are provided in Tables 12 and 13, shown below.

[0401] [Table 12]

[0402] [Table 13]

[0403] A personalized treatment recommendation was then assembled for each patient based on the in silico results using an automated decision algorithm incorporating simulations of different drug therapies. The recommendation combined the level of instability achieved with the selected drug choice with the control, and a text statement was automatically generated to reflect the best therapy for that patient (FIG. 25A-C). Specifically, FIG. 25A-C show personalized patient treatment recommendations for three exemplary patients that can be printed by a clinical decision support system incorporating techniques from this study. Such printed or digital recommendations can be used in patient-physician consultations. The recommendations generated by the software include one or more of the following: individual absolute and relative radar plots, statements about the benefits obtained from drug therapy, and one or two heat maps representing protein signatures of the treated state and of the untreated or baseline state.

[0404] Patient "John Doe" is an example of a patient with a highly unstable initial condition that can be reliably improved by drug therapy (Figure 25A). Simulated treatment for patient P491 demonstrated a statistically significant benefit for the combination therapy. The top five baseline protein levels matched four of the five unstable specimens and one stable specimen, providing strong support for an unstable condition. After the recommended treatment, two minimal disease specimens, two stable specimens, and only one unstable specimen matched, reflecting improvement with treatment.

[0405] Patient "Bill Smith" represents a patient starting from a more stable initial state where no drug therapy is recommended (FIG. 25B). Patient 265's baseline protein levels are consistent with four minimal disease specimens and one stable specimen, indicating stability with no improvement after simulated treatments. The recommended therapy is to maintain current therapy, not any of the simulated therapies.

[0406] Patient "David Jones" represents a patient who receives only modest improvement from medical therapy, but based on a highly unstable starting point, surgical intervention is selected as the best course of action (Figure 25C). Simulated treatments for patient P450 demonstrated a statistically significant benefit for the combination therapy.

[0407] Heat maps for specific protein level signatures are also included, including protein expression relative to the baseline state, with the treated state added in cases where a statistically significant improvement with treatment was found. The extent to which this is clinically significant is determined from the difference in clinical picture. Treatments show strength according to the difference between asymptomatic and symptomatic. Results for P491 and P265 show the extent to which the simulation capabilities may be applied (Figures 25A-C).

[0408] Other embodiments Although the present invention has been described with reference to detailed descriptions thereof, it should be understood that the above description is intended to illustrate, rather than limit, the scope of the invention, which is defined by the appended claims. Below are numbered embodiments which are intended to further illustrate, rather than limit, the scope of the invention.

[0409] 1. A method for making a therapy recommendation for a patient with known or suspected atherosclerotic cardiovascular disease, comprising: receiving non-invasively obtained data of plaque from the patient; accessing a systems biological model of atherosclerotic cardiovascular disease, where (i) the systems biological model represents a plurality of pathways associated with atherosclerotic cardiovascular disease and (ii) the systems biological model includes disease-associated molecular levels for each molecule in the systems biological model; updating the systems biological model using personalized molecular levels derived from the non-invasively obtained data from the patient to generate a patient-specific systems biological model; obtaining information regarding one or more therapy candidates for the patient; updating the patient-specific systems biological model with information regarding an intended effect of each therapy candidate; simulating a therapy response to each therapy candidate in the systems biological model to obtain a simulated therapy effect for each therapy candidate; comparing the simulated therapy effects in the systems biological model before and after the therapy response simulation for each therapy candidate; selecting one or more therapy candidates as preferred therapies based on the comparison; and providing a report recommending the preferred therapy for the patient.

[0410] 2. The method of embodiment 1, wherein the step of simulating the therapeutic response comprises setting, in at least one network, a reduced molecular level for plaque instability and setting an increased molecular level for plaque stability.

[0411] 3. The method of embodiment 1, wherein the step of updating the system biological model using personalized molecular levels further comprises the step of using disease gene transcription levels derived from non-invasively obtained data.

[0412] 4. The method of embodiment 1, wherein the non-invasively obtained data is imaging data.

[0413] 5. The method of embodiment 4, wherein the imaging data is radiology imaging data.

[0414] 6. The method of embodiment 5, wherein the radiology imaging data is obtained by computed tomography (CT), dual-energy computed tomography (DECT), spectral computed tomography (spectral CT), computed tomography angiography (CTA), cardiovascular computed tomography angiography (CCTA), magnetic resonance imaging (MRI), multi-contrast magnetic resonance imaging (multi-contrast MRI), ultrasound (US), positron emission tomography (PET), intravascular ultrasound (IVUS), optical coherence tomography (OCT), near-infrared spectroscopy (NIRS), or single-photon emission computed tomography (SPECT) diagnostic imaging, or any combination thereof.

[0415] 7. The method of embodiment 4, further comprising processing the non-invasively obtained imaging data to obtain quantitative plaque morphology data, including anatomical structure data, tissue composition data, or both.

[0416] 8. The method of embodiment 7, wherein the anatomical data comprises data regarding any one or more levels of remodeling, wall thickening, ulceration, stenosis, dilation, or plaque burden.

[0417] 9. The method of embodiment 7, wherein the tissue composition data comprises data regarding any one or more levels of calcification, necrotic lipid core (LRNC), intraplaque hemorrhage (IPH), matrix, fibrous cap, or perivascular adipose tissue (PVAT).

[0418] 10. The method of embodiment 1, wherein the pathways are compartmentalized into cell-specific networks.

[0419] 11. The method of embodiment 10, wherein the cell-specific network includes at least an endothelial cell network, a macrophage network, and a vascular smooth muscle cell network.

[0420] 12. The method of any one of the preceding embodiments, wherein the candidate therapy is a dyslipidemia management agent.

[0421] 13. The method of embodiment 12, wherein the dyslipidemia management medication is a high-dose statin.

[0422] 14. The method of embodiment 13, wherein the high-dose statin is atorvastatin.

[0423] 15. The method of embodiment 12, wherein the dyslipidemia management medication is an enhanced lipid-lowering medication.

[0424] 16. The method of embodiment 15, wherein the enhanced lipid-lowering agent is a proprotein convertase subtilisin kexin type 9 (PCSK9) inhibitor or a cholesteryl ester transfer protein (CETP) inhibitor.

[0425] 17. The method of embodiment 12, wherein the dyslipidemia management drug is a hypertriglyceridemia-lowering drug or a hypercholesterolemia-lowering drug.

[0426] 18. The method of any one of embodiments 1 to 11, wherein the candidate therapy is a drug that affects the inflammatory cascade.

[0427] 19. The method of embodiment 18, wherein the agent that affects the inflammatory cascade is an anti-inflammatory agent.

[0428] 20. The method of embodiment 19, wherein the anti-inflammatory agent is an inhibitor of IL-1.

[0429] 21. The method of embodiment 20, wherein the inhibitor of IL-1 is canakinumab.

[0430] 22. The method of embodiment 19, wherein the anti-inflammatory agent inhibits the activity of TNF.

[0431] 23. The method of embodiment 19, wherein the anti-inflammatory agent inhibits IL12 / 23.

[0432] 24. The method of embodiment 19, wherein the anti-inflammatory agent inhibits IL17.

[0433] 25. The method of embodiment 18, wherein the agent that affects the inflammatory cascade is an inhibitor of inflammatory cytokines induced by danger signals.

[0434] 26. The method of embodiment 18, wherein the agent that affects the inflammatory cascade is a resolvin precursor.

[0435] 27. The method of embodiment 26, wherein the resolvin precursor is an omega-3 fatty acid.

[0436] 28. The method of embodiment 27, wherein the omega-3 fatty acid is eicosapentaenoic acid (EPA), docosahexaenoic acid (DHA), or docosapentaenoic acid (DPA).

[0437] 29. The method of any one of embodiments 1 to 11, wherein the candidate therapy is an immunomodulatory agent.

[0438] 30. The method of embodiment 29, wherein the immunomodulatory agent triggers innate immunity.

[0439] 31. The method of embodiment 29, wherein the immunomodulatory agent is an immune tolerance stimulant.

[0440] 32. The method of embodiment 31, wherein the immune tolerance stimulant increases the activity of Tregs.

[0441] 33. The method of any one of embodiments 1 to 11, wherein the candidate therapy is an antihypertensive agent.

[0442] 34. The method of embodiment 33, wherein the antihypertensive agent is an ACE inhibitor.

[0443] 35. The method of embodiment 18, wherein the candidate therapy is an anticoagulant.

[0444] 36. The method of embodiment 35, wherein the anticoagulant reduces thrombin generation and / or limits the activity of thrombin.

[0445] 37. The method of any one of embodiments 1 to 11, wherein the candidate therapy is a regulator of intracellular signaling.

[0446] 38. The method of any one of embodiments 1 to 11, wherein the candidate therapy is an antidiabetic agent.

[0447] 39. The method of embodiment 38, wherein the antidiabetic agent is metformin.

[0448] 40. The method of any one of embodiments 1 to 11, wherein the candidate therapy is a drug-eluting stent.

[0449] 41. The method of embodiment 40, wherein the drug-eluting stent is coated with a drug that inhibits cell cycle progression by inhibiting DNA synthesis.

[0450] 42. The method of any one of embodiments 1 to 11, wherein the candidate therapy is a drug-coated balloon.

[0451] 43. The method of embodiment 42, wherein the drug-coated balloon is coated with a drug that inhibits neointimal growth by delivering an anti-proliferative substance to the vessel wall.

[0452] 44. The method of any one of embodiments 1 to 11, wherein the candidate therapy is a combination of one or more of a lipid-lowering agent, an anti-inflammatory agent, and an anti-diabetic agent.

[0453] 45. The method of any one of embodiments 1 to 42, wherein the method further comprises a step of quantifying the patient's actual response to each potential therapy.

[0454] 46. ​​The method of any one of embodiments 1 to 43, wherein the method further comprises detecting one or more candidate contraindications associated with each candidate therapy.

[0455] 45. The method of any one of embodiments 1 to 44, wherein the method further comprises identifying likely side effects for each potential therapy.

[0456] 48. The method of any one of embodiments 1 to 45, wherein the method further comprises identifying potential toxicities for each potential therapy.

[0457] 49. The method of any one of embodiments 1 to 46, wherein the method further comprises a step of identifying potential future side effects in response to each potential therapy.

[0458] 50. The method of any one of embodiments 1 to 49, wherein the therapeutic response to each therapeutic candidate is simulated in the system biological model by determining a known set of molecules affected by the therapeutic candidate, determining a therapeutic effect molecular level for each molecule in the known set of molecules based on one or more known mechanisms of action of the therapeutic candidate on the known set of molecules, and estimating a therapeutic effect molecular level for other molecules represented in the system biological model other than the known set of molecules based on a simulated effect of the determined therapeutic effect molecular level of the known set of molecules on one or more of the other molecules represented in the network.

[0459] 51. The method of embodiment 48, wherein the method comprises a step of comparing the determined therapeutic effect molecule levels and the estimated therapeutic effect molecule levels in the system biological model before and after therapeutic response simulation for each therapeutic candidate.

[0460] 52. The method of any one of embodiments 1 to 51, wherein the system biological model includes one or more pathways represented in Table 5 or Table 6.

[0461] 53. A method for screening therapeutic candidate agents for treating atherosclerotic cardiovascular disease, comprising: receiving non-invasively obtained data related to plaque from each of a plurality of subjects diagnosed with atherosclerotic cardiovascular disease; accessing a systems biological model of atherosclerotic cardiovascular disease, where (i) the systems biological model represents a plurality of pathways associated with atherosclerotic cardiovascular disease and (ii) the systems biological model includes disease-associated molecular levels for each molecule in the systems biological model; updating the systems biological model using the disease-associated molecular levels derived from the non-invasively obtained data from the subjects to generate a validated systems biological model; updating the validated systems biological model with information about the therapeutic candidate agents based on a known mechanism of action of the therapeutic candidate agents; simulating a therapeutic response to the therapeutic candidate agents in the updated and validated systems biological model to obtain a simulated therapeutic effect; comparing the therapeutic effect in the updated and validated systems biological model before and after simulating the therapeutic response with the therapeutic candidate agents; and determining whether the therapeutic candidate agents have a therapeutic effect based on the comparison.

[0462] 54. The method of embodiment 53, further comprising a step of quantifying the actual response at a cohort level.

[0463] 55. The method of any one of embodiments 53 or 54, wherein the selection method allows for the selection of cases that will increase the statistical power of the clinical trial.

[0464] 56. The method of any one of embodiments 53 or 54, wherein the selection method allows for the selection of cases that reduce the statistical power of the clinical trial.

[0465] 57. A method of selecting patient candidates to participate in a clinical trial testing the safety, or efficacy, or both, of a therapeutic candidate for patients with known or suspected atherosclerotic cardiovascular disease, comprising the steps of receiving non-invasively obtained data regarding plaque from the candidate subject; accessing a systems biology model of atherosclerotic cardiovascular disease; updating the systems biology model using personalized molecular levels derived from the non-invasively obtained data from the candidate subject to generate a subject-specific systems biology model; and updating the subject-specific systems biology model with information regarding the therapeutic candidate based on a known mechanism of action of the therapeutic candidate. simulating a therapeutic response by the candidate subject to the candidate therapeutic agent in the updated subject-specific systems biological model to obtain a simulated therapeutic effect for the candidate therapeutic agent; comparing the updated subject-specific systems biological model with the simulated therapeutic effect for each of the two or more combinations to the updated subject-specific systems biological model without the simulated therapeutic effect; and providing a report indicating whether the candidate subject's atherosclerotic cardiovascular disease is likely to be improved or not affected by the candidate therapeutic agent for the subject and / or whether the candidate subject will experience side effects from the candidate therapeutic agent.

[0466] 58. A computer-implemented method comprising: receiving a first input indicative of biological pathways associated with atherosclerotic cardiovascular disease; generating a first network based on the first input, the first network including nodes representing reference molecular levels and edges representing molecular interactions in one or more cell types; receiving a second input indicative of calibration data from a plurality of subjects diagnosed with the disease; determining disease-associated molecular levels for the molecules in the first network from the second input; and generating a second network based on the first network and the disease-associated molecular levels, wherein the second network calibrated using the second input represents an in silico systems biological model of the disease and includes disease-associated molecular levels for each molecule in the second network.

[0467] 59. The computer-implemented method of embodiment 58, wherein the step of receiving a plurality of first inputs comprises the step of querying a pathway database to identify biological pathways associated with atherosclerotic cardiovascular disease.

[0468] 60. The computer-implemented method of embodiment 58, wherein the one or more cell types comprise endothelial cells, vascular smooth muscle cells, macrophages, and lymphocytes.

[0469] 61. The computer-implemented method of embodiment 58, wherein the first network comprises (i) a core network representing molecular interactions specific to each respective cell type, (ii) a mid network representing molecular interactions across a subset of cell types, and (iii) a full network representing molecular interactions found in all cell types.

[0470] 62. The computer-implemented method of embodiment 58, wherein the edge representing an intermolecular interaction represents any one of translation, activation, inhibition, indirect effect, state change, binding, dissociation, phosphorylation, dephosphorylation, glycosylation, ubiquitination, and methylation.

[0471] 63. The computer-implemented method of embodiment 58, wherein the step of receiving the second input comprises, for each subject, obtaining at least plaque computed tomography angiography imaging data from the subject, plaque morphology data, and proteomic data corresponding to the subject.

[0472] 64. The computer-implemented method of embodiment 63, further comprising receiving transcriptomic data for at least a portion of the subjects.

[0473] 65. The computer-implemented method of embodiment 58, wherein the molecule is a protein, a gene, or a metabolite.

[0474] 66. The computer-implemented method of embodiment 65, wherein the first network includes nodes representing baseline levels of proteins and genes and edges representing protein-protein interactions, gene-gene interactions, and protein-gene interactions in one or more cell types.

[0475] 67. The computer-implemented method of embodiment 58, wherein the disease molecular levels are either measured molecular levels from the subject, or estimated molecular levels based on a virtual tissue model, or non-invasively obtained imaging data from the subject, or both.

[0476] 68. The computer-implemented method of embodiment 58, wherein the step of determining disease molecular levels for molecules in the first network comprises the steps of identifying disease molecular levels for a set of molecules from a second input, where the disease molecular levels of the set of molecules are provided by a second input from the subject, and estimating disease molecular levels for molecules in the first network other than the set of molecules based on the disease molecular levels of a subset of the set of molecules, where the subset of the set of molecules is represented by adjacent nodes in the first network.

[0477] 69. The computer-implemented method of embodiment 58, wherein the step of generating the second network comprises the steps of: indicating a disease molecular level for each node in the first network whose disease molecular level is obtained from calibration data from the subject; and indicating a disease molecular level for each node in the first network whose disease molecular level is estimated.

[0478] 70. A computer-implemented method of making a therapy recommendation for a patient with known or suspected atherosclerotic cardiovascular disease, comprising: receiving non-invasively obtained imaging data of atherosclerotic plaques from the patient; accessing a trained in silico system biological model of atherosclerotic cardiovascular disease, the trained in silico system biological model comprising a network with a disease molecular level for each of a plurality of nodes, each node representing a different molecule; calibrating the system biological model for the patient using the disease molecular levels derived from the imaging data; determining a known set of molecules affected by a therapy candidate; determining a therapy effect molecular level for each molecule in the known set of molecules based on one or more effects of the therapy candidate on the known set of molecules; estimating a therapy effect molecular level for other molecules represented in the in silico system biological model other than the known set of molecules based on a simulated effect of the determined therapy effect molecular level of the known set of molecules on one or more of the other molecules represented in the network; A computer-implemented method comprising: simulating a therapeutic response to each of a set of therapeutic candidates in a trained in silico system biological model by comparing a determined therapeutic effect molecular level with an estimated therapeutic effect molecular level in the silico system biological model; determining a preferred therapy based on the comparison; and providing a report indicating the preferred therapy for the patient.

[0479] 71. The computer-implemented method of embodiment 70, wherein the step of calibrating the network using disease molecule levels derived from the imaging data comprises the steps of: comparing the patient's computed tomography angiography imaging data with a plurality of computed tomography angiography imaging data of a plurality of subjects, where the plurality of computed tomography angiography imaging data of the plurality of subjects were inputs for training the system biological model; and predicting disease molecule levels for the molecules in the network based on the comparison.

[0480] 72. The computer-implemented method of embodiment 70, wherein the candidate therapy is a dyslipidemia management drug.

[0481] 73. The computer-implemented method of embodiment 72, wherein the dyslipidemia management medication is a high-dose statin.

[0482] 74. The computer-implemented method of embodiment 73, wherein the high-dose statin is atorvastatin.

[0483] 75. The computer-implemented method of embodiment 72, wherein the dyslipidemia management medication is an enhanced lipid-lowering medication.

[0484] 76. The computer-implemented method of embodiment 75, wherein the enhanced lipid-lowering agent is a proprotein convertase subtilisin kexin type 9 (PCSK9) inhibitor or a cholesteryl ester transfer protein (CETP).

[0485] 77. The computer-implemented method of embodiment 72, wherein the dyslipidemia management medication is a hypertriglyceridemia-lowering medication or a hypercholesterolemia-lowering medication.

[0486] 78. The computer-implemented method of embodiment 70, wherein the candidate therapy is a drug that affects the inflammatory cascade.

[0487] 79. The computer-implemented method of embodiment 78, wherein the drug that affects the inflammatory cascade is an anti-inflammatory drug.

[0488] 80. The computer-implemented method of embodiment 79, wherein the anti-inflammatory agent is an inhibitor of IL-1.

[0489] 81. The computer-implemented method of embodiment 80, wherein the inhibitor of IL-1 is canakinumab.

[0490] 82. The computer-implemented method of embodiment 79, wherein the anti-inflammatory drug inhibits the activity of TNF.

[0491] 83. The computer-implemented method of embodiment 79, wherein the anti-inflammatory drug inhibits IL12 / 23.

[0492] 84. The computer-implemented method of embodiment 79, wherein the anti-inflammatory drug inhibits IL17.

[0493] 85. The computer-implemented method of embodiment 78, wherein the drug that affects the inflammatory cascade is an inhibitor of inflammatory cytokines induced by danger signals.

[0494] 86. The computer-implemented method of embodiment 78, wherein the drug that affects the inflammatory cascade is a resolvin precursor.

[0495] 87. The computer-implemented method of embodiment 86, wherein the resolvin precursor is an omega-3 fatty acid.

[0496] 88. The computer-implemented method of embodiment 87, wherein the omega-3 fatty acid is eicosapentaenoic acid (EPA), docosahexaenoic acid (DHA), or docosapentaenoic acid (DPA).

[0497] 89. The computer-implemented method of embodiment 70, wherein the candidate therapy is an immunomodulatory agent.

[0498] 90. The computer-implemented method of embodiment 89, wherein the immunomodulatory agent triggers natural immunity.

[0499] 91. The computer-implemented method of embodiment 89, wherein the immunomodulatory agent is an immune tolerance stimulant.

[0500] 92. The computer-implemented method of embodiment 91, wherein the immune tolerance stimulant increases Treg activity.

[0501] 93. The computer-implemented method of embodiment 70, wherein the candidate therapy is an antihypertensive drug.

[0502] 94. The computer-implemented method of embodiment 93, wherein the antihypertensive drug is an ACE inhibitor.

[0503] 95. The computer-implemented method of embodiment 70, wherein the candidate therapy is an anticoagulant.

[0504] 96. The computer-implemented method of embodiment 95, wherein the anticoagulant reduces thrombin generation and / or limits the activity of thrombin.

[0505] 97. The computer-implemented method of embodiment 70, wherein the candidate therapy is a modulator of intracellular signaling.

[0506] 98. The computer-implemented method of embodiment 70, wherein the candidate therapy is an antidiabetic drug.

[0507] 99. The computer-implemented method of embodiment 98, wherein the antidiabetic drug is metformin.

[0508] 100. The computer-implemented method of embodiment 70, wherein the candidate therapy is a drug-eluting stent.

[0509] 101. The computer-implemented method of embodiment 100, wherein the drug-eluting stent is coated with a drug that inhibits cell cycle progression by inhibiting DNA synthesis.

[0510] 102. The computer-implemented method of embodiment 70, wherein the candidate therapy is a drug-coated balloon.

[0511] 103. The computer-implemented method of embodiment 102, wherein the drug-coated balloon is coated with a drug that inhibits neointimal growth by delivering an anti-proliferative substance to the vessel wall.

[0512] 104. The computer-implemented method of embodiment 70, wherein the candidate therapies are a combination of one or more of lipid-lowering drugs, anti-inflammatory drugs, and anti-diabetic drugs.

[0513] 105. The computer-implemented method of embodiment 70, wherein determining the therapeutic effect molecule levels comprises setting the therapeutic effect molecule levels of the set of molecules to a reference level.

[0514] 106. A system comprising a memory configured to store instructions and a processor for executing the instructions to perform operations, the operations comprising: receiving a first input indicative of biological pathways associated with atherosclerotic cardiovascular disease; generating a first network based on the first input, the first network including nodes representing reference molecular levels and edges representing molecular interactions in one or more cell types; receiving a second input indicative of calibration data from a plurality of subjects diagnosed with the disease; determining disease molecular levels for the molecules in the first network from the second input; and generating a second network based on the first network and the disease molecular levels, the second network calibrated using the second input representing an in silico system biological model of the disease and including disease molecular levels for each molecule in the second network.

[0515] 107. One or more computer readable media storing instructions executable by a processing device and, when executed, causing the processing device to perform operations comprising: receiving a first input indicative of biological pathways associated with atherosclerotic cardiovascular disease; generating a first network based on the first input, the first network including nodes representing reference molecular levels and edges representing molecular interactions in one or more cell types; receiving a second input indicative of calibration data from a plurality of subjects diagnosed with the disease; determining disease molecular levels for the molecules in the first network from the second input; and generating a second network based on the first network and the disease molecular levels, the second network calibrated using the second input representing an in silico systems biological model of the disease and including disease molecular levels for each molecule in the second network.

[0516] 108. A system comprising a memory configured to store instructions and a processor to execute the instructions to perform operations, the operations including receiving non-invasively obtained imaging data of atherosclerotic plaque from a patient; accessing a trained in silico system biological model of atherosclerotic cardiovascular disease, the trained in silico system biological model comprising a network with a disease molecular level for each of a plurality of nodes, each node representing a different molecule; calibrating the system biological model for the patient using the disease molecular levels derived from the imaging data; determining a known set of molecules affected by a therapy candidate; determining a therapy effect molecular level for each molecule in the known set of molecules based on one or more effects of the therapy candidate on the known set of molecules; estimating a therapy effect molecular level for other molecules represented in the in silico system biological model other than the known set of molecules based on a simulated effect of the determined therapy effect molecular level of the known set of molecules on one or more of the other molecules represented in the network; A system comprising: simulating a therapeutic response to each of a set of therapeutic candidate therapies in a trained in silico system biological model by comparing a determined therapeutic effect molecular level with an estimated therapeutic effect molecular level in the silico system biological model; determining a preferred therapy based on the comparison; and providing a report indicating the preferred therapy for the patient.

[0517] 109. One or more computer readable media storing instructions executable by a processing device and, when executed, causing the processing device to perform operations, the operations including receiving non-invasively obtained imaging data of atherosclerotic plaque from a patient; accessing a trained in silico system biological model of atherosclerotic cardiovascular disease, the trained in silico system biological model comprising a network with a disease molecular level for each of a plurality of nodes, each node representing a different molecule; calibrating the system biological model for the patient using the disease molecular levels derived from the imaging data; determining a known set of molecules affected by a therapy candidate; determining a therapy effect molecular level for each molecule in the known set of molecules based on one or more effects of the therapy candidate on the known set of molecules; estimating a therapy effect molecular level for other molecules represented in the in silico system biological model other than the known set of molecules based on a simulated effect of the determined therapy effect molecular level of the known set of molecules on one or more of the other molecules represented in the network; A computer-readable medium comprising: simulating a therapeutic response to each of a set of therapeutic candidate therapies in a trained in silico system biological model by comparing a determined therapeutic effect molecular level with an estimated therapeutic effect molecular level in the silico system biological model; determining a preferred therapy based on the comparison; and providing a report indicating the preferred therapy for the patient.

[0518] 110. A method for making a recommendation of a combination of any two or more therapies selected from lipid-lowering therapy, anti-inflammatory therapy, and anti-diabetic therapy for a patient with known or suspected atherosclerotic cardiovascular disease, comprising the steps of receiving non-invasively obtained data on plaque from the patient and accessing a systems biological model of atherosclerotic cardiovascular disease, (i) the systems biological model represents a plurality of pathways associated with atherosclerotic cardiovascular disease, (ii) the plurality of pathways includes pathways corresponding to two or all three of a) one or more of glycosylated (glyLDL), oxidized (oxLDL), and minimally modified (mmLDL), or VLDL, b) one or more of IL-1, IL1β, TNF, IL12 / 23, IL17, or other cytokine molecules, and c) one or more of MTOR, NFκβ1, ICAM1, or VCAM1, respectively, and (iii) the systems biological model includes disease-related molecule levels for each molecule in the systems biological model. updating the system biological model using personalized molecular levels derived from non-invasively obtained data from the patient to generate a patient-specific system biological model; updating the patient-specific system biological model with information regarding the effect of lipid-lowering drugs on LDL levels, the effect of anti-inflammatory drugs on inflammation levels, and / or the effect of anti-diabetic drugs on glucose levels based on a known mechanism of action of each of the drugs; simulating the patient's therapeutic response to any two or more combinations of lipid-lowering drugs, anti-inflammatory drugs, and anti-diabetic drugs in the updated patient-specific system biological model to obtain simulated therapeutic effects for the two or more combinations; comparing the updated patient-specific system biological model with the simulated therapeutic effects for each of the two or more combinations with the updated patient-specific system biological model without the same; and providing a report recommending to the patient a therapeutic drug combination that provides the greatest level of improvement based on the comparison.

[0519] 111. 1. A method of identifying one or more contraindications associated with any two or more combinations of lipid-lowering, anti-inflammatory, and anti-diabetic therapies in a patient with known or suspected atherosclerotic cardiovascular disease, comprising the steps of receiving non-invasively obtained data relating to plaque from the patient; and accessing a systems biology model of atherosclerotic cardiovascular disease, wherein (i) the systems biology model represents a plurality of pathways associated with atherosclerotic cardiovascular disease, (ii) the plurality of pathways includes one or more pathways corresponding to two or all three of a) one or more of glycosylated (glyLDL), oxidized (oxLDL), and minimally modified (mmLDL), or VLDL, b) one or more of IL-1, IL1β, TNF, IL12 / 23, or IL17, and c) one or more of MTOR, NFκβ1, ICAM1, or VCAM1, respectively; and (iii) the systems biology model includes disease-related molecule levels for each molecule in the systems biology model; and updating the system biological model using personalized molecular levels derived from non-invasively obtained data from the patient to generate a system biological model of the lipid-lowering drug; updating the patient-specific system biological model with information regarding the effect of lipid-lowering drugs on LDL levels, the effect of anti-inflammatory drugs on inflammation levels, and / or the effect of anti-diabetic drugs on glucose levels based on a known mechanism of action of each of the drugs; simulating a therapeutic response by the patient to any two or more combinations of lipid-lowering drugs, anti-inflammatory drugs, and anti-diabetic drugs in the updated patient-specific system biological model to obtain a simulated therapeutic effect for the two or more combinations; comparing the updated patient-specific system biological model with the simulated therapeutic effect for each of the two or more combinations to the updated patient-specific system biological model without the same; and identifying one or more contraindications associated with any two or more combinations of lipid-lowering drugs, anti-inflammatory drugs, and anti-diabetic drugs based on the comparison;and providing a report to the patient indicating one or more contraindications associated with any two or more combinations of antidiabetic drugs.

[0520] 112. The method of embodiment 110 or 111, wherein the lipid-lowering drug is a statin or an enhanced lipid-lowering drug.

[0521] 113. The method of embodiment 110 or 111, wherein the anti-inflammatory agent is an inhibitor of IL-1, IL1β, TNF, IL12 / 23, IL17, or other cytokine proteins.

[0522] 114. The method of embodiment 110 or 111, wherein the antidiabetic drug is metformin.

[0523] 115. The method of embodiment 110 or 111, wherein the step of simulating the therapeutic response to any two or more combinations of lipid-lowering drugs, anti-inflammatory drugs, and antidiabetic drugs in the patient-specific system biological model comprises the steps of determining a set of molecules known to be affected by one or more lipid-lowering drugs, anti-inflammatory drugs, and / or antidiabetic drugs, determining a therapeutic effect molecular level for each molecule in the set of molecules based on one or more known mechanisms of action of any one or more of the lipid-lowering drugs, anti-inflammatory drugs, and antidiabetic drugs on the set of molecules, and estimating therapeutic effect molecular levels for molecules represented in the patient-specific system biological model other than those in the set of molecules based on the simulated effect of the determined therapeutic effect molecular level of the set of molecules on one or more of the other molecules represented in the network.

[0524] 116. The method of embodiment 110 or 111, wherein at least one network comprises one or more pathways represented in Table 5 or Table 6 that are affected by any one or more of LDL levels, inflammation levels, and / or glucose levels.

[0525] 117. A method of making lipid-lowering therapy recommendations for a patient with known or suspected atherosclerotic cardiovascular disease, comprising the steps of receiving non-invasively obtained data relating to plaque from the patient; accessing a systems biology model of atherosclerotic cardiovascular disease, where (i) the systems biology model represents a plurality of pathways associated with atherosclerotic cardiovascular disease, (ii) the plurality of pathways correspond to one or more of glycosylated low-density lipoprotein (glyLDL), oxidized LDL (oxLDL), minimally modified LDL (mmLDL), or very low-density lipoprotein (VLDL), and (iii) the systems biology model includes disease-associated molecule levels for each molecule in the systems biology model; and generating a patient-specific systems biology model. updating the patient-specific system biological model with personalized molecular levels derived from non-invasively obtained data from the patient to achieve a therapeutic effect; updating the patient-specific system biological model with information regarding the effect of the lipid-lowering drug on LDL levels based on a known mechanism of action of the lipid-lowering drug; simulating a therapeutic response by the patient to the lipid-lowering drug in the updated patient-specific system biological model to obtain a simulated therapeutic effect; comparing the updated patient-specific system biological model with the simulated therapeutic effect to the updated patient-specific system biological model without the simulated therapeutic effect; and when the comparison indicates an improvement for the patient, providing a report recommending a lipid-lowering drug to the patient.

[0526] For therapies involving anti-inflammatory therapies, anti-diabetic therapies, and combination therapies, similar methods can be implemented by modifying pathways in a biological systems model to include pathways associated with the target of a particular type of therapy, as disclosed herein.

[0527] 118. A method for identifying one or more contraindications associated with lipid-lowering therapy in a patient with known or suspected atherosclerotic cardiovascular disease, comprising the steps of receiving non-invasively obtained data relating to plaque from the patient; accessing a systems biology model of atherosclerotic cardiovascular disease, where (i) the systems biology model represents a plurality of pathways associated with atherosclerotic cardiovascular disease, (ii) the plurality of pathways correspond to one or more of glycosylated low-density lipoprotein (glyLDL), oxidized LDL (oxLDL), minimally modified LDL (mmLDL), or very low-density lipoprotein (VLDL), and (iii) the systems biology model includes disease-associated molecular levels for each molecule in the systems biology model; and generating a patient-specific systems biology model. updating the system biological model using personalized molecular levels derived from non-invasively obtained data from the patient; updating the patient-specific system biological model with information about the effect of the lipid-lowering drug on LDL levels based on a known mechanism of action of the lipid-lowering drug; simulating a therapeutic response by the patient to the lipid-lowering drug in the updated patient-specific system biological model to obtain a simulated therapeutic effect; comparing the updated patient-specific system biological model with the simulated therapeutic effect and the updated patient-specific system biological model without the simulated therapeutic effect; identifying one or more contraindications associated with the lipid-lowering drug based on the comparison; and providing a report to the patient indicating the contraindications associated with the lipid-lowering drug.

[0528] 119. A method for screening a candidate dyslipidemic drug for atherosclerotic cardiovascular disease, comprising the steps of receiving non-invasively derived data relating to plaque from each of a plurality of subjects diagnosed with atherosclerotic cardiovascular disease; accessing a systems biology model of atherosclerotic cardiovascular disease, wherein (i) the systems biology model represents a plurality of pathways associated with atherosclerotic cardiovascular disease, (ii) the plurality of pathways includes one or more pathways corresponding to potential targets of the candidate dyslipidemic drug; and (iii) the systems biology model includes disease-associated molecular levels for each molecule in the systems biology model; and computing the disease-associated molecular levels derived from the non-invasively derived data from the subjects to generate a validated systems biology model. updating the validated system biology model using a system biology model based on a known mechanism of action of the candidate dyslipidemia therapeutic agent; updating the validated system biology model with information about an effect of the candidate dyslipidemia therapeutic agent on low density lipoprotein (LDL) based on a known mechanism of action of the candidate dyslipidemia therapeutic agent; simulating a therapeutic response to the candidate dyslipidemia therapeutic agent in the updated and validated system biology model to obtain a simulated therapeutic effect; comparing the therapeutic effect in the updated and validated system biology model before and after simulating the therapeutic response with the candidate dyslipidemia therapeutic agent; and providing a report indicating that the candidate dyslipidemia therapeutic agent is a potential therapeutic agent when the comparison indicates that the candidate dyslipidemia therapeutic agent results in an improvement of the disease state.

[0529] 120. The method of any one of embodiments 117, 118, or 119, wherein the lipid-lowering drug is a statin.

[0530] 121. The method of any one of embodiments 117, 118, or 119, wherein the lipid-lowering agent is an enhanced lipid-lowering agent.

[0531] 122. The method of any one of embodiments 117, 118, or 119, wherein the enhanced lipid-lowering agent is a proprotein convertase subtilisin kexin type 9 (PCSK9) inhibitor or a cholesteryl ester transfer protein (CETP) inhibitor.

[0532] 123. The method of any one of embodiments 117, 118, or 119, further comprising the step of recommending a combination of a lipid-lowering drug and one or both of an anti-inflammatory drug and an anti-diabetic drug.

[0533] 124. The method of any one of embodiments 117, 118, or 119, wherein the step of simulating the therapeutic response to a lipid-lowering drug in the patient-specific systems biological model comprises the steps of determining a set of molecules known to be affected by the lipid-lowering drug, determining a therapeutic effect molecular level for each molecule in the set of molecules based on one or more known mechanisms of action of the lipid-lowering drug on the set of molecules, and estimating therapeutic effect molecular levels for molecules represented in the patient-specific systems biological model other than those in the set of molecules based on the simulated effect of the determined therapeutic effect molecular level of the set of molecules on one or more of the other molecules represented in the network.

[0534] 125. The method of any one of embodiments 117, 118, or 119, wherein the system biological model includes one or more pathways represented in Table 5 or Table 6 that are affected by LDL.

[0535] 126. A method of making anti-inflammatory therapy recommendations for a patient with known or suspected atherosclerotic cardiovascular disease, comprising the steps of receiving non-invasively obtained data relating to plaque from the patient; accessing a systems biology model of atherosclerotic cardiovascular disease, where (i) the systems biology model represents a plurality of pathways associated with atherosclerotic cardiovascular disease, (ii) the plurality of pathways correspond to one or more of IL-1, IL1β, TNF, IL12 / 23, IL17, or other cytokine molecules, and (iii) the systems biology model includes disease-related molecular levels for each molecule in the systems biology model; and receiving non-invasively obtained data relating to plaque from the patient to generate a patient-specific systems biology model. updating the patient-specific system biological model with personalized molecular levels derived from data acquired in a targeted manner; updating the patient-specific system biological model with information regarding the effect of the anti-inflammatory drug on inflammation based on a known mechanism of action of the anti-inflammatory drug; simulating a therapeutic response by the patient to the anti-inflammatory drug in the updated patient-specific system biological model to obtain a simulated therapeutic effect; comparing the updated patient-specific system biological model with the simulated therapeutic effect to the updated patient-specific system biological model without the same; and providing a report recommending an anti-inflammatory drug to the patient when the comparison indicates an improvement for the patient.

[0536] 127. A method for identifying one or more contraindications associated with anti-inflammatory therapy in a patient with known or suspected atherosclerotic cardiovascular disease, comprising the steps of receiving non-invasively obtained data relating to plaque from the patient; accessing a systems biology model of atherosclerotic cardiovascular disease, where (i) the systems biology model represents a plurality of pathways associated with atherosclerotic cardiovascular disease, (ii) the plurality of pathways correspond to one or more of IL-1, IL1β, TNF, IL12 / 23, IL17, or other cytokine molecules, and (iii) at least one network includes disease-associated molecule levels for each molecule in the systems biology model; and accessing the non-invasively obtained data relating to plaque from the patient to generate a patient-specific systems biology model. updating the patient-specific system biological model with personalized molecular levels derived from the data collected; updating the patient-specific system biological model with information regarding the effect of the anti-inflammatory drug on inflammation based on a known mechanism of action of the anti-inflammatory drug; simulating a therapeutic response by the patient to the anti-inflammatory drug in the updated patient-specific system biological model to obtain a simulated therapeutic effect; comparing the updated patient-specific system biological model with the simulated therapeutic effect and the updated patient-specific system biological model without the simulated therapeutic effect; identifying any one or more contraindications associated with the anti-inflammatory drug based on the comparison; and providing a report to the patient indicating the contraindications associated with the anti-inflammatory drug.

[0537] 128. A method for screening anti-inflammatory drug candidates for atherosclerotic cardiovascular disease, comprising the steps of receiving non-invasively obtained data on plaque from each of a plurality of subjects having known or suspected atherosclerotic cardiovascular disease; accessing a systems biology model of atherosclerotic cardiovascular disease, where (i) the systems biology model represents a plurality of pathways associated with atherosclerotic cardiovascular disease, (ii) the plurality of pathways includes one or more pathways corresponding to potential targets of the anti-inflammatory drug candidate, and (iii) the systems biology model includes disease-related molecule levels for each molecule in the systems biology model; and accessing the non-invasively obtained data from the subjects to generate a validated systems biology model. updating the system biological model using disease-related molecular levels derived from the data; updating the validated system biological model with information regarding the effect of the anti-inflammatory drug candidate on inflammation based on a known mechanism of action of the anti-inflammatory drug candidate; simulating a therapeutic response to the anti-inflammatory drug candidate in the updated and validated system biological model to obtain a simulated therapeutic effect; comparing the therapeutic effect in the updated and validated system biological model before and after simulating the therapeutic response to the anti-inflammatory drug candidate; and providing a report indicating that the anti-inflammatory drug candidate is a potential therapeutic agent when the comparison indicates that the anti-inflammatory drug candidate results in improvement of the disease condition.

[0538] 129. The method of any one of embodiments 126, 127, or 128, wherein the anti-inflammatory agent is colchicine or an inhibitor of IL-1.

[0539] 130. The method of embodiment 129, wherein the inhibitor of IL-1 is canakinumab.

[0540] 131. The method of any one o...

Claims

1. A method implemented on a computer for making a therapy recommendation to a patient with known or suspected atherosclerotic cardiovascular disease, comprising: receiving imaging data non-invasively obtained from plaques from said patient; accessing a systems biological model of atherosclerotic cardiovascular disease, comprising: (i) said systems biological model representing a plurality of pathways associated with atherosclerotic cardiovascular disease; (ii) said systems biological model including disease-related molecular levels for each molecule represented by at least one network in said systems biological model; updating said systems biological model using personalized molecular levels derived from said imaging data non-invasively obtained from said patient to generate a patient-specific systems biological model; obtaining information regarding one or more therapy candidates for said patient; updating said patient-specific systems biological model using information regarding the intended effect of each therapy candidate; simulating the therapy response for each therapy candidate in said systems biological model to obtain a simulated therapy effect for each therapy candidate; comparing the simulated therapy effects in said systems biological model before and after said therapy response simulation for each therapy candidate; selecting, based on said comparison, one of said one or more therapy candidates as a preferred therapy; and providing a report recommending said preferred therapy for said patient. A method implemented on a computer.

2. The step of simulating the therapy response comprises, in the at least one network, setting a reduced molecular level related to plaque instability and setting an increased molecular level related to plaque stability, the computer-implemented method according to claim 1.

3. The molecule is a gene, protein, or metabolite, and the step of updating the system biological model using the personalized molecular level comprises using a disease gene transcription level, a disease protein level, or a combination of both, derived from the non-invasively obtained imaging data, the computer-implemented method according to claim 1.

4. The imaging data is radiological imaging data obtained by diagnostic images of computed tomography (CT), dual-energy computed tomography (DECT), spectral computed tomography (spectral CT), computed tomography angiography (CTA), cardiovascular computed tomography angiography (CCTA), magnetic resonance imaging (MRI), multi-contrast magnetic resonance imaging (multi-contrast MRI), ultrasound (US), positron emission tomography (PET), intravascular ultrasound (IVUS), optical coherence tomography (OCT), near-infrared spectroscopy (NIRS), or single-photon emission computed tomography (SPECT), the computer-implemented method according to claim 1.

5. The method further comprises the step of processing the non-invasively obtained imaging data to obtain quantitative plaque morphology data including anatomical structure data, tissue composition data, or both, i) the anatomical structure data comprises data related to any one or more levels of remodeling, wall thickening, ulceration, stenosis, dilation, or plaque burden, or ii) the tissue composition data comprises data regarding one or more levels of any one of calcification, necrotic lipid core (LRNC), intraplaque hemorrhage (IPH), matrix, fibrous capsule, or perivascular adipose tissue (PVAT), or The computer-implemented method according to claim 1, which is both i) and ii).

6. The computer-implemented method according to claim 1, wherein the pathway is compartmentalized into cell-specific networks, and the cell-specific networks at least include an endothelial cell network, a macrophage network, and a vascular smooth muscle cell network.

7. The computer-implemented method according to any one of claims 1 to 6, wherein the therapy candidate is a lipid disorder management drug.

8. The therapy candidate is a drug that affects the inflammatory cascade, or The computer-implemented method according to any one of claims 1 to 7, wherein the drug that affects the inflammatory cascade is an inhibitor of inflammatory cytokines induced by danger signals or a resolvin precursor.

9. The therapy candidate is an immunomodulatory drug, and the immunomodulatory drug triggers innate immunity, or The computer-implemented method according to any one of claims 1 to 7, wherein the immunomodulatory drug is an immune tolerance-stimulating drug.

10. The therapy candidate is an antihypertensive drug, an anticoagulant, a drug for regulating intracellular signal transduction, an antidiabetic drug, a drug-eluting stent, or The computer-implemented method according to any one of claims 1 to 7, wherein the therapy candidate is a drug-coated balloon.

11. The method further comprises the step of quantifying the actual response of the patient to each therapy candidate, and / or the method further comprises the step of detecting one or more contraindication candidates associated with each therapy candidate, and / or the method further comprises the step of identifying likely side effects for each therapy candidate, and / or The method further comprises the step of identifying potential toxicity for each therapy candidate, and / or The computer-implemented method according to any one of claims 1 to 10, wherein the method further comprises the step of identifying future side effect candidates in the response to each therapy candidate. **Claim 12** The therapy response for each therapy candidate is determining a known set of molecules affected by the therapy candidate, defining a therapeutic effect molecular level for each molecule in the known set of molecules based on one or more known mechanisms of action of the therapy candidate on the known set of molecules, estimating the therapeutic effect molecular level for other molecules represented in the system biological model other than the known set of molecules based on the simulated effect of the defined therapeutic effect molecular levels of the known set of molecules on one or more of the other molecules represented in the network Thereby, the computer-implemented method according to any one of claims 1 to 11, which is simulated in the system biological model. **Claim 13** A computer-implemented method for selecting a therapeutic agent candidate for treating atherosclerotic cardiovascular disease, comprising: receiving non-invasively obtained imaging data regarding plaques from each of a plurality of subjects diagnosed with atherosclerotic cardiovascular disease; accessing a system biological model of atherosclerotic cardiovascular disease, wherein (i) the system biological model represents a plurality of pathways associated with atherosclerotic cardiovascular disease, (ii) the system biological model includes disease-related molecular levels for each molecule in the system biological model, steps, updating the system biological model using disease-related molecular levels derived from the non-invasively obtained data from the subject to generate a validated system biological model; updating the validated system biological model using information on the therapeutic agent candidate based on the known mechanism of action of the therapeutic agent candidate; simulating the therapeutic response to the therapeutic agent candidate in the updated and validated system biological model to obtain a simulated therapeutic effect; comparing the therapeutic effects in the updated and validated system biological model before and after simulating the therapeutic response to the therapeutic agent candidate; determining whether the therapeutic agent candidate has a therapeutic effect based on the comparison, a method implemented by a computer.

14. a memory configured to store instructions, a system comprising a processor for executing the instructions to implement the method according to any one of claims 1 to 12.

15. One or more computer-readable media, executable by a processing device and storing instructions which, when so executed, cause the processing device to implement the method according to any one of claims 1 to 12.

Images