Method and device for predicting and diagnosing risk levels of cardiovascular diseases by type

Abstract

A device for predicting and diagnosing risk levels of cardiovascular diseases by type, according to an embodiment, comprises at least one processor, wherein the at least one processor predicts the risk levels of cardiovascular diseases by type using a diagnostic model that includes: a first model that analyzes input fundus images to predict cardiovascular risk levels; and a second model that receives the cardiovascular risk score predicted by the first model and type-specific factors for the diseases to predicts risk levels of cardiovascular diseases by type. The at least one processor provides diagnostic information on the basis of the cardiovascular risk levels predicted by the first model and the risk levels of cardiovascular diseases by type predicted by the second model.

Description

Diagnostic method and device for predicting risk by type of cardiovascular disease

[0001] This application relates to a diagnostic method and device for predicting the risk of each type of cardiovascular disease using machine learning that has learned fundus photographs and cardiovascular biomarkers.

[0002]

[0003] Fundus photography (i.e., fundus images or retinal images) is an image of the back surface of the retina, including the retina, retinal blood vessels, optic nerve, macula, and choroid. While fundus photography has traditionally been primarily used to diagnose ophthalmic diseases, recent rapid advances in artificial intelligence (AI) have significantly expanded its application to assess the risk of systemic diseases such as cardiovascular disease. This technology is not only noninvasive and cost-effective, but is also recognized as a valuable tool in disease screening, particularly for large populations. AI excels at analyzing retinal vascular changes associated with cardiovascular disease, leading to a new paradigm in medical image analysis. The applicant has filed Korean patent applications, including Application Nos. 10-2018-0166720, 10-2018-0166721, and 10-2018-0166722, regarding a method for assisting the diagnosis of cardiovascular disease using existing fundus photography. The patent application relates to a technology that provides supplementary information related to an integrated risk assessment of cardiovascular disease, focusing on more accurately predicting the risk of cardiovascular disease through fundus photography and supporting clinical decision-making based on this.

[0004] Meanwhile, cardiovascular disease includes specific types such as myocardial infarction (MI), stroke, peripheral artery disease (PAD), heart failure (HF), and atrial fibrillation (AF). While the applicant's existing patent application, mentioned above, aims to predict the overall risk of cardiovascular disease, if the individual risk for each specific type of cardiovascular disease could be predicted, this could provide more detailed supplementary information. This could provide crucial support for medical professionals in making more precise clinical decisions tailored to each patient's condition.

[0005]

[0006] The problem to be solved by the present invention is to provide a method for predicting the risk of cardiovascular disease by type by receiving a fundus photograph of a target patient as input and predicting the risk of cardiovascular disease by type by using a machine learning model that has learned the correlation between the fundus photograph and the risk of cardiovascular disease by type, and a device implementing the method. Specifically, the present invention aims to provide a device for predicting the risk of cardiovascular disease by type by additionally considering disease type-specific factors, including age, gender, blood biomarkers, smoking status, and obesity, in an integrated cardiovascular risk score based on features extracted from the fundus photograph by the machine learning model.

[0007] The problem to be solved by the present invention is to provide a device for predicting the risk of cardiovascular disease by type by additionally considering the type of cardiovascular plaque of the target patient, i.e., whether it is soft plaque or hard plaque.

[0008] The present invention aims to provide a device for predicting risk by type of cardiovascular disease, which includes a model for determining which clinical test is recommended for a patient based on the risk by type of cardiovascular disease.

[0009] The problems to be solved by this application are not limited to the problems described above, and problems not mentioned can be clearly understood by a person having ordinary skill in the technical field to which this application belongs from this specification and the attached drawings.

[0010]

[0011] According to one aspect of the present application, a device for predicting and diagnosing risk by type of cardiovascular disease according to one embodiment includes a first model (cardiovascular risk prediction model) that analyzes an input fundus photograph to predict cardiovascular risk, and a second model (cardiovascular disease type-specific risk prediction model) that additionally receives a cardiovascular risk score predicted by the first model and specific factors by type of disease to predict the risk by type of cardiovascular disease, and may include a device for predicting and diagnosing risk by type of cardiovascular disease, characterized in that the device provides a risk grade and related information by type of cardiovascular disease by synthesizing the cardiovascular risk predicted by the first model and the cardiovascular disease type-specific risk predicted by the second model.

[0012] According to one aspect of the present application, a device for predicting and diagnosing risk by type of cardiovascular disease according to one embodiment includes a first model for predicting cardiovascular risk by analyzing an input fundus photograph, a second model for predicting risk by type of cardiovascular disease by receiving a cardiovascular risk score predicted by the first model and specific factors by type of disease, and a third model (recommended test judgment model) for determining a clinical test recommended to a patient based on the risk by type of cardiovascular disease predicted by the second model, and the device may include a device for predicting and diagnosing risk by type of cardiovascular disease, characterized in that the device provides a risk grade and related information for each type of cardiovascular disease and information related to recommended tests by synthesizing the cardiovascular risk predicted by the first model and the risk by type of cardiovascular disease predicted by the second model.

[0013] According to one aspect of the present application, a device for predicting and diagnosing risk by type of cardiovascular disease according to one embodiment includes a first model for predicting cardiovascular risk and cardiovascular plaque type by analyzing an input fundus photograph, a second model for predicting risk by type of cardiovascular disease by additionally receiving a cardiovascular risk score predicted by the first model and specific factors by type of disease, and a third model for determining a clinical test recommended to a patient based on the risk by type of cardiovascular disease predicted by the second model, and the device may include a device for predicting and diagnosing risk by type of cardiovascular disease, characterized in that the device provides a risk grade and related information for each type of cardiovascular disease and information related to recommended tests by synthesizing the cardiovascular risk and plaque type predicted by the first model and the risk by type of cardiovascular disease predicted by the second model.

[0014]

[0015] According to the present invention, a machine learning model that has learned the correlation between fundus photographs and cardiovascular disease risk by type can be used to predict not only the patient's integrated cardiovascular risk score but also the risk by type of cardiovascular disease based on the fundus photographs of the target patient. Furthermore, the machine learning model can predict the patient's cardiovascular plaque type. This allows for the development of a personalized treatment plan optimized for each patient and the early identification of patients at high risk for specific cardiovascular diseases, thereby strengthening preventive measures.

[0016] The effects of the present application are not limited to the effects described above, and effects not mentioned can be clearly understood by a person having ordinary skill in the art to which the present application pertains from this specification and the attached drawings.

[0017]

[0018] FIG. 1 is a block diagram illustrating a diagnostic device according to one embodiment.

[0019] Figure 2 is a diagram illustrating that cardiovascular risk (Reti-CVD) can be a biomarker for predicting individual risks of cardiovascular stroke, myocardial infarction, and atrial fibrillation.

[0020] Figures 3 and 4 are diagrams illustrating the correlation between Reti-CVD and various types of cardiovascular diseases.

[0021] FIG. 5 and FIG. 6 are diagrams for explaining the performance of predicting cardiovascular disease for diabetic patients according to the present embodiment.

[0022] Figure 7 is a diagram for explaining a method for predicting risk by type of cardiovascular disease according to one embodiment.

[0023] Figure 8 is a diagram for explaining the prediction performance of atrial fibrillation risk according to one embodiment.

[0024] Figure 9 is a diagram for explaining a method for predicting risk by type of cardiovascular disease according to one embodiment.

[0025] Fig. 10 is a diagram for explaining a method for predicting risk by type of cardiovascular disease according to another embodiment.

[0026]

[0027] The above-described purposes, features, and advantages of the present application will become more apparent through the following detailed description, taken in conjunction with the accompanying drawings. However, as the present application is susceptible to various modifications and various embodiments, specific embodiments will be illustrated in the drawings and described in detail below.

[0028] Throughout the specification, identical reference numbers, in principle, indicate identical components. Furthermore, components with identical functions within the scope of the same concept shown in the drawings of each embodiment are described using the same reference numbers, and redundant descriptions thereof will be omitted.

[0029] If a detailed description of a known function or configuration related to this application is deemed to unnecessarily obscure the gist of this application, such detailed description will be omitted. Furthermore, numbers (e.g., "first," "second," etc.) used throughout the description of this specification are merely identifiers used to distinguish one component from another.

[0030] In addition, the suffixes "module" and "part" for components used in the following examples are given or used interchangeably only for the convenience of writing the specification, and do not have distinct meanings or roles in themselves.

[0031] In the examples below, singular expressions include plural expressions unless the context clearly indicates otherwise.

[0032] In the examples below, terms such as “include” or “have” mean that a feature or component described in the specification is present, and do not preclude the possibility that one or more other features or components may be added.

[0033] For convenience of explanation, the sizes of components in the drawings may be exaggerated or reduced. For example, the sizes and thicknesses of each component shown in the drawings are arbitrarily shown for convenience of explanation, and the present invention is not necessarily limited to what is shown.

[0034] In some embodiments, where implementations are otherwise feasible, the order of specific processes may differ from the order described. For example, two processes described in succession may be performed substantially simultaneously, or in a reverse order from the order described.

[0035] In the following examples, when components, etc. are said to be connected, this includes not only cases where the components are directly connected, but also cases where components are indirectly connected by intervening components. For example, when it is said in this specification that components, etc. are electrically connected, this includes not only cases where the components, etc. are directly electrically connected, but also cases where components, etc. are indirectly electrically connected by intervening components, etc.

[0036] Below, a diagnostic device and method for predicting the risk of cardiovascular disease by type based on ocular images and assisting medical professionals in making decisions are described. In this specification, the term "diagnosis" may refer to a diagnostic aid intended to assist in the diagnosis of a disease, rather than directly diagnosing the disease. While the term "diagnosis" is used hereinafter for convenience of explanation, it can also refer to a diagnostic aid. The machine learning model described herein can be designed based on various machine learning libraries.

[0037]

[0038] Cardiovascular diseases include hypertension, atherosclerosis (or atherosclerosis), coronary artery disease (CAD), valvular disease, pericardial disease, myocarditis, cardiomyopathy, arrhythmia, heart failure (HF), rheumatic heart disease, congenital heart disease, aortic aneurysm, peripheral vascular disease, cerebrovascular disease, and pulmonary vascular disease, and are specifically classified as angina pectoris, myocardial infarction (MI), pulmonary artery hypertension (PAH), Cor Pulmonale, Left Heart Failure, Right Heart Failure, Aortic Valve Stenosis (AS), Aortic Valve Regurgitation (AR), Pulmonary Valve Stenosis (PS), Pulmonary Valve Regurgitation (PR), Tricuspid Valve Stenosis (TS), Tricuspid Valve Regurgitation (TR), Mitral Valve Stenosis (MS), Mitral Valve Regurgitation, Peripheral Artery Disease (PAD), Aortic Aneurysm (e.g.,It may include at least one of abdominal aortic aneurysm, thoracic aortic aneurysm), pericarditis, stroke, cerebral infarction, cerebral hemorrhage, cerebral aneurysm, thrombotic disease (e.g., deep vein thrombosis, pulmonary embolism), atrial fibrillation (AF), atrial tachycardia (AT), paroxysmal supraventricular tachycardia (PSVT), ventricular tachycardia (VT), bradycardia, ischemic stroke, hemorrhagic stroke, and transient ischemic attack (TIA). Cardiovascular disease may include complications. Additionally, complications may include heart attack, death due to cardiovascular disease, heart attack, cardiogenic shock, kidney disease, aspiration pneumonia, dysphagia, motor dysfunction, speech dysfunction, cognitive dysfunction, sleep disorders, emotional disorders, neuropathic pain, urinary tract infections, malnutrition, deep vein thrombosis, bedsores, falls, pain, seizures, depression, etc. Additionally, detailed subtypes by type of cardiovascular disease may be included. For example, heart failure is divided into heart failure with reduced ejection fraction (HFrEF), heart failure with borderline ejection fraction (HFmrEF), and heart failure with preserved ejection fraction (HFpEF). In the case of stroke, it can be divided into ischemic stroke, thrombotic cerebral infarction (cerebral thrombosis), cerebral embolism (embolic cerebral infarction), lacunar cerebral infarction, hemorrhagic stroke, intracerebral hemorrhage, subarachnoid hemorrhage, transient ischemic attack, brainstem stroke, and stroke of unknown cause. In the case of atrial fibrillation, it includes subtypes of paroxysmal atrial fibrillation, persistent atrial fibrillation, long-term persistent atrial fibrillation, and permanent atrial fibrillation, and in the case of myocardial infarction, it includes type 1 (spontaneous myocardial infarction),It is divided into subtypes: type 2 (myocardial infarction due to ischemic imbalance), type 3 (myocardial infarction resulting in death when biomarker values ​​are not available), type 4a (myocardial infarction related to percutaneous coronary intervention (PCI)), type 4b (myocardial infarction related to stent thrombosis), and type 5 (myocardial infarction related to coronary artery bypass grafting (CABG)).

[0039]

[0040] 1. Biomarkers for cardiovascular disease subtypes

[0041] Fundus photography provides a noninvasive way to observe the vascular structure of the human body, which provides the basis for predicting the risk of cardiovascular disease (CVD). Machine learning models trained on fundus photographs and cardiovascular biomarkers are used to predict the risk of future cardiovascular disease. Major cardiovascular disease biomarkers include the Coronary Artery Calcium (CAC) score, the Pooled Cohort Equation (PCE) score, the QRISK score, the modified Framingham Score (FRS), the Carotid Intima-Media Thickness (CIMT) score, the Brachial-Ankle Pulse Wave Velocity (baPWV) score, and the Ankle-Bracial Index (ABI). The machine learning model generates a cardiovascular risk score, Reti-CVD, for the input fundus photograph. Based on this score, patients are classified into three risk groups: low, intermediate, and high. This classification criterion used the reference value established in the applicant's previous study, “Deep-learning-based cardiovascular risk stratification using coronary artery calcium scores predicted from retinal photographs”, The Lancet Digital Health 3(5). The applicant revealed through a previous study that the Reti-CVD score can be used as a biomarker to predict the risk of cardiovascular disease that will occur in the future. The present invention newly reveals that the Reti-CVD score is closely related to the risk of various types of cardiovascular disease that will occur in the future, and provides a detailed algorithm for predicting the risk of each type of cardiovascular disease based on Reti-CVD.

[0042]

[0043] The diagnostic device can use a machine learning model to predict various information and risks of this specification, such as cardiovascular risk, risk by type of cardiovascular disease, and risk by subtype of type.

[0044] Fig. 1 is a block diagram illustrating a diagnostic device according to one embodiment. Referring to Fig. 1, the diagnostic device (10) may include a processor (12), a storage module (11), and a communication module (13).

[0045] The processor (12) may include one or more of a CPU (Central Processing Unit), a RAM (Random Access Memory), a GPU (Graphics Processing Unit), one or more microprocessors, and other electronic components capable of processing input data according to predetermined logic. In addition, there may be at least one processor (12).

[0046] The processor (12) can read the system program and various processing programs stored in the storage module (11). For example, the processor (12) can deploy processes, methods, etc. for performing the diagnosis described below on RAM and perform various processing according to the deployed program. For example, the processor (12) can process the algorithm described in this specification. The processor (12) can predict cardiovascular risk using a machine learning model. In addition, the processor (12) can implement various methods described in this specification.

[0047] The storage module (11) can store a diagnostic model. The storage module (11) can store a machine learning model, parameters and variables of the machine learning model, algorithms described in this specification, etc.

[0048] The storage module (11) can be implemented as a nonvolatile semiconductor memory, hard disk, flash memory, RAM, ROM (Read Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), or other types of tangible nonvolatile recording media.

[0049] The storage module (11) can store various processing programs, parameters for performing program processing, or data resulting from such processing. For example, the storage module (11) can store programs for performing the diagnosis described below, parameters, and data obtained by performing such programs. In addition, the storage module (11) can store various machine learning models described below.

[0050] Although not shown, the diagnostic device (20) may further include an input module. The input module may obtain user input. For example, the input module may obtain user input. In addition, the input module may obtain physical information of the subject described in this specification (at least one of height, weight, age, gender, race, smoking status, blood pressure (e.g., blood pressure level, high blood pressure), diabetes (or blood sugar level), and cholesterol level). In addition, the input module may obtain specific factors for each disease type to be described later. The processor (12) may obtain information input through the input module.

[0051] The communication module (13) can communicate with various devices (e.g., learning devices, client devices, etc.). For example, the diagnostic device (10) can be implemented in the form of a server that communicates with a client device. In addition, the diagnostic device (10) can be implemented in the form of a client device.

[0052]

[0053] <Example 1 - Prediction of the risk of stroke, myocardial infarction, and atrial fibrillation>

[0054] Figure 2 illustrates that the Reti-CVD score can serve as a biomarker for predicting individual risk for stroke, myocardial infarction, and atrial fibrillation. (a) shows the incidence of stroke, myocardial infarction, and atrial fibrillation in low-, intermediate-, and high-risk groups according to the Reti-CVD score. Additionally, (b) shows the adjusted hazard ratio (HR), HR trend, and C-index for each risk group. Here, CI represents the confidence interval.

[0055] (1) Stroke: The Reti-CVD score showed a significant correlation with the risk of stroke. Analysis results showed that the risk of stroke increased as the Reti-CVD score increased (adjusted HR trend, 1.50; 95% CI, 1.16-1.95; p=0.002). This suggests that the Reti-CVD score can be used as a useful biomarker for identifying high-risk patients requiring early preventive intervention.

[0056] (2) Myocardial Infarction (MI): The Reti-CVD score also showed a significant correlation with the risk of myocardial infarction (MI). As the Reti-CVD score increased, the risk of MI also increased (adjusted HR trend, 1.22; 95% CI, 1.01-1.47; p=0.036). This score can be used to identify individuals at high risk of myocardial infarction early and strengthen preventive measures.

[0057] (3) Atrial Fibrillation (AF): The Reti-CVD score was also closely associated with the risk of developing atrial fibrillation (AF) (adjusted HR trend, 1.32; 95% CI, 1.15–1.51; p<0.001). This suggests that the Reti-CVD score may be useful in assessing the risk of developing AF, and Reti-CVD may be an important tool for early risk assessment of AF and establishing preventive management strategies.

[0058]

[0059] <Example 2 - Risk Prediction for 14 Types of Cardiovascular Diseases, Including Myocardial Infarction and Peripheral Arterial Disease>

[0060] This study analyzed the correlation between the Reti-CVD score and 13 different cardiovascular disease (CVD) and arterial hypertension (HTN). A cross-sectional analysis was conducted on 45,980 participants from the UK Biobank. Logistic regression analysis was used to identify differential correlations with various CVDs, adjusting for factors such as hypertension, diabetes, dyslipidemia, and smoking.

[0061] Figures 3 and 4 are diagrams illustrating the correlation between Reti-CVD and various cardiovascular disease types. Referring to Figures 3 and 4, (a) and (b) are tables illustrating the relationship between the Reti-CVD score and the risk of each cardiovascular disease type. In (a) and (b), OR represents the odds ratio, LCL represents the lower control limit, and UCL represents the upper control limit. This table confirms that the high-risk group according to Reti-CVD is significantly associated with specific cardiovascular diseases compared to the low-risk group.

[0062] (1) Coronary artery disease or myocardial infarction (MI): A higher Reti-CVD score significantly increased the risk of developing coronary artery disease. In the high-risk group, the odds ratio (OR) was 10.37 (95% CI, 7.58-14.18), and a highly significant increase was observed across the overall trends.

[0063] (2) Peripheral vascular disease or peripheral arterial disease (PAD): As the Reti-CVD score increased, the risk of developing peripheral vascular disease also increased. The odds ratio for the high-risk group was 9.65 (95% CI, 2.94-31.64), demonstrating a statistically significant correlation.

[0064] (3) Atrial fibrillation (AF): A strong correlation was observed between the Reti-CVD score and the risk of developing atrial fibrillation, with the odds ratio in the high-risk group being 9.36 (95% CI, 6.51-13.45).

[0065] (4) Aortic Valve Stenosis (AVS): The higher the Reti-CVD score, the higher the risk of developing aortic valve stenosis, and the odds ratio for the high-risk group was 8.13 (95% CI, 1.87-35.35).

[0066] (5) Other diseases: A significant correlation with the Reti-CVD score was confirmed in various cardiovascular diseases such as stroke, heart failure (HF), and pulmonary embolism (PE).

[0067]

[0068] Examples 1 and 2 focused on analyzing the correlation between various cardiovascular disease types and related conditions using the Reti-CVD score and predicting the risk for each type. These results can be utilized to detect the possibility of cardiovascular disease that patients are currently unaware of and assess their risk. Specifically, based on correlation data between the Reti-CVD score and various CVD conditions, the diagnostic device can detect cardiovascular disease that may be present without the patient's knowledge. This is useful for detecting conditions that patients may not experience clear symptoms or that are not diagnosed in their early stages. Furthermore, based on the research results, the diagnostic device can analyze the Reti-CVD score and existing patient health data to prioritize recommendations for the most likely cardiovascular disease for the patient. This system can guide patients and healthcare professionals to consider additional testing for potential risk conditions, promoting early diagnosis and preventive treatment. For example, if a patient is at high risk for atrial fibrillation (AF) based on their Reti-CVD score, the diagnostic tool may recommend additional electrocardiogram testing or monitoring, as they likely have undiagnosed AF. Similarly, if another patient is at intermediate risk for peripheral vascular disease (PVD) or peripheral arterial disease (PAD), the diagnostic tool may guide them to more intensive testing for these conditions.

[0069]

[0070] 2. Cardiovascular predictive biomarkers for patients with underlying diseases

[0071] Reti-CVD can be a cardiovascular risk index that can more precisely predict the risk of secondary cardiovascular disease by considering a patient's gender, race, and pre-existing medical conditions. This score can be derived from a model trained on fundus photographs and coronary artery calcium (CAC) data. While the Reti-CVD score primarily assesses cardiovascular disease risk by analyzing a patient's fundus photographs, it can be expanded to become a predictive tool that comprehensively considers a patient's pre-existing medical conditions. This can more accurately predict the risk of secondary cardiovascular disease in high-risk patients, such as those with chronic conditions like diabetes, hypertension, and hyperlipidemia, or those who have previously experienced cardiovascular disease (heart failure, coronary artery disease, stroke, etc.). For example, the incidence of cardiovascular disease can differ by gender (male / female) and race. By incorporating these demographic variables into the Reti-CVD score, it is possible to individually predict cardiovascular risk in specific genders or races. Furthermore, patients with chronic diseases such as hypertension, hyperlipidemia, and diabetes may be at increased risk for additional cardiovascular disease. The Reti-CVD-based risk prediction model according to the present invention can predict a patient's risk of developing secondary cardiovascular disease (e.g., heart failure, coronary artery disease, stroke, etc.) by considering these pre-existing conditions. Patients who already have cardiovascular disease (e.g., heart failure, coronary artery disease, stroke, etc.) are at an additional risk for developing new types of cardiovascular disease. The Reti-CVD score may be useful in more accurately assessing the risk of developing secondary cardiovascular disease in this patient group.

[0072]

[0073] <Example 1 - Prediabetic and diabetic patients>

[0074] In this study, we used data from UK Biobank for prediabetic and diabetic patients, classified them into three groups (low-risk, intermediate-risk, and high-risk) based on the Reti-CVD score, and followed the patients to assess fatal and non-fatal cardiovascular disease types (coronary artery disease, ischemic stroke, and transient ischemic attack).

[0075] Figures 5 and 6 are diagrams illustrating the cardiovascular disease prediction performance for diabetic patients according to the present embodiment. These diagrams show the results of classifying patients into low-risk groups (n=550), intermediate-risk groups (n=276), and high-risk groups (n=275) in a 2:1:1 ratio according to the 50th and 75th percentiles based on the Reti-CVD score. This cardiovascular disease prediction score was derived based on retinal images of patients with prediabetes and diabetes. To evaluate how well the Reti-CVD score predicts fatal and non-fatal cardiovascular events, survival analysis using the Cox proportional hazards model and hazard ratios (HRs) was performed using longitudinal data from the UK Biobank. As shown in Figures 5 and 6, 138 (12.5%) of 1,101 patients with prediabetes and diabetes experienced cardiovascular disease events. During a median follow-up period of 11 years, these events occurred in 8.2% (45 / 550) of the low-risk group, 15.2% (42 / 276) of the intermediate-risk group, and 18.5% (51 / 275) of the high-risk group. Even after adjusting for factors such as age, sex, use of antihypertensive medications, use of statins, and smoking history, a significant association was observed between the CVD prediction score and the incidence of cardiovascular events. In particular, the risk of cardiovascular events was significantly increased in the intermediate-risk group (hazard ratio 1.57; 95% CI, 1.00-2.47) and high-risk group (hazard ratio 1.88; 95% CI, 1.19-2.98) compared to the low-risk group. In addition, a gradual increasing hazard ratio trend was observed (hazard ratio trend 1.36; 95% CI, 1.09-1.70). These results suggest that the Reti-CVD score may serve as a useful tool for risk stratification among patients with prediabetes and diabetes, and demonstrate significant potential in the management of high-risk patients.

[0076]

[0077] 3. Methods for predicting risk by type of cardiovascular disease

[0078] 3.1 Direct prediction model for cardiovascular disease risk by type (classification model)

[0079] The diagnostic device can perform the following operations to directly predict the risk of each type of cardiovascular disease. The machine learning (deep learning) model implemented in the diagnostic device can be trained using input fundus images and labels indicating the presence or absence of cardiovascular disease types such as atrial fibrillation (AF), coronary artery disease (CAD), peripheral artery disease (PAD), heart failure (HF), and stroke. Through this training process, features associated with cardiovascular disease are learned and recognized. Based on the analysis results, the processor of the diagnostic device can predict the likelihood of developing a type of cardiovascular disease in the subject of the corresponding fundus image. Deep learning algorithms that predict the risk of multiple cardiovascular disease types using fundus images can be designed in various ways. These include single-model, parallel, and multi-task models. The single-model approach uses a single deep learning model to simultaneously predict multiple cardiovascular disease types (e.g., atrial fibrillation (AF), coronary artery disease (CAD), peripheral artery disease (PAD), heart failure (HF), and stroke). This model receives a fundus photograph as input and has output nodes that individually predict the likelihood of occurrence for various cardiovascular disease types. The final layer of the model contains multiple output nodes, each returning a probability representing the likelihood of occurrence for a specific cardiovascular disease type. In a parallel modeling approach, individual deep learning models are built for each cardiovascular disease type. For example, there are independent models for predicting atrial fibrillation (AF), coronary artery disease (CAD), and peripheral artery disease (PAD). Each model receives a fundus photograph as input and predicts only its target cardiovascular disease type. The multitask modeling approach is similar to a single model, but it reinforces the learning of common features for predicting multiple cardiovascular disease types. A single model learns common features and, based on these features, makes individual predictions for each cardiovascular disease type.This model is designed to learn features related to multiple cardiovascular disease types from a common layer, then branch out for each cardiovascular disease type. Each branched path makes a detailed prediction tailored to that cardiovascular disease type.

[0080]

[0081] Figure 7 is a diagram for explaining a method for predicting risk by type of cardiovascular disease according to one embodiment.

[0082] Referring to FIG. 7, the processor of the diagnostic device may include a step of obtaining a fundus photograph (S100) and a step of obtaining cardiovascular disease diagnosis information (S200).

[0083] In step S100, the processor of the diagnostic device may acquire a fundus photograph. Furthermore, depending on the embodiment, the processor of the diagnostic device may perform preprocessing, enhancement, serialization, etc. on the acquired fundus photograph. Various techniques may be applied for this, and therefore, a detailed description thereof will be omitted.

[0084] In addition, in step S200, the processor of the diagnostic device may obtain cardiovascular disease diagnosis information. In the present specification, the diagnosis information may be expressed as diagnosis assistance information. The cardiovascular disease diagnosis information may include risk prediction information for each cardiovascular disease type. In the present specification, the risk prediction information for each cardiovascular disease type may include risk prediction information for each cardiovascular disease type, such as stroke, myocardial infarction (MI), atrial fibrillation (AF), heart failure (HF), and peripheral artery disease (PAD). For example, the risk prediction information for each cardiovascular disease type may include information on a probability value (score) and / or grade of occurrence of the corresponding cardiovascular disease type in a subject in a fundus photograph. Additionally, the risk prediction information for each type of cardiovascular disease may include information regarding the probability (score) and / or grade of developing the corresponding cardiovascular disease type within a given period (e.g., within 10 years or 5 years) for the subject in the fundus photograph. Furthermore, secondary auxiliary information, as described below, may be acquired in step S200. Specific implementation examples are described below.

[0085] <Example 1 - Reti-AF single model>

[0086] This example demonstrates the conceptual operation of the Reti-AF algorithm, which predicts the risk of atrial fibrillation (AF). The Reti-AF algorithm uses a patient's fundus photograph (retinal image) as input and learns the features of the retinal image to recognize patterns associated with the presence of atrial fibrillation (AF). Based on the analysis results, the algorithm predicts whether the input image is associated with AF or not. The Reti-AF algorithm can be executed by a processor in a diagnostic device.

[0087] In order to evaluate the performance of Reti-AF, a deep learning algorithm according to this embodiment, univariable Cox regression analysis and multivariable Cox regression analysis were performed. Fig. 8 is a diagram for explaining the prediction performance of atrial fibrillation risk according to one embodiment. Referring to Fig. 8, (a) is a table summarizing the contents related to atrial fibrillation risk (Reti-AF) through univariable Cox regression analysis, and (b) is a table summarizing the contents related to atrial fibrillation risk (Reti-AF) through multivariable Cox regression analysis.

[0088] Univariate Cox regression analysis showed that a higher Reti-AF score (when not adjusted for variables such as age and gender) was significantly associated with an increased incidence of AF. The hazard ratio (HR) of the high-risk group was 5.50 (95% CI, 4.08–7.41), demonstrating that Reti-AF is effective in predicting AF risk. Multivariate Cox regression analysis showed that Reti-AF remained significantly associated with AF incidence even after adjusting for variables such as age, gender, smoking status, diabetes, hyperlipidemia, and hypertension. In particular, age was an independently significant variable for AF incidence, enhancing the predictive performance of Reti-AF. The C-index of the multivariate model was 0.76, demonstrating improved performance compared to the univariate analysis. In conclusion, the Reti-AF algorithm showed significant performance in predicting the risk of AF incidence.

[0089]

[0090] 3.2 Cardiovascular biomarker (Reti-CVD)-based cardiovascular disease risk prediction model

[0091] 3.2.1 Basic Algorithm Configuration

[0092] Patients with existing cardiovascular disease (e.g., heart failure, coronary artery disease, stroke) are at increased risk for developing new cardiovascular subtypes. Therefore, it is necessary to predict the risk of developing each cardiovascular subtype.

[0093] FIG. 9 is a diagram illustrating a method for predicting risk by type of cardiovascular disease according to one embodiment. FIG. 9 shows a process of assessing a patient's overall risk of cardiovascular disease (first model (1100)) through a Reti-CVD prediction model (diagnostic model (1000)) and predicting the risk by type of cardiovascular disease (second model (1200)) through an analysis including specific factors by disease type. The first model (1100) and the second model (1200) may be included in the diagnostic model (1000).

[0094] Cardiovascular risk prediction model ('first model'): In this step, the patient's overall cardiovascular risk is assessed through the first model (1100). The first model (1100) learns the patient's fundus photograph and coronary artery calcium (CAC) score, and can predict the cardiovascular risk score, Reti-CVD, based on the input fundus photograph. In addition, the first model (1100) can predict various disease type-specific factors in addition to the Reti-CVD score. For example, the disease type-specific factors may include at least one of the parameters representing the subject's physical information, such as biological age, gender, height, weight, BMI index, and information on obesity (obesity score, BMI, body mass, body fat percentage, body muscle mass, obesity, etc.).

[0095] Laboratory markers specific to each disease type include hematocrit, red blood cell count, white blood cell count, hemoglobin, platelet count, total iron-binding capacity (TIBC), iron, ferritin (storage iron protein), total protein, albumin, asparagine transferase (AST), aminotransferase (ALT), γ-GT, alkaline phosphatase (ALP), globulin, hepatitis antigen, hepatitis antibody, blood sugar, glycated hemoglobin (HbA1C), blood urea nitrogen (BUN), creatinine (Cr), uric acid, total cholesterol, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, triglyceride (TG), apolipoprotein B (ApoB), and blood It may include at least one of the diagnostic numerical parameters such as lipoprotein(a) (Lp(a)), C-Reactive Protein (CRP), Erythrocyte Sedimentation Rate (ESR), bicarbonate level, pH level, hematuria, urine occult blood level, urine protein level, urine glucose level, urine pH, urine bilirubin level, urine ketone body level, urine urobilinogen level, urine glucose level, urine white blood cell count, urine creatinine level, urine albumin level, urine albumin / creatinine ratio. In addition, disease type-specific factors may include at least one of the electrocardiogram (ECG), systolic blood pressure (SBP), diastolic blood pressure (DBP), family history information, or patient clinical history information (Patient Clinical History, etc.).

[0096] Meanwhile, the first model (1100) learns the type of cardiovascular plaque, i.e., the presence or absence of soft plaque and hard plaque, along with the fundus photograph, and can predict the type of cardiovascular plaque based on the input fundus photograph. Details related to the prediction of the type of cardiovascular plaque are described in detail in section 3.2.4, Algorithm Reflecting the Type of Cardiovascular Plaque.

[0097]

[0098] Cardiovascular disease type-specific risk prediction model ('second model'): In this step, the data derived from the cardiovascular risk prediction model is used to predict the risk of each type of cardiovascular disease (e.g., atrial fibrillation, coronary artery disease, stroke, etc.). The risk of each disease type is evaluated by synthesizing the Reti-CVD score obtained from the first model (1100) and disease type-specific factors. The data input to the second model (1200) may be a feature vector of the fundus photograph extracted by the first model (1100) instead of the Reti-CVD score. Meanwhile, disease type-specific factors may be predicted by the first model as mentioned above, and in this case, the prediction result (output result) of the type-specific factors may be input to the second model (1200).

[0099] The second model (1200) may further include an input module. The input module can acquire disease-specific factors through user input. As shown in Fig. 9, if separate patient data (e.g., gender, age, blood test results, smoking status, obesity status, etc.) exists, the patient data can be input into a cardiovascular disease-specific risk prediction model. This model individually analyzes the likelihood of each cardiovascular disease to predict which cardiovascular disease a patient is more susceptible to.

[0100]

[0101] Outputting Cardiovascular Disease Diagnostic Information: The final step involves obtaining and outputting cardiovascular disease diagnostic information by synthesizing the overall cardiovascular risk (integrated score) and the risk levels for each type. This diagnostic information can include the current patient status, risk levels for each type, and related descriptive information. Additionally, it can provide supplementary information (e.g., secondary supplementary information, described below) for developing personalized prevention and treatment plans.

[0102]

[0103] <Example - Specific Patient Case>

[0104] The cardiovascular disease type-specific risk prediction model independently assesses the type-specific risk for each patient by considering the Reti-CVD score and type-specific covariates.

[0105]

[0106] 1. Patient A Case Overall Status High Reti-CVD score, hypertension, smoking history, female, 65 years old Stroke Risk Prediction: Very High Explanation: Patient A is predicted to have a very high risk of stroke due to the combination of a high Reti-CVD score, hypertension, smoking history, age (65 years), and being female. In the stroke model, all of these factors significantly contribute to the increased risk, so the patient is very likely to develop a stroke. Myocardial Infarction (MI) Risk Prediction: High Explanation: Patient A has a high Reti-CVD score, hypertension, smoking history, and age (65 years), which puts her at high risk for MI. However, menopause status and LDL cholesterol level are major factors, and being postmenopausal or having high LDL cholesterol level can further increase the risk of myocardial infarction. Atrial fibrillation (AF) risk prediction: Intermediate Explanation: Patient A is predicted to have an intermediate risk of atrial fibrillation due to high Reti-CVD, hypertension, smoking history, and age (65 years). Obesity is a major factor, but the risk can remain at an intermediate level in the absence of obesity. Heart failure (HF) risk prediction: High Explanation: Patient A's age (65 years), hypertension, and smoking history are major risk factors for heart failure. The high Reti-CVD score and the combination of risk factors predict a high risk of heart failure. Peripheral artery disease (PAD) risk prediction: High Explanation: Patient A's smoking history, hypertension, age (65 years), and being female all increase the risk of peripheral artery disease. Together with the high Reti-CVD score, the risk of PAD is predicted to be high.

[0107]

[0108] 2. Patient B Case Overall Status Intermediate Reti-CVD score, low cholesterol, obesity, family history of heart disease, male, 50 years old. Stroke Risk Prediction: Intermediate Explanation: Patient B has an intermediate Reti-CVD score and obesity, but does not have hypertension or a history of smoking, which are major risk factors for stroke. Considering that he is male and 50 years old, his stroke risk is predicted to be intermediate. Myocardial Infarction (MI) Risk Prediction: High Explanation: Patient A has obesity, a family history of heart disease, and is male, which puts Patient B at high risk for MI. Although his Reti-CVD score is intermediate and his LDL cholesterol level is low, obesity and a family history are important predictors of MI, so his risk is predicted to be high. Atrial Fibrillation (AF) Risk Prediction: High Explanation: Patient B has a high risk of atrial fibrillation due to his obesity and age of 50. Despite the intermediate Reti-CVD score, being male, obesity, and age predicted a high risk of atrial fibrillation. Heart failure (HF) risk prediction: Intermediate Explanation: Patient B has a family history of obesity and heart disease, but considering his relatively young age (50 years) and intermediate Reti-CVD score, the predicted risk of HF is intermediate. Peripheral artery disease (PAD) risk prediction: Intermediate Explanation: Patient B may be at a somewhat higher risk of PAD due to obesity, but considering his intermediate Reti-CVD score and age (50 years), the predicted risk is intermediate. Low LDL cholesterol also reduces the risk.

[0109]

[0110] 3.2.2 Detailed algorithm of the cardiovascular disease type-specific risk prediction model (Model 2)

[0111] For each cardiovascular disease type (stroke, myocardial infarction (MI), atrial fibrillation (AF), heart failure (HF), and peripheral arterial disease (PAD)), the model can consider the Reti-CVD score and type-specific covariates. For example, for stroke, specific covariates could include blood pressure, history of transient ischemic attack (TIA), atrial fibrillation status, and smoking status; for MI, LDL cholesterol, family history of heart disease, smoking, and hypertension; and for AF, age, history of heart disease, obesity, and hypertension. Additionally, for HF, age, history of heart disease, and smoking could be included; and for PAD, smoking, diabetes, hypertension, hyperlipidemia, and age could be included.

[0112] The diagnostic device's processor may use a multivariate Cox proportional hazards model or other appropriate statistical or machine learning model for each specific disease type. The following models may be included in the second model (1200). The following are examples of risk equations for each disease type performed by each model.

[0113] (1) Stroke model:

[0114] Stroke Risk = exp(β1 X Reti-CVD + β2 X Blood Pressure + β3 X TIA History + β4 X Atrial Fibrillation Status + β5 X Smoking + …)

[0115] (2) Myocardial infarction (MI) model:

[0116] Myocardial infarction risk = exp(β1 X Reti-CVD + β2 X LDL Cholesterol + β3 X Family History of Heart Disease + β4 X Smoking + β5 X Hypertension + …)

[0117] (3) Atrial fibrillation (AF) model:

[0118] Atrial fibrillation risk = exp(β1 X Reti-CVD + β2 X Age + β3 X History of Heart Disease + β4 X Obesity + β5 X Hypertension + …)

[0119] (4) Heart failure (HF) model:

[0120] Heart failure risk = exp(β1 X Reti-CVD + β2 X Age + β3 X Smoking + β4 X History of Heart Disease + …)

[0121] (5) Peripheral arterial disease (PAD) model:

[0122] Peripheral arterial disease risk = exp(β1 X Reti-CVD + β2 X Smoking + β3 X Diabetes + β4 X Hypertension + β5 X Hyperlipidemia + β6 X Age + …)

[0123]

[0124] In the above formula, β1, β2, etc. represent coefficients for each covariate, and each coefficient reflects the effect of the covariate on the risk. Each model outputs a risk estimate (e.g., hazard ratio or probability) for a specific disease type (stroke, myocardial infarction (MI), atrial fibrillation (AF), heart failure (HF), peripheral artery disease (PAD)). These risk estimates are calculated independently, and although each model uses the Reti-CVD score as a common input, the output values ​​can be produced individually. Furthermore, these risk estimates represent individual probabilities or hazard ratios for different outcomes and are not summed. Each risk can be interpreted independently based on a combination of the Reti-CVD score and disease type-specific factors. Each model can operate independently. That is, the risk for each cardiovascular type (stroke, myocardial infarction, atrial fibrillation, heart failure, peripheral arterial disease) is calculated based on the Reti-CVD score and type-specific covariates. These models do not influence each other, and risk prediction for each type can be performed independently.

[0125]

[0126] 3.2.3 Algorithm including recommended test judgment model

[0127] Fig. 10 is a diagram for explaining a method for predicting risk by type of cardiovascular disease according to another embodiment.

[0128] Referring to Figure 10, a recommended test judgment model can be added to the basic algorithm configuration introduced in section 3.2.1. The aforementioned content can be applied to the first model (cardiovascular risk prediction model, 1100) and the second model (cardiovascular disease type-specific risk prediction model, 1200), so a detailed description thereof will be omitted. The third model (1300) can also be included in the diagnostic model (1000).

[0129] Recommended Test Decision Model ('Third Model'): The third model (1300) determines what additional clinical tests are required for a patient based on the risk of each type of cardiovascular disease, and the processor of the diagnostic device can provide information on the additional clinical tests. Here, the additional clinical tests may include an electrocardiogram (ECG), an echocardiogram, an ankle-brachial index (ABI), an ankle-brachial pressure index (ABPI), a blood test, a urine test, an ultrasound, etc.

[0130] For example, if you are predicted to be at high risk for stroke, additional blood pressure monitoring or brain imaging may be recommended. If you are at high risk for myocardial infarction, additional blood tests (e.g., LDL levels) or cardiac function tests may be needed. If you are at high risk for heart failure, further evaluation of cardiac function may be necessary, such as a cardiac ultrasound or B-type natriuretic peptide (BNP) testing. If you are at high risk for peripheral artery disease (PAD), lower extremity Doppler ultrasound, ankle-brachial index (ABI), or other vascular tests may be recommended to assess blood flow. Furthermore, if you are predicted to be at high risk for AF, ECG monitoring (e.g., 24-hour Holter monitoring) may be recommended to detect paroxysmal atrial fibrillation. Patients with heart failure, peripheral artery disease, or atrial fibrillation may also be encouraged to monitor their own health through relevant mobile service apps, such as heart rate monitoring apps or pain recording apps.

[0131]

[0132] <Example - Recommended Test Output for Patients A and B>

[0133] This example demonstrates the process of determining whether additional clinical testing is necessary based on the predicted risk for cardiovascular disease types for patients A and B. Based on each patient's condition, appropriate additional testing is recommended for disease types predicted to be at high risk, enabling precise assessment and management of the patient's cardiovascular health.

[0134]

[0135] 1. Patient A Case Overall Status High Reti-CVD score, hypertension, smoking history, female, 65 years old Stroke Risk Prediction: Very High Recommended Testing: Because Patient A is predicted to be at a very high risk of stroke, additional clinical testing is recommended. Specifically, blood pressure monitoring and brain imaging tests (e.g., MRI or CT) are recommended. These tests will help identify the possibility of stroke early and develop a preventive treatment plan. Myocardial Infarction (MI) Risk Prediction: High Recommended Testing: Because Patient A is predicted to be at a high risk of MI, additional blood tests (e.g., checking LDL cholesterol levels) and cardiac function tests (e.g., echocardiography) are recommended. These tests help to more accurately assess risk factors for myocardial infarction and, if necessary, to modify treatment plans. Atrial Fibrillation (AF) Risk Prediction: Moderate Recommended Tests: Patient A is predicted to be at intermediate risk for AF, but considering his age (65 years), ECG monitoring (e.g., 24-hour Holter monitoring) may be considered. This may be useful for early detection of paroxysmal AF. Heart Failure (HF) Risk Prediction: High Recommended Tests: Patient A is predicted to be at high risk for heart failure, so further evaluation of cardiac function using echocardiography and BNP (B-type natriuretic peptide) testing is necessary. These tests will help monitor the progression of heart failure and plan appropriate treatment. Peripheral Arterial Disease (PAD) Risk Prediction: High Recommended Tests: Patient A is predicted to be at high risk for peripheral arterial disease, so evaluation of blood flow status using lower extremity Doppler ultrasound, ankle-brachial index testing, or other vascular tests is recommended. This allows us to check for stenosis of peripheral blood vessels and provide immediate treatment if necessary.

[0136]

[0137] 2. Patient B Case Overall Status Intermediate Reti-CVD score, low cholesterol, obesity, family history of heart disease, male, 50 years old. Stroke Risk Prediction: Intermediate Recommended Tests: Patient B is predicted to have an intermediate risk of stroke, so no further testing is necessary. Regular blood pressure monitoring and lifestyle modifications are recommended. Periodic retinal imaging can be used to continuously monitor his condition. Myocardial Infarction (MI) Risk Prediction: High Recommended Tests: Patient B is predicted to have a high risk of MI, so noninvasive cardiac testing (e.g., exercise / pharmacological stress testing) and cardiac ultrasound should be considered. This will help in the early detection and prevention of heart disease. Atrial Fibrillation (AF) Risk Prediction: High Recommended Tests: Patient B is predicted to have a high risk of atrial fibrillation, so ECG monitoring (e.g., 24-hour Holter monitoring) is recommended. This can help detect early signs of atrial fibrillation, allowing for early correction of the heart rhythm. Heart failure (HF) risk prediction: Moderate Recommended tests: Patient B is predicted to have an intermediate risk of heart failure, so periodic echocardiography and BNP tests are recommended to monitor cardiac function. Additional tests may be determined based on the patient's current condition. Peripheral artery disease (PAD) risk prediction: Moderate Recommended tests: Patient B is predicted to have an intermediate risk of peripheral artery disease, so periodic lower extremity Doppler ultrasound tests can be used to monitor blood flow. No additional tests are necessary, but lifestyle modifications to manage risk factors are recommended.

[0138] When the processor of the diagnostic device obtains new data on the recommended test, the processor of the diagnostic device re-enters the patient's data into the second model (1200), so that the risk of cardiovascular disease by type can be re-evaluated based on this, as shown in FIG. 10.

[0139]

[0140] 3.2.4 Algorithm reflecting cardiovascular plaque types

[0141] The algorithm reflecting the cardiovascular plaque type can be combined with the basic algorithm configuration introduced in 3.2.1 above or the algorithm that further includes the recommended test judgment model in 3.2.3 above. Since the aforementioned content can be applied to the first model (cardiovascular risk prediction model, 1100) and the second model (cardiovascular disease type-specific risk prediction model, 1200), a detailed description thereof will be omitted. As briefly mentioned above, the first model (1100) can learn the correlation between the input fundus photograph and the cardiovascular plaque type. Cardiovascular plaque types can be obtained from CT coronary angiography (CT-CA), cardiovascular magnetic resonance (CMR), 18F-FDG PET (Positron Emission Tomography), 18F-NaF PET (Positron Emission Tomography), coronary angiography, intravascular ultrasound (IVUS), IVUS-RF analysis, optical coherence tomography (OCT), optical frequency domain imaging (OFDI), intracoronary thermography, Raman spectroscopy, and near-infrared spectroscopy (NIRS). Specifically, a cardiovascular risk prediction model can be trained using a label indicating a soft plaque and its corresponding retinal image and a label indicating a hard plaque and its corresponding retinal image.Accordingly, the cardiovascular risk prediction model can obtain a probability value (and / or grade) for the probability that a subject in a retinal image has a soft plaque, or a probability value (and / or grade) for the probability that a subject in a retinal image has a hard plaque. The first model can predict plaque type as a probability value along with Reti-CVD and disease type-specific factors, and the probability value can be expressed as a score.

[0142] The plaque types predicted from the first model (1100) can be reflected in predicting the risk of cardiovascular disease by type. For example, soft plaques are associated with a higher risk of acute events due to the potential for rupture and embolism, and may be particularly associated with the risk of stroke. Furthermore, soft plaques increase the risk of heart failure and, due to the potential for embolization of soft plaques, may increase the risk of peripheral artery disease due to the increased risk of acute limb ischemia. In contrast, hard plaques are associated with progressive occlusion, which can lead to myocardial infarction over time. Furthermore, hard plaques increase the risk of heart failure and, due to the increased risk of chronic limb ischemia and intermittent claudication caused by progressive stenosis of the vessel, may increase the risk of peripheral artery disease.

[0143] Accordingly, the second model (1200) can predict the risk of each type of cardiovascular disease, such as stroke, myocardial infarction, heart failure, and peripheral artery disease, by considering the type of plaque predicted from the first model (1100) as a covariate for each detailed disease type. In addition, the third model (1300) can determine which clinical test is recommended for the patient by considering the type of plaque. For example, if the second model predicts a high risk of stroke and / or the third model predicts the presence of soft plaque or hard plaque, the third model (1300) can generate necessary information, such as blood biomarkers (e.g., blood coagulation test, lipid profile), blood pressure information, carotid artery condition information, and information necessary for diagnosing atrial fibrillation (age, blood pressure information, etc.), and recommend a clinical test to obtain the information. In addition, if a high risk of myocardial infarction disease is confirmed and / or a hard plaque is present, the third model (1300) can generate necessary information such as blood biomarkers (e.g., lipid profile), family history, etc., and recommend clinical tests to obtain the information. In addition, if a high risk of heart failure disease is confirmed based on heart failure disease diagnosis information and / or a soft plaque or hard plaque is present, the third model (1300) can generate necessary information such as blood biomarkers (e.g., BNP level), previous myocardial infarction disease history, and overall cardiac function indicators, and recommend clinical tests to obtain the information. In addition, if a high risk of peripheral artery disease is confirmed based on peripheral artery disease diagnosis information and / or a soft plaque or hard plaque is present, the third model can generate necessary information such as smoking information, diabetes information, and blood pressure information, and recommend clinical tests to obtain the information.

[0144]

[0145] 3.2.5 Risk Prediction Algorithm for Subtypes of Cardiovascular Disease

[0146] There are various forms (subtypes) of cardiovascular disease. Because each subtype differs in its pathological mechanism, symptoms, prognosis, and treatment, it is important to clearly distinguish them. The present invention goes beyond predicting the existing integrated cardiovascular risk using a machine learning model based on fundus photographs and provides an algorithm for predicting the risk of each type of cardiovascular disease. The processor of the diagnostic device can predict the risk of each subtype of cardiovascular disease based on the algorithm of the present invention. Since the basic model configuration is applicable to the description in Figure 9, a detailed description will be omitted.

[0147] For example, in the case of heart failure (HF), there are subtypes of heart failure with reduced ejection fraction (HFrEF), heart failure with preserved ejection fraction (HFpEF), and borderline heart failure with reduced ejection fraction (HFmrEF). Heart failure with reduced ejection fraction (HFrEF) is a heart failure condition in which the left ventricular ejection fraction (LVEF) is reduced to 40% or less, and the contractile function of the myocardium is reduced, significantly reducing the heart's ability to pump blood. In this case, disease-specific factors include previous myocardial infarction (MI), chronic hypertension, valvular disease, diabetes, and smoking, and echocardiography, cardiac MRI, and BNP (B-type natriuretic peptide) tests may be recommended. Heart failure with preserved left ventricular ejection fraction (HFpEF) is heart failure in which the left ventricular ejection fraction (LVEF) is preserved at 50% or higher. This is mainly due to a problem with the diastolic function of the ventricle, so the heart cannot relax sufficiently and its ability to fill with blood is reduced. In this case, obesity, hypertension, diabetes, older age, and female gender may be considered as disease-specific factors, and echocardiography, cardiac MRI, and BNP test may be recommended. In addition, heart failure with borderline left ventricular ejection fraction (HFmrEF) is heart failure in which the left ventricular ejection fraction (LVEF) is at the borderline level of 40-49%, and is considered an intermediate form between HFrEF and HFpEF. Older age, hypertension, diabetes, obesity, and heart valve disease may be considered disease-specific factors, and echocardiography, cardiac MRI, and MBP test may be recommended.

[0148] As with heart failure, stroke can be divided into ischemic stroke, thrombotic cerebral infarction (cerebral thrombosis), cerebral embolism (embolic cerebral infarction), lacunar infarction, hemorrhagic stroke, intracerebral hemorrhage, subarachnoid hemorrhage, transient ischemic attack, brainstem stroke, and stroke of unknown cause. For atrial fibrillation, it includes subtypes of paroxysmal atrial fibrillation, persistent atrial fibrillation, long-standing persistent atrial fibrillation, and permanent atrial fibrillation, and for myocardial infarction, it is divided into subtypes of type 1 (spontaneous myocardial infarction), type 2 (myocardial infarction due to ischemic imbalance), type 3 (myocardial infarction resulting in death when biomarker values ​​are not available), type 4a (myocardial infarction associated with percutaneous coronary intervention (PCI)), type 4b (myocardial infarction associated with stent thrombosis), and type 5 (myocardial infarction associated with coronary artery bypass grafting (CABG)).

[0149] Accordingly, the second model (cardiovascular disease risk prediction model) according to the present invention can predict subtype-specific risk by incorporating specific factors for each subtype into cardiovascular risk. Furthermore, the third model (recommended test judgment model) can recommend necessary clinical tests for each patient's cardiovascular disease, considering each subtype's specific risk factors.

[0150]

[0151] 3.2.6 Secondary Auxiliary Information

[0152] Diagnostic information may include the aforementioned cardiovascular risk score (Reti-CVD), plaque-based diagnostic information, disease-specific factors for each patient, risk and related information for each sub-disease type, and recommended test information. This section describes secondary auxiliary information additionally generated based on the diagnostic information. Specifically, the secondary auxiliary information includes secondary guidance information obtained based on the diagnostic information. For example, the secondary guidance information may include prescription information, action information, and management information.

[0153] Prescription information may refer to information on medications (e.g., prescription drugs) recommended to a subject to maintain or improve the risk of cardiovascular disease based on diagnostic information. In addition, prescription information may include information on the prescribed medication, timing of administration, and dosage. For example, prescription information may include information on prescriptions for one or more of a statin (including various agents such as simvastatin, atorvastatin, and rosuvastatin), an HMG-CoA reductase inhibitor, a PCSK9 inhibitor, a fibric acid derivative combination therapy, aspirin, a bile acid sequestrant, and nicotinic acid, an omega-3 fatty acid, ezetimibe, and a fibrate. Additionally, if the diagnostic information includes information that the subject has soft plaques and / or hard plaques, the prescribing information may include information about high-intensity statin therapy or a calcium channel blocker or ACE inhibitor for hypertension management.

[0154] Additionally, action information may refer to information about future recommended actions for the subject to maintain or improve cardiovascular disease risk based on diagnostic information. For example, additional screening information may include information about secondary diagnoses or medical treatments. For example, additional screening information may include information about additional required tests, hospitals / medical professionals who can perform the additional tests, and recommended procedures / surgeries.

[0155] Additionally, management information may include information on non-medical treatments recommended for the subject to maintain or improve cardiovascular disease risk based on diagnostic information. For example, management information may include lifestyle information, dietary information, exercise information, and information on non-prescription medications such as nutritional supplements to reduce the risk of cardiovascular disease. Furthermore, if the diagnostic information includes information indicating the presence of soft plaque in the subject, management information may include information on strict lifestyle modifications (e.g., diet, exercise, smoking cessation, etc.).

[0156] Additionally, if the diagnostic information includes information indicating that the subject has hard plaque, the management information may include information on long-term lifestyle improvement. Furthermore, in one embodiment, the diagnostic device may be linked to an external monitoring device. Here, the monitoring device may refer to a device that monitors the subject's lifestyle or behavior. For example, the monitoring device may include a portable device, a wearable device, a wellness measurement device, etc. Furthermore, the monitoring device may be the client device described above. For example, the monitoring device may include an imaging unit and may capture images of the inside and outside of the eye through the imaging unit.

[0157] In addition, the monitoring device can monitor various information such as the subject's activity level, exercise method, exercise time, food intake, intake amount, health supplement intake information, sleep time, sleep habits, heart rate, blood pressure, blood sugar level, body water content, oxygen content, body temperature, oxygen saturation, pulse wave, whether the subject has visited a hospital, whether the subject has undergone an examination, whether the subject has undergone a procedure / surgery, and eye images.

[0158] The processor of the diagnostic device can communicate with the monitoring device via a communication module, either wired or wirelessly.

[0159] The processor of the diagnostic device can provide guidance information to the monitoring device. The monitoring device can then provide various information to the subject based on the guidance information and the monitored information. For example, the monitoring device can obtain management information (e.g., lifestyle information, dietary information, exercise information) from the diagnostic device as guidance information, compare the management information with the monitored information, determine whether the monitored information matches the management information, and provide the determination result and / or additional information based on the result.

[0160] For example, if the monitored exercise time is less than the exercise time indicated in the management information, the monitoring device may provide the subject with information to exercise according to the management information. Furthermore, if the food intake in the monitored information corresponds to the food intake indicated in the management information, the monitoring device may provide the subject with information indicating that the food intake is appropriately consuming according to the management information.

[0161] Additionally, the processor of the diagnostic device can obtain monitored information from the monitoring device. The processor of the diagnostic device can provide various information to the subject based on the guide information and the monitored information. For example, the processor of the diagnostic device can compare the monitored information with the guide information, determine whether the monitored information matches the management information, and provide the determination result and / or additional information based on the comparison. The examples of the monitoring device described above can be applied to the operation of the processor of the diagnostic device.

[0162] Additionally, the processor of the diagnostic device can generate guidance information by reflecting monitoring information received from the monitoring device. For example, the processor of the diagnostic device can obtain status information (e.g., exercise status, lifestyle, eating habits) of the subject based on the monitoring information, and can modify guidance information determined as diagnostic information based on the subject's status information to suit the subject.

[0163] In one embodiment, the processor of the diagnostic device may provide guidance information using a predetermined database. For example, the diagnostic device may include a database that matches the score and / or grade of the diagnostic information with the guide information. For example, if the diagnostic information is expressed in three grades, the database may include guide information matching a low-risk grade (e.g., no prescription information, action information - information about the next treatment schedule, management information - providing dietary information, providing exercise information), guide information matching a medium-risk grade (e.g., no prescription information, action information - providing additional test information, management information - providing dietary information, providing exercise information, providing non-prescription drug information), and guide information matching a high-risk grade (e.g., prescription information - providing statin prescription information, action information - providing additional test information, providing recommended procedure / surgery information, management information - providing dietary information, providing exercise information, providing non-prescription drug information). The processor of the diagnostic device may provide guide information matching the diagnostic information based on the database.

[0164] In another embodiment, the processor of the diagnostic device may provide guide information using a machine learning model. For example, the diagnostic device may include a guide information model based on a machine learning model or a neural network model. The guide information model may be learned based on the score and / or grade of the diagnostic information and the guide information. In addition, the guide information model may also be learned together with the subject's physical information (at least one of height, weight, age, gender, race, smoking status, blood pressure (e.g., blood pressure level, high blood pressure), diabetes (or blood sugar level), and cholesterol level). Accordingly, the processor of the diagnostic device may input the score and / or grade of the diagnostic information and the subject's physical information into the guide information model to obtain guide information about the subject. In addition, depending on the embodiment, the guide information model may be included in the aforementioned diagnostic model or may be configured independently of the diagnostic model.

[0165]

[0166] Various embodiments of the present specification may be implemented as software including instructions stored in a machine-readable storage medium that can be read by a machine (e.g., a computer). The machine is a device that can call instructions stored in the storage medium and operate according to the called instructions, and may include an electronic device according to the disclosed embodiments. When the instructions are executed by a processor, the processor can perform a function corresponding to the instructions directly or under the control of the processor using other components. The instructions may include code generated or executed by a compiler or interpreter. The machine-readable storage medium may be provided in the form of a non-transitory storage medium. Here, the term "non-transitory storage medium" does not mean that it does not include a signal and is tangible, and does not distinguish whether data is stored semi-permanently or temporarily in the storage medium. For example, the "non-transitory storage medium" may include a buffer in which data is temporarily stored.

[0167] According to one embodiment, the method according to the various embodiments disclosed in the present specification may be provided as a computer program product. The computer program product may be traded as a product between a seller and a buyer. The computer program product may be distributed in the form of a device-readable storage medium (e.g., Compact Disc Read Only Memory, CD-ROM) or online through an application store (e.g., Play Store™). In the case of online distribution, at least a portion of the computer program product, e.g., a downloadable app, may be temporarily stored or temporarily generated in a storage medium such as the memory of a manufacturer's server, an application store's server, or a relay server.

[0168] Although the embodiments described above have been described by way of limited examples and drawings, those skilled in the art will appreciate that various modifications and variations can be made based on the above teachings. For example, appropriate results can still be achieved even if the described techniques are performed in a different order than described, and / or components of the described systems, structures, devices, circuits, etc. are combined or combined in a different manner than described, or are replaced or substituted with other components or equivalents.

[0169] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims described below.

Claims

1. In a diagnostic device for predicting risk by type of cardiovascular disease, Contains at least one processor, At least one processor uses a diagnostic model to predict the risk of each type of cardiovascular disease, The above diagnostic model is, A first model that predicts cardiovascular risk by analyzing the input fundus photograph; and A second model is included that receives the cardiovascular risk score predicted by the first model and the specific factors for each disease type, and predicts the risk for each type of cardiovascular disease. At least one processor, It is characterized in that it provides diagnostic information based on the cardiovascular risk predicted by the first model and the risk by type of cardiovascular disease predicted by the second model. A diagnostic device that predicts risk by type of cardiovascular disease.

2. In paragraph 1, A device for predicting risk by type of cardiovascular disease, characterized in that the above cardiovascular disease subtypes include at least one of myocardial infarction (MI), peripheral arterial disease (PAD), stroke, atrial fibrillation (AF), and heart failure (HF).

3. In paragraph 1, A diagnostic device for predicting risk of cardiovascular disease by type, characterized in that the specific factors for each disease type include at least one of the patient's age, sex, blood biomarker, urine test result, smoking status, obesity, electrocardiogram (ECG), or family history.

4. In paragraph 1, A cardiovascular disease type-specific risk prediction diagnostic device characterized in that the specific factors for each disease type are predicted by the first model or input from the outside.

5. In paragraph 1, A device for predicting risk by type of cardiovascular disease, characterized in that the above diagnostic information includes a risk grade and related information for each cardiovascular type.

6. In a diagnostic device for predicting risk by type of cardiovascular disease, Contains at least one processor, The at least one processor provides cardiovascular diagnostic information using a diagnostic model, The above diagnostic model is, The first model predicts cardiovascular risk by analyzing the input fundus photograph; A second model that receives the cardiovascular risk score predicted by the first diagnostic model and specific factors for each disease type and predicts the risk for each type of cardiovascular disease; and A third model is included that determines the clinical test recommended to the patient based on the risk of each type of cardiovascular disease predicted by the second model. At least one processor, A cardiovascular disease type-specific risk prediction diagnostic device characterized in that it provides cardiovascular diagnosis information based on the cardiovascular risk predicted by the first model and the cardiovascular disease type-specific risk predicted by the second model.

7. In paragraph 6, A device for predicting risk by type of cardiovascular disease, characterized in that the above cardiovascular disease subtypes include at least one of myocardial infarction (MI), peripheral arterial disease (PAD), stroke, atrial fibrillation (AF), and heart failure (HF).

8. In paragraph 6, A diagnostic device for predicting risk of cardiovascular disease by type, characterized in that the specific factors for each disease type include at least one of the patient's age, sex, blood biomarker, smoking status, obesity, urine test results, electrocardiogram (ECG), or family history.

9. In paragraph 6, A cardiovascular disease type-specific risk prediction diagnostic device characterized in that the specific factors for each disease type are predicted by the first model or input from an external source.

10. In paragraph 6, A diagnostic device for predicting risk of cardiovascular disease by type, characterized in that the additional clinical test includes at least one of an electrocardiogram (ECG), an echocardiogram, an ankle-brachial index (ABI), an ankle-brachial pressure index (ABPI), a blood test, a urine test, and an ultrasound.

11. A device for predicting risk by type of cardiovascular disease, characterized in that the cardiovascular diagnosis information in paragraph 6 includes risk grades and related information for each disease type or information related to recommended tests.

12. In a diagnostic device for predicting risk by type of cardiovascular disease, Contains at least one processor, The at least one processor provides cardiovascular diagnostic information using a diagnostic model, The above diagnostic model is, A first model that predicts cardiovascular risk and cardiovascular plaque type by analyzing the input fundus photograph; A second model that receives the cardiovascular risk score predicted by the first model and specific factors for each disease type and predicts the risk for each type of cardiovascular disease; and A third model is included that determines the clinical test recommended to the patient based on the risk of each type of cardiovascular disease predicted by the second model. A cardiovascular disease type-specific risk prediction diagnostic device characterized in that it provides cardiovascular diagnosis information based on the results predicted from the first to third models.

13. In paragraph 12, A device for predicting risk by type of cardiovascular disease, characterized in that the above cardiovascular disease subtypes include at least one of myocardial infarction (MI), peripheral arterial disease (PAD), stroke, atrial fibrillation (AF), and heart failure (HF).

14. In paragraph 12, A diagnostic device for predicting risk of cardiovascular disease by type, characterized in that the specific factors for each disease type include at least one of the patient's age, sex, blood biomarker, smoking status, obesity, urine test results, electrocardiogram (ECG), or family history.

15. In paragraph 12, A cardiovascular disease type-specific risk prediction diagnostic device characterized in that the specific factors for each disease type are predicted by the first model or input from an external source.

16. In paragraph 12, A diagnostic device for predicting risk of cardiovascular disease by type, characterized in that the additional clinical test includes at least one of an electrocardiogram (ECG), an echocardiogram, an ankle-brachial index (ABI), an ankle-brachial pressure index (ABPI), a blood test, a urine test, and an ultrasound.

17. A device for predicting risk by type of cardiovascular disease, characterized in that the cardiovascular diagnosis information in paragraph 12 includes risk grade and related information for each type of cardiovascular disease, plaque type, or recommended test related information.

18. A cardiovascular disease type-specific risk prediction diagnostic device characterized in that, in the case where a disease type-specific factor is input from outside to the second model in the 15th paragraph, the cardiovascular disease type-specific risk is re-evaluated based on the disease type-specific factor obtained from outside.

19. In a diagnostic device for predicting risk by type of cardiovascular disease, Contains at least one processor, At least one processor uses a diagnostic model to predict the risk of each type of cardiovascular disease, The above diagnostic model is, A first model for predicting cardiovascular risk by analyzing the input fundus photograph; A second model is included that receives the cardiovascular risk score predicted by the first model and specific factors for each subtype of disease type as input to predict the subtype risk for each disease type. At least one processor, It is characterized in that it provides diagnostic information based on the cardiovascular risk predicted by the first model and the risk of each detailed subtype predicted by the second model. A diagnostic device that predicts risk by type of cardiovascular disease.

20. In paragraph 19, The above cardiovascular disease types are characterized by including at least one of myocardial infarction (MI), peripheral artery disease (PAD), stroke, atrial fibrillation (AF), and heart failure (HF). A diagnostic device that predicts risk by type of cardiovascular disease.

21. In paragraph 20, The heart failure includes one or more subtypes of heart failure with reduced left ventricular ejection fraction (HFrEF), heart failure with borderline ejection fraction (HFmrEF), and heart failure with preserved ejection fraction (HFpEF); the stroke includes one or more subtypes of ischemic stroke, thrombotic cerebral infarction (cerebral thrombosis), cerebral embolism (embolic cerebral infarction), lacunar cerebral infarction, hemorrhagic stroke, intracerebral hemorrhage, subarachnoid hemorrhage, transient ischemic attack, brainstem stroke, and stroke of unknown etiology; the atrial fibrillation includes one or more subtypes of paroxysmal atrial fibrillation, persistent atrial fibrillation, long-standing persistent atrial fibrillation, and permanent atrial fibrillation; and the myocardial infarction includes one or more subtypes of type 1 (spontaneous myocardial infarction), type 2 (myocardial infarction due to ischemic imbalance), type 3 (myocardial infarction resulting in death when biomarker values ​​are not available), and type 4a (percutaneous coronary artery bypass grafting). characterized by including at least one subtype of type 4b (myocardial infarction associated with percutaneous coronary intervention (PCI), type 4b (myocardial infarction associated with stent thrombosis), and type 5 (myocardial infarction associated with coronary artery bypass grafting (CABG)). A diagnostic device that predicts risk by type of cardiovascular disease.

Images