Sensing systems and methods for providing diabetes decision support using continuously monitored analyte data - Patents.com
A continuous analyte monitoring system using glucose and lactate measurements with machine learning models addresses the inaccuracy of existing diabetes management by providing precise glycemic event predictions and treatment recommendations, improving patient outcomes during physical activity.
Patent Information
- Application Number
- JP2024501909
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2021-11-04
- Filing Date
- 2022-11-02
- Publication Date
- 2024-11-19
AI Technical Summary
Existing diabetes management systems fail to accurately predict and manage glycemic events during physical activity due to reliance on glucose levels and trends alone, which do not account for various influencing factors, leading to inaccurate guidance and potential deterioration of the patient's condition.
A continuous analyte monitoring system that combines glucose and lactate measurements with machine learning models to provide real-time or retrospective diabetes decision support, predicting glycemic events and offering treatment recommendations based on comprehensive physiological data.
Enhances the accuracy of predicting and managing glycemic events by considering multiple physiological factors, enabling early intervention and reducing the risk of complications such as hypoglycemia and hyperglycemia.
Smart Images

Figure 2024542916000001_ABST
Abstract
Description
[Technical field]
[0001] (CROSS REFERENCE TO RELATED APPLICATIONS) This application claims the benefit of and priority to U.S. Provisional Patent Application No. 63 / 263,540, filed November 4, 2021. The foregoing application is hereby incorporated by reference in its entirety. [Background technology]
[0002] Diabetes mellitus is a metabolic condition in which the pancreas cannot make enough insulin (type I or insulin-dependent) and / or insulin effectiveness is reduced (type 2 or non-insulin-dependent). Insulin is a hormone that allows the body to use glucose as energy or store it as fat. In the diabetic state, patients suffer from high glucose, called "hyperglycemia," which can cause many negative physiological effects associated with microvascular deterioration, such as nerve damage (neuropathy), kidney failure, skin ulcers, diabetic ketoacidosis, or bleeding into the vitreous humor of the eye. Conversely, a state of low blood sugar is called "hypoglycemia." Severe hypoglycemia can lead to myocardial damage, neurocognitive dysfunction, and in certain cases, even stroke or death.
[0003] In view of the above, it is very important for diabetic patients to be constantly aware of their glycemic status and know what steps to take to manage their diabetic condition in order to stay within the target glucose range and avoid physical complications. Thus, diabetic patients can benefit from continuous monitoring of glucose levels and trends along with real-time diabetic management guidance determined based on the monitored data. However, managing diabetes still presents many challenges to patients, clinicians, and caregivers, as a diverse set of factors affect a patient's glycemic status, and these factors may not always be reflected in the patient's glucose levels and glucose trends, thereby affecting the accuracy of the diabetic management guidance provided. [Brief description of the drawings]
[0004] So that the above-listed features of the present disclosure can be understood in detail, a more particular description, briefly summarized above, can be made by reference to embodiments, some of which are illustrated in the drawings. It should be noted, however, that the attached drawings merely illustrate certain exemplary embodiments of the present disclosure and therefore should not be considered as limiting its scope, since the description may be allowed for other equally valid embodiments. [Figure 1] 1 illustrates aspects of an exemplary decision support system that may be used in connection with implementing embodiments of the present disclosure. [Diagram 2] FIG. 1 conceptually illustrates an exemplary continuous analyte monitoring system including an exemplary continuous analyte sensor having sensor electronics, in accordance with certain aspects of the present disclosure. [Diagram 3] 2 illustrates example inputs and example metrics calculated based on the inputs for use by the decision support system of FIG. 1 according to certain embodiments disclosed herein. [Figure 4A] 1 illustrates a flow diagram of an exemplary method for providing decision support using a continuous analyte monitoring system configured to continuously measure at least glucose and lactate levels, according to some exemplary aspects of the present disclosure. [Figure 4B] 1 illustrates a flow diagram of an exemplary method for providing decision support using a continuous analyte monitoring system configured to continuously measure at least glucose and lactate levels, according to some exemplary aspects of the present disclosure. [Diagram 5] FIG. 1 is a flow diagram depicting a method for training a machine learning model to provide predictions of physical activity induced glycemic events according to certain embodiments of the present disclosure. [Figure 6] 6 is a block diagram illustrating a computing device configured to perform the operations of FIGS. 4A-5 in accordance with certain embodiments disclosed herein. FIG.
[0005] For ease of understanding, wherever possible, the same reference numerals have been used to designate identical elements that are common to each of the figures. It is contemplated that elements disclosed in one embodiment may be beneficially utilized in other embodiments without specific recitation. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0006] Managing diabetes presents many challenges to patients, clinicians, and caregivers because a diverse constellation of factors influences a patient's glycemic state and these factors may not always be reflected in a patient's glucose levels and glucose trends.
[0007] In particular, physical activity (e.g., exercise, training, outdoor activities, sporting events, and other events involving physical exertion of the patient) presents a particularly complex challenge in determining diabetes management guidance for diabetic patients. For example, physical activity may induce hyperglycemic or hypoglycemic events in the patient that may require immediate medical intervention. However, glucose levels and trends alone do not provide a comprehensive view of the patient's physiological state during physical activity and in certain cases may resemble, for example, the levels and trends of nutrition-related events rather than the patient's physical activity. As a result, basing diabetes management guidance solely on the glucose levels and / or glucose trends of a patient during physical activity may result in inaccurate or erroneous guidance being provided.
[0008] Because regular physical activity, in combination with diet and medication, is important for disease management in patients with metabolic disorders such as diabetes, there is a need in the art for improved systems and methods for characterizing a patient's physiological state during physical activity for improved diabetes management assistance, such as providing more accurate predictions of hyperglycemia and hypoglycemia.
[0009] In view of the above, diabetic patients may benefit from real-time diabetes management guidance that is determined based on the patient's physiological state. Traditionally, a patient's physiological state is determined based on glucose levels and glucose trends, which then informs predictions of hyperglycemic and / or hypoglycemic events (hereinafter "adverse glycemic events" or "glycemic events") and the type of guidance provided to the patient. However, managing diabetes presents many challenges to patients, clinicians, and caregivers, as a diverse collection of factors can affect a patient's glucose levels and glucose trends, and thus the accuracy of the guidance provided.
[0010] The inventors have recognized that, among other things, physical activity, e.g., workouts, training, sporting events, and other events involving a patient's physical exertion, presents a particularly complex challenge in determining a patient's physiological state when utilizing only glucose levels and glucose trends, thus making it difficult to provide accurate guidance for managing the patient's diabetes. For example, it is known that physical activity can induce adverse glycemic events in patients requiring medical intervention. However, glucose levels and trends alone do not provide a comprehensive view of a patient's physiological state during such physical activity. For example, in some cases, other confounding factors may contribute to the observed glucose levels and trends. In one example, the observed levels and trends may in certain cases resemble, for example, levels and trends of nutrition-related events rather than the patient's physical activity. Thus, a diagnostic system utilizing only glucose levels and trends may mischaracterize a patient's physiological state during physical activity, leading to inaccurate predictions of adverse glycemic events due to physical activity and inadequate guidance on how to manage such events, resulting in a worsening of the patient's condition.
[0011] To better characterize the physiological state of a patient engaged in physical activity, certain available diagnostic systems utilize additional data from various types of physical activity sensors, such as heart rate monitors and accelerometers. However, these physical activity sensors are more accurate in suggesting physical stress (i.e., intensity) levels for certain types of physical activity, e.g., cardiovascular activities such as running, compared to others, such as weightlifting, yoga, etc. Thus, the data provided from such physical activity sensors may not paint a complete picture of the patient's physical stress level during a particular type of activity, leading to suboptimal or inaccurate characterization of the patient's physiological state.
[0012] Because regular physical activity, in combination with diet and medication, is important for disease management in patients with metabolic disorders such as diabetes, there is a need in the art for improved systems and methods for characterizing a patient's physiological state during physical activity for improved diabetes management assistance, such as providing more accurate predictions of physical activity-induced adverse glycemic events, etc.
[0013] Accordingly, certain embodiments described herein provide a technical solution to the technical problem described above by providing a continuous analyte monitoring system configured to generate and analyze a combination of at least interstitial glucose and lactate measurements obtained from one or more continuous analyte sensors to provide a more accurate prediction of a diabetic patient's current or future physical activity induced glycemic events, as well as decision support for managing the patient's diabetes associated with physical activity, e.g., exercise. The decision support may include risk assessment, diagnosis, and / or recommendations for treatment of adverse glycemic events induced as a result of physical activity, or for nocturnal adverse glycemic events.
[0014] Continuous monitoring of at least a combination of glucose and lactate levels allows for a better characterization of the physiological state of a patient, especially one engaged in physical activity. The primary function of glucose is to provide energy for cellular function. Glucose is broken down into various by-products during cellular respiration, which produces an energy source called adenosine 5'-triphosphate (ATP). ATP is the primary molecule for storing and transferring energy in cells.
[0015] Depending on the presence of oxygen, glucose can be broken down via aerobic cellular respiration (in the presence of oxygen) or anaerobic cellular respiration (in the absence of oxygen). During aerobic cellular respiration, glucose is first converted to pyruvate in a process called glycolysis, releasing some initial ATP. Pyruvate is then converted to acetyl-CoA, which is processed through the citric acid cycle to release additional ATP. Electron carriers from these reactions transfer their electrons to a series of protein complexes in the inner membrane of mitochondria called the electron transport chain, and the flow of these electrons through the electron transport chain drives the synthesis of additional ATP.
[0016] During anaerobic cellular respiration, glucose is first converted to pyruvate via anaerobic glycolysis. Pyruvate is then converted to lactate, the conjugate base of lactate, by the enzyme lactate dehydrogenase (LDHA). The lactate produced is then used as fuel by surrounding tissues or transported to other tissues for use as an energy source. This process produces less energy than the full energy potential of a glucose molecule, but is very rapid and provides an immediate energy source.
[0017] Aerobic cellular respiration occurs during normal metabolism and low to moderate levels of physical activity when oxygen is readily available for use in the breakdown of glucose for energy. However, anaerobic cellular respiration occurs during elevated metabolism and moderate to high levels of physical activity when muscle oxygen demand exceeds oxygen supply. Thus, in certain circumstances, the presence of lactate may be a better indicator of the type and intensity of physical activity engaged in by a user than, for example, glucose levels alone. Thus, continuous monitoring of lactate, in addition to glucose, may be required to assess the physical activity engaged in by a user in order to accurately predict glycemic events resulting from a user engaging in such physical activity.
[0018] In certain embodiments described herein, the continuous analyte monitoring system may provide decision support to the patient based on various collected data including analyte data, patient information, secondary sensor data (e.g., non-analyte data), patient input, etc. For example, the analyte data may include continuously monitored lactate data in addition to other continuously monitored analyte data such as glucose, ketones, glycerol, electrolytes such as sodium and potassium, calculated measurements such as anion gap, and other suitable analytes. The continuously monitored lactate data may be indicative of or may be used to determine the patient's lactate level, lactate production level, and / or lactate clearance rate.
[0019] As explained above, the collected data also includes patient information, which may include age, sex, glucose threshold levels, previous exercise-related or nutrition-related event data, fitness level, activity frequency, etc. Secondary sensor data may include accelerometer data, step count data, exercise equipment power meter data, global positioning system (GPS) data, heart rate, heart rate reserve, heart rate variability (HRV), electrocardiogram (EKG) data, electromyogram (EMG), respiratory rate, body temperature, blood pressure, galvanic skin response, oxygen uptake data (e.g., VO 2 max), sleep, impedance data, etc.
[0020] According to certain embodiments of the present disclosure, the decision support system presented herein is designed to provide a diabetes decision support to predict current or future physical activity induced glycemic events for a diabetic patient and to assist the patient in managing diabetes. Providing a diabetes decision support may involve, for example, using a large amount of collected data, including the analyte data, patient information, and secondary sensor data mentioned above, (1) automatically detecting a patient engaged in physical activity, (2) determining relevant parameters of the physical activity engaged by the patient (e.g., intensity, duration, and / or type of physical activity), (3) predicting current or future glycemic events based in part on the physical activity engaged by the predicting patient, and (4) making patient-specific treatment decisions or recommendations for diabetes. In other words, the decision support system presented herein may provide information useful in directing and improving care for a diabetic patient.
[0021] In certain embodiments, the decision support systems described herein may use various algorithms or artificial intelligence (AI) models, such as machine learning models, trained based on patient-specific and / or population data to provide real-time or retrospective decision support to a patient based on collected information about the patient. For example, certain aspects are directed to algorithms and / or machine learning models designed to predict a patient's current or future physical activity-induced glycemic events based in part on the physical activity engaged in by the patient. The algorithms and / or machine learning models may be used in combination with one or more continuous analyte sensors configured to continuously measure at least glucose and lactate levels to provide real-time or retrospective predictions of the patient's current or future glycemic events. In particular, the algorithms and / or machine learning models may take into account parameters such as the patient's normal lactate and glucose ranges (e.g., normal minimum and maximum levels) when predicting glycemic events. Based on these parameters, the algorithms and / or machine learning models may provide predictions of the patient's current or future physical activity-induced glycemic events, as well as treatment recommendations to manage (e.g., prevent or mitigate) the predicted glycemic events. The algorithms and / or machine learning models may take into account population data, personalized patient-specific data, or a combination of both when predicting a patient's glycemic events.
[0022] According to certain embodiments, prior to deployment, the machine learning model is trained with training data including, for example, population data. As described in more detail herein, the population data may be provided in the form of a data set including data records or samples of the medical history of a diabetic patient. Each data record is then characterized (e.g., refined into a set of one or more features or predictive variables) and labeled. Data labeling is the process of adding one or more meaningful and informative labels to provide context to the data for learning by the machine learning model. In certain embodiments, each data record is labeled with one or more of a glycemic event (e.g., including a type of glycemic event, a time associated with the glycemic event, a glucose level associated with the glycemic event, a duration of the glycemic event, etc.), a treatment provided in response to the glycemic event (e.g., including a type of treatment, an amount / dosage of the treatment, an outcome of the treatment, etc.), etc. The features associated with each data record may be used as input to the machine learning model, and the generated output may be compared to the label assigned to each of the data records. The model may calculate a loss based on the difference between the generated output and the provided label. This loss may then be used to modify the internal parameters or weights of the model. By iteratively processing the features associated with each data record corresponding to each historical patient, the model can be iteratively refined to generate accurate predictions of glycemic events for patients engaging in physical activity.
[0023] The combination of a continuous analyte monitoring system with machine learning models and / or algorithms for predicting glycemic events in diabetic patients engaged in physical activity, as provided by the decision support system described herein, allows for real-time or retrospective prediction and provision of treatment decisions or recommendations to enable early intervention. In particular, the decision support system may be used to provide early warning of hyperglycemic or hypoglycemic events. Early detection of such events may allow for earlier intervention to prevent or mitigate such events from leading to further physical complications, which in some cases may even lead to hospitalization and death.
[0024] In addition, through the combination of a continuous analyte monitoring system with machine learning models and / or algorithms for predicting glycemic events in diabetic patients engaged in physical activity, the decision support system described herein may provide the necessary accuracy and reliability that diabetic patients expect from a continuous analyte monitoring system. For example, continuous monitoring of multiple analytes, e.g., glucose and lactate, combined with machine learning models and / or algorithms for predicting glycemic events affected by physical activity may better characterize a patient's physiological state during and / or after physical activity compared to utilizing glucose levels and / or glucose trends alone. Thus, the decision support system described herein may provide more reliable predictions of glycemic events in diabetic patients engaged in physical activity.
[0025] Exemplary decision support system including an exemplary continuous analyte sensor for predicting glycemic events in patients engaging in physical activity - Patents.com 1 illustrates an exemplary decision support system 100 for predicting current or future physical activity induced glycemic events of a user 102 (individually referred to herein as user(s) and collectively referred to herein as users(s)) using a continuous analyte monitoring system 104 configured to continuously measure at least glucose and lactate levels. The user may be a diabetic patient, in certain embodiments, or in some cases, a caregiver for the patient. In certain embodiments, the system 100 includes the continuous analyte monitoring system 104, a display device 107 running an application 106, a decision support engine 114, a user database 110, a medical history record database 112, a training server system 140, and the decision support engine 114, each of which is described in more detail below.
[0026] As used herein, the term "analyte" is a broad term used in its ordinary sense, including, but not limited to, referring to a substance or chemical component within a biological fluid (e.g., blood, interstitial fluid, cerebrospinal fluid, lymphatic fluid, or urine) that can be analyzed. Analytes can include naturally occurring substances, man-made substances, metabolites, and / or reaction products. Analytes measured by the device and method include, but are not limited to, glucose, acarboxyprothrombin; acylcarnitines; adenine phosphoribosyltransferase; adenosine deaminase; albumin; α-fetoprotein; amino acid profile (arginine (Krebs cycle), histidine / urocanic acid, homocysteine, phenylalanine / tyrosine, tryptophan); andrenostenedione; antipyrine; arabinitol enantiomers; arginase; benzoylecgonine (cocaine); biotinidase; biopterin; c-reactive protein; carnitine; carnosinase; CD4; ceruloplasmin; chenodeoxycholic acid; chloroquine; cholesterol; cholinesterase; conjugated 1-β hydroxy-cholic acid; cortisol; creatine kinase; creatine kinase MM isoenzyme; cyclosporine; Porin A; d-penicillamine; deethylchloroquine; dehydroepiandrosterone sulfate; DNA (acetylation polymorphism, alcohol dehydrogenase, α1-antitrypsin, cystic fibrosis, Duchenne / Becker muscular dystrophy, glucose-6-phosphate dehydrogenase, hemoglobin A, hemoglobin S, hemoglobin C, hemoglobin D, hemoglobin E, hemoglobin F, D-Punjab, β-thalassemia, hepatitis B virus, HCMV, HIV-1, HTLV-1, Leber's hereditary optic neuropathy, MCAD, RNA, PKU, Plasmodium vivax, sex differentiation, 21-deoxycortisol); desbutylhalofantrine; dihydropteridine reductase; diphtheria / tetanus antitoxin; erythrocyte arginase; erythrocyte protoporphyrin; esterase D; fatty acids / acylglycines; free beta-human chorionic gonadotropin; free erythrocyte porphyrin; free thyroxine (FT4);Free tri-iodothyronine (FT3); Fumarylacetoacetase; Galactose / gal-1-phosphate; Galactose-1-phosphate uridyltransferase; Gentamicin; Glucose-6-phosphate dehydrogenase; Glutathione; Glutathione peroxidase; Glycocholate; Glycerol; Glycosylated hemoglobin; Halofantrine; Hemoglobin variants; Hexosaminidase A; Human erythrocyte carbonic anhydrase enzyme I;17-α-hydroxyprogesterone;hypoxanthine phosphoribosyltransferase;immunoreactive trypsin;lactate;lead;lipoproteins ((a), B / A-1, β);lysozyme;mefloquine;netilmicin;phenobarbitone;phenytoin;phytanic acid / pristanic acid;progesterone;prolactin;prolidase;purine nucleoside phosphorylase;potassium;quinine;reverse triiodothyronine (reverse tri-iodothyronine, rT3); selenium; serum pancreatic lipase; sisomicin; somatomedin C; specific antibodies (adenovirus, antinuclear antibody, anti-zeta antibody, arbovirus, pseudorabies virus, dengue virus, guinea worm, tapeworm, ameba histolytica, enterovirus, giardiasis, Helicobacter pylori, hepatitis B virus, herpes virus, HIV-1, IgE (atopic disease), influenza virus, Leishmania donovani, Leptospirosis, measles / mumps / rubella, Mycobacterium leprae, Mycoplasma pneumoniae, myoglobin, Onchocerca volvulus, parainfluenza virus, Malaria leiomyosa, poliovirus, Pseudomonas aeruginosa, respiratory syncytial virus, Rickettsia (tsutsugamushi disease), Schistosoma mansoni, Toxoplasma gondii, Treponema pallidum, Trypanosoma cruzi / langeri, Vesicular stomatitis virus, Wuchereria bancrofti, Yellow fever virus); specific antigens (Hepatitis B virus, HIV-1); succinylacetone; sulfadoxine; theophylline; thyrotropin (TSH); thyroxine (T4); thyroxine-binding globulin; trace elements; transferrin; UDP-galactose-4-epimerase; urea; uroporphyrinogen I synthase; vitamin A; leukocytes; and zinc protoporphyrin.
[0027] Salts, sugars, proteins, fats, vitamins, and hormones naturally occurring in blood or interstitial fluids can also constitute analytes in certain embodiments. Analytes can be naturally occurring in biological fluids, such as metabolites, hormones, antigens, antibodies, and the like. Alternatively, analytes can be introduced into the body or can be exogenous, such as contrast agents for imaging, radioisotopes, chemical agents, fluorocarbon-based synthetic blood, or drugs or pharmaceutical compositions, including insulin; glucagon, ethanol; cannabis (marijuana, tetrahydrocannabinol, hashish); inhalants (nitrous oxide, amyl nitrite, butyl nitrite, chlorohydrocarbons, hydrocarbons); cocaine (crack cocaine); stimulants (amphetamine, methamphetamine, ritalin, silurt, preludine, didrex, prestate, borane, sandrex, pregin); antidepressants ( Medications that may be used include, but are not limited to, barbiturates, methaqualone, valium, librium, miltuna, serax, equanil, tranxine, and other tranquilizers; hallucinogens (phencyclidine, lysergic acid, mescaline, peyote, psilocybin); narcotics (heroin, codeine, morphine, opium, meperidine, percocet, percodan, tasionex, fentanyl, darvon, talwin, lomotil); synthetic narcotics (fentanyl, meperidine, amphetamine, methamphetamine, and analogs of phencyclidine, e.g., ecstasy); anabolic steroids; and nicotine. Metabolites of drugs and pharmaceutical compositions are also contemplated analytes. Analytes such as neurochemicals and other chemicals produced in the body, such as, for example, ascorbic acid, uric acid, dopamine, noradrenaline, 3-methoxytyramine (3MT), 3,4-dihydroxyphenylacetic acid (DOPAC), homovanillic acid (HVA), 5-hydroxytryptamine (5HT), and 5-hydroxyindoleacetic acid (FHIAA), as well as intermediates of the citric acid cycle, can also be analyzed.
[0028] Analytes measured and analyzed by the devices and methods described herein include glucose, lactate, ketones, and in some cases other analytes may be considered, including but not limited to those listed above.
[0029] In certain embodiments, the continuous analyte monitoring system 104 is configured to continuously measure one or more analytes and transmit the analyte measurements to the display device 107 for use by the application 106. In some embodiments, the continuous analyte monitoring system 104 transmits the analyte measurements to the display device 107 through a wireless connection (e.g., a Bluetooth connection). In certain embodiments, the display device 107 is a smartphone. However, in certain other embodiments, the display device 107 may instead be any other type of computing device, such as a laptop computer, a smart watch, a fitness tracker, a tablet, or any other computing device capable of running the application 106. The continuous analyte monitoring system 104 may be described in more detail with respect to FIG. 2.
[0030] The application 106 is a mobile health application configured to receive and analyze analyte measurements from the analyte monitoring system 104. For example, the application 106 stores information about the user in the user profile 118, including the user's analyte measurements, for processing and analysis, and for use by the decision support engine 112 to provide decision support recommendations or guidance to the user.
[0031] The decision support engine 114 refers to a set of software instructions having one or more software modules, including a data analysis module (DAM) 116. In certain embodiments, the decision support engine 114 executes entirely on one or more computing devices in a private or public cloud. In such embodiments, the application 106 communicates with the decision support engine 114 via a network (e.g., the Internet). In some other embodiments, the decision support engine 114 executes partially on one or more local devices, such as the display device 107, and partially on one or more computing devices in a private or public cloud. In some other embodiments, the decision support engine 114 executes entirely on one or more local devices, such as the display device 107. As discussed in more detail herein, the decision support engine 114 may provide decision support recommendations to the user via the application 106. The decision support engine 114 provides decision support recommendations based on information contained in the user profile 118.
[0032] The user profile 118 may include information collected about the user from the application 106. For example, the application 106 provides a set of inputs 128 including analyte measurements associated with one or more analytes received from the continuous analyte monitoring system 104, which are stored in the user profile 118. In certain embodiments, the inputs 128 provided by the application 106 include other data in addition to the analyte measurements. For example, the application 106 may obtain the additional inputs 128 through manual user input, one or more other non-analyte sensors or devices, other applications running on the display device 107, etc. The non-analyte sensors and devices include, but are not limited to, one or more of an insulin pump, a respiration sensor, a sensor or device provided by the display device 107 (e.g., an accelerometer, a camera, a global positioning system (GPS), a heart rate monitor, etc.) or other user accessories (e.g., a smart watch or fitness tracker), or any other sensor or device that provides relevant information about the user (e.g., a sensor on exercise equipment). The inputs 128 of the user profile 118 provided by the application 106 are described in further detail below with respect to FIG.
[0033] The DAM 116 of the decision support engine 114 is configured to process the set of inputs 128 to determine one or more metrics 130. The metrics 130, which will be discussed in more detail below with respect to FIG. 3, may, at least in some cases, generally be indicative of the user's health or condition, such as one or more of the user's physiological condition, trends associated with the user's health or condition, etc. In certain embodiments, the metrics 130 may then be used by the decision support engine 112 as inputs to provide guidance to the user. As shown, the metrics 130 are also stored in the user profile 118.
[0034] The user profile 118 also includes demographic information 120, disease information 122, and / or medication information 124. In certain embodiments, such information may be provided through user input or may be obtained from a particular data store (e.g., electronic medical record, etc.). In certain embodiments, the demographic information 120 may include one or more of the user's age, body mass index (BMI), ethnicity, gender, etc. In certain embodiments, the disease information 122 may include information regarding the user's one or more diseases, including relevant information regarding the user's diabetic status and / or other conditions (e.g., liver disease, renal disease, etc.). In certain embodiments, the disease information 122 may also include length of time since diagnosis, level of disease control, level of adherence to disease management therapy, other types of diagnoses (e.g., heart disease, obesity), etc. In certain embodiments, disease information 122 may include other measures of health (e.g., heart rate, stress, sleep, etc.) or health status (e.g., cardiovascular endurance, muscle strength and / or power, muscular endurance, and other measures of health status).
[0035] In certain embodiments, medication information 124 may include information regarding the amount and type of medication taken by the user.In certain embodiments, the dosing information includes insulin (e.g., short-acting insulin, rapid-acting insulin (insulin aspart, insulin glulisine, insulin lispro), intermediate-acting insulin (insulin isophane), long-acting insulin degludec, indulin detemir, insulin glargine, insulin), combination insulin), amylinomimetics, alpha-glucosidase inhibitors (e.g., acarbose, miglitol), biguanides (e.g., metformin-alogliptin, metformin-dopamine), metformin-dopamine, ... metformin-canagliflozin, metformin-dapagliflozin, metformin-empagliflozin, metformin-glipizide, metformin-glyburide, metformin-linagliptin, metformin-pioglitazone, metformin-repaglinide, metformin-rosiglitazone, metformin-saxagliptin, metformin-sitagliptin), dopamine agonists (e.g., bromocriptine), dipeptidyl peptidase-4 (dipeptidyl peptidase-4 (DPP-4) inhibitors (e.g., alogliptin, alogliptin-pioglitazone, linagliptin, linagliptin-empagliflozin, saxagliptin, sitagliptin, simvastatin), glucagon-like peptide-1 (GLP-1) receptor agonists (e.g., albiglutide, dulaglutide, exenatide, repagliptin, semagligliflozin), meglitinides (e.g., nateglinide, repaglinide), sodium-glucose transporters (sodium-glucose The information may include information regarding the consumption of one or more medications for the management of the user's diabetic condition, such as glutamatidine diabetic reductase (SGLT) 2 inhibitors (e.g., dapagliflozin, canagliflozin, exenatide, liraglutide, semagligliflozin), sulfonylureas (e.g., glimepiride, glimepiride-pioglitazone, glimepiride-rosiglitazone, gliclazide, glipizide, glyburide, chlorpropamide, tolazamide, tolbutamide), thiazolidinediones (e.g., rosiglitazone, pioglitazone), and other medications.In certain embodiments, the medication information may include information regarding the consumption of one or more medications for the management or treatment of other diseases or conditions of the user, including medications for cholesterol, high blood pressure, heart disease, etc., in addition to supplements for promoting general health and metabolism, such as vitamins.
[0036] The data stored in the user profile 118 may maintain time series data collected about a user (e.g., a patient) over the period of time that the user utilizes the continuous analyte monitoring system 104 and application 106. For example, analyte data for a user who has used the continuous analyte monitoring system 104 and application 106 for five years to manage the user's diabetic condition may have the analyte data associated with the user maintained in the user profile 118 for the five years.
[0037] Additionally, the data stored in the user profile 118 may provide time series data collected over the lifetime of the user. For example, the data may include information associated with the user throughout the user's lifetime that is indicative of the user's physiological state, the user's fitness level, the user's glucose level, the user's lactate level, the user's ketone level, the state / condition of one or more of the user's organs, the user's habits (e.g., alcohol consumption, activity level, food consumption, etc.), prescribed medications, etc.
[0038] In certain embodiments, the user profile 118 is dynamic because at least a portion of the information stored in the user profile 118 may be revised or updated over time and / or new information may be added to the user profile 118 by the decision support engine 114 and / or the application 106. Thus, the information in the user profile 118 stored in the user database 110 provides an up-to-date repository of information about the user.
[0039] User database 110, in some embodiments, refers to a storage server operating in, for example, a public or private cloud. User database 110 may be implemented as any type of data store, such as a relational database, a non-relational database, a key-value data store, a file system including a hierarchical file system, etc. In some example implementations, user database 110 is distributed. For example, user database 110 may comprise multiple persistent storage devices that are distributed. Additionally, user database 110 may be replicated such that the storage devices are geographically distributed.
[0040] The user database 110 includes user profiles 118 associated with multiple users, including users who similarly interact or have previously interacted with the applications 106 on their devices. The user profiles stored in the user database 110 are accessible to the applications 106 as well as the decision support engine 114. The user profiles in the user database 110 may be accessible to the applications 106 and the decision support engine 114 via one or more networks (not shown). As described above, the decision support engine 114, and more specifically, the data analysis module (DAM) 116 of the decision support engine 114, may fetch inputs 128 from the user's profile 118 stored in the user database 110 and then calculate one or more metrics 130, which may be stored in the user's profile 118 as application data 126.
[0041] In certain embodiments, the user profiles 118 stored in the user database 110 may also be stored in the medical history record database 112. The user profiles 118 stored in the medical history record database 112 may provide a repository of up-to-date information and medical history information for each user of the application 106. Thus, the medical history record database 112 essentially provides all data related to each user of the application 106, the data being stored using a timestamp. The timestamp associated with any information stored in the medical history record database 112 may identify, for example, when the information was obtained and / or updated.
[0042] Additionally, in certain embodiments, historical record database 112 may include data for one or more patients who are not users of continuous analyte monitoring system 104 and / or application 106. For example, historical record database 112 may include information (e.g., user profiles) about one or more patients, diagnosed or not diagnosed with diabetes, analyzed, for example, by a health care physician (or other known method). The data stored in historical record database 112 may be referred to herein as population data.
[0043] Although depicted as separate databases for conceptual clarity, in some embodiments, the user database 110 and the medical history record database 112 may operate as a single database, which may be a storage server running on a public or private cloud.
[0044] As mentioned above, the decision support system 100 is configured to provide predictions of current or future glycemic events for a user with diabetes engaging in physical activity, as well as specific treatment decisions or recommendations for diabetes management, using a continuous analyte monitoring system 104 configured to continuously measure at least glucose and lactate levels. In certain embodiments, the decision support engine 114 is configured to provide real-time or retrospective diabetes decision support, thus enabling the provision of early predictions and / or early treatment recommendations to prevent or mitigate predicted glycemic events. In particular, the decision support engine 114 may be used to collect information associated with the user in a user profile 118, to detect when the user is engaged in physical activity, to determine parameters of the physical activity, to predict current or future glycemic events resulting from the physical activity, and to perform its analysis to provide one or more recommendations for treatment based at least in part on the prediction. The user profile 118 may be accessible to the decision support engine 114 via one or more networks (not shown) to perform such analysis.
[0045] In certain embodiments, the decision support engine 114 may utilize one or more trained machine learning models that enable the decision support engine 114 to perform analysis on information collected / received from the user profile 118. In the illustrated embodiment of FIG. 1, the decision support engine 114 may utilize trained machine learning models provided by a training server system 140. Although depicted as separate servers for conceptual clarity, in some embodiments the training server system 140 and the decision support engine 114 may operate as a single server or system. That is, models may be trained and used by a single server, or trained by one or more servers and deployed for use on one or more other servers or systems.
[0046] The training server system 140 is configured to train the machine learning models using training data, which may include data (e.g., from user profiles) associated with one or more diabetic patients (e.g., users or non-users of the continuous analyte monitoring system 104 and / or application 106). The training data may be stored in the medical history record database 112 and may be accessible to the training server system 140 via one or more networks (not shown) for training the machine learning models.
[0047] Training data refers to a characterized and labeled dataset. For example, the dataset may include multiple data records, each containing information corresponding to a different user profile stored in the user database 110, and each data record is characterized and labeled. In machine learning and pattern recognition, a feature is an individual measurable characteristic or feature. Generally, the features that best characterize the patterns in the data are selected to create a predictive machine learning model. Data labeling is the process of adding one or more meaningful and informative labels to provide context to the data for learning by the machine learning model. As an illustrative example, each relevant feature of a user reflected in the corresponding data record may be a feature used in training the machine learning model. Such characteristics may include age, gender, health status, normal glucose range, normal lactate range (e.g., lactate baseline and / or peak), start and end times associated with the physical activity, duration of the physical activity, type and / or intensity of the physical activity (e.g., based on glucose level delta and / or lactate level delta), time and / or duration associated with the glycemic event, glucose information associated with the glycemic event (e.g., glucose levels associated with a hypoglycemic or hyperglycemic event, etc.), treatment administered, glucose information after treatment, etc.
[0048] The model is then trained by the training server system 140 using the characterized and labeled training data. In particular, the features of each data record may be used as input to the machine learning model, and the generated output may be compared to the label associated with the corresponding data record. The model may calculate a loss based on the difference between the generated output and the provided label. This loss is then used to modify the internal parameters or weights of the model. By iteratively processing each data record corresponding to each medical history patient, the model may be iteratively refined to generate accurate predictions of glycemic events for patients engaged in physical activity.
[0049] As illustrated in FIG. 1, the training server system 140 deploys these trained models to the decision support engine 114 for use during runtime. For example, the decision support engine 114 may retrieve a user profile 118 associated with the user and stored in the user database 110, use information in the user profile 118 as input to the trained models, and output a prediction indicative of the user's current or future physical activity-induced glycemic events (e.g., shown as output 144 in FIG. 1). The output 144 generated by the decision support engine 114 may also provide one or more recommendations for treatment based on the prediction. The output 144 may be provided to the user (e.g., via the application 106), to the user's caregiver (e.g., a parent, relative, guardian, teacher, physical therapist, fitness trainer, nurse, etc.), to the user's physician or health care provider, or in some cases, to any other person interested in the user's health for the purpose of improving the user's health, such as by implementing the recommended treatment. The output 144 generated by the decision support engine 114, along with the actual glycemic events and / or treatments taken by the user, are stored in the user database 110 and utilized to train or retrain the trained model. In certain embodiments, the output 144 generated by the decision support engine 114, which may be indicative of the user's physiological state and / or current treatments recommended for the user, may be stored in the user profile 118.
[0050] 2 is a diagram 200 that conceptually illustrates an example continuous analyte monitoring system 104 including an example continuous analyte sensor with sensor electronics according to certain aspects of the present disclosure. For example, the system 104 can be configured to continuously monitor one or more analytes of a user according to certain aspects of the present disclosure.
[0051] The continuous analyte monitoring system 104 in the illustrated embodiment includes a sensor electronics module 204 and one or more continuous analyte sensors 202 (individually referred to herein as a continuous analyte sensor 202 and collectively referred to herein as continuous analyte sensors 202) associated with the sensor electronics module 204. The sensor electronics module 204 may wirelessly communicate (e.g., directly or indirectly) with one or more of the display devices 210, 220, 230, and 240. In certain embodiments, the sensor electronics module 204 may also wirelessly communicate (e.g., directly or indirectly) with one or more medical devices, such as a medical device 208 (individually referred to herein as a medical device 208 and collectively referred to herein as medical devices 208), and / or one or more other non-analyte sensors 206 (individually referred to herein as a non-analyte sensor 206 and collectively referred to herein as non-analyte sensors 206).
[0052] In certain embodiments, the continuous analyte sensor 202 may comprise one or more sensors for detecting and / or measuring an analyte. The continuous analyte sensor 202 may be a multi-analyte sensor configured to continuously measure two or more analytes, such as at least glucose and lactose, or a single-analyte sensor configured to continuously measure a single analyte as a non-invasive, subcutaneous, transcutaneous, transdermal, and / or intravascular device. In certain embodiments, the continuous analyte sensor 202 may be configured to continuously measure an analyte level of a user using one or more techniques, such as enzymatic, chemical, physical, electrochemical, spectrophotometric, polarimetric, colorimetric, iontophoretic, radiometric, immunochemical, etc. The term "continuous" as used herein may mean fully continuous, semi-continuous, periodic, etc. In certain aspects, the continuous analyte sensor 202 provides a data stream indicative of the concentration of one or more analytes in a user. The data stream may include a raw data signal that is then converted into a calibrated and / or filtered data stream that is used to provide an estimated analyte value to a user.
[0053] In certain embodiments, the continuous analyte sensor 202 may be a multi-analyte sensor configured to continuously measure multiple analytes in the user's body. For example, in certain embodiments, the continuous multi-analyte sensor 202 may continuously measure glucose, lactate, ketones, glycerol, potassium (e.g., hyperkalemia), sodium, and / or CO2 in the user's body. 2 Or, there may be a single sensor configured to measure the anion gap.
[0054] In certain embodiments, the sensor electronics module 204 includes electronic circuitry associated with measuring and processing the continuous analyte sensor data, including predictive algorithms associated with processing and calibrating the sensor data. The sensor electronics module 204 can be physically connected to the continuous analyte sensor 202 and can be integral (non-releasably attached) or releasably attached to the continuous analyte sensor 202. The sensor electronics module 204 can include hardware, firmware, and / or software that enables measurement of the level of the analyte via the continuous analyte sensor 202. For example, the sensor electronics module 204 can include a potentiostat, a power supply for providing power to the sensor, other components useful for signal processing and data storage, and a telemetry module for transmitting data from the sensor electronics module, for example, to one or more display devices. The electronics can be affixed to a printed circuit board (PCB) or the like and can take a variety of forms. For example, the electronics can take the form of an integrated circuit (IC), such as an Application-Specific Integrated Circuit (ASIC), a microcontroller, and / or a processor.
[0055] Display devices 210, 220, 230, and / or 240 are configured to display displayable sensor data, including analyte data, that may be transmitted by sensor electronics module 204. Each of display devices 210, 220, 230, or 240 may include a display, such as a touchscreen display 212, 222, 232, and / or 242, for displaying sensor data to a user and / or receiving input from a user. For example, a graphical user interface (GUI) may be presented to the user for such purposes. In some embodiments, the display device may include other types of user interfaces, such as a voice user interface, instead of or in addition to a touchscreen display, for communicating sensor data to a user of the display device and / or receiving user input. Display devices 210, 220, 230, and 240 may be examples of display device 107 illustrated in FIG. 1 used to display sensor data to a user of the system of FIG. 1 and / or receive input from a user.
[0056] In some embodiments, one, some, or all of the display devices are configured to display or otherwise communicate the sensor data as it is communicated from the sensor electronics module (e.g., in a data package sent to the display device based on their respective preferences) without any additional possible processing required for calibration and / or real-time display of the sensor data.
[0057] The display devices may include custom display devices designed specifically to display a particular type of displayable sensor data associated with the analyte data received from the sensor electronics module. Display device 210 is an example of such a custom device. In some embodiments, one of the display devices is a smart phone, such as display device 220 representing a mobile phone, configured to display a graphical representation of the continuous sensor data (e.g., including current and historical data) using a commercially available operating system (OS). Other display devices may include display device 230 representing a tablet, display device 240 representing a smart watch or fitness tracker, medical device 208 (e.g., an insulin delivery device or blood glucose meter), and / or other handheld devices, such as a desktop or laptop computer (not shown).
[0058] Because different display devices have different user interfaces, the contents of the data package (e.g., amount, format, and / or type of data displayed, alarms, etc.) can be customized (e.g., programmed differently by the manufacturer and / or end user) for each particular display device. Thus, in certain implementations, multiple different display devices can wirelessly communicate directly with the sensor electronics module (e.g., the on-skin sensor electronics module 204 physically connected to the continuous analyte sensor 202) during a sensor session to enable multiple different types and / or levels of display and / or functionality associated with the displayable sensor information.
[0059] As mentioned, the sensor electronics module 204 may be in communication with a medical device 208. The medical device 208 may be a passive device in some exemplary embodiments of the present disclosure. For example, the medical device 208 may be an insulin pump for administering insulin to a user. For various reasons, it may be desirable for such an insulin pump to receive and track glucose values transmitted from the continuous analyte monitoring system 104, and the continuous analyte sensor 202 comprises at least a glucose sensor and a lactate sensor.
[0060] Additionally, as mentioned, the sensor electronics module 204 may also communicate with other non-analyte sensors 206. The non-analyte sensors 206 may include, but are not limited to, altimeter sensors, accelerometer sensors, global positioning system (GPS) sensors, temperature sensors, respiration sensors, electromyogram (EMG) sensors, galvanic skin response (GSR) sensors, impedance sensors, and the like. The non-analyte sensors 206 may also include monitors such as heart rate monitors, blood pressure monitors, pulse oximeters, caloric intake monitors, and medication delivery devices. One or more of these non-analyte sensors 206 may provide data to the decision support engine 114, which is described further below. In some aspects, a user may manually provide a portion of the data for processing by the training server system 140 and / or the decision support engine 114 of FIG. 1.
[0061] In certain embodiments, the non-analyte sensors 206 may further include sensors for measuring skin temperature, core body temperature, sweat rate, and / or sweat composition.
[0062] In certain embodiments, the non-analyte sensor 206 may be combined in any other configuration, such as, for example, combined with one or more continuous analyte sensors 202. As an illustrative example, a non-analyte sensor, such as a temperature sensor, may be combined with a continuous glucose sensor 202 to form a glucose / temperature sensor that is used to transmit sensor data to the sensor electronics module 204 using a common communications circuit. As another illustrative example, a non-analyte sensor, such as a temperature sensor, may be combined with a multi-analyte sensor 202 configured to measure glucose and lactate to form a glucose / lactate / temperature sensor that is used to transmit sensor data to the sensor electronics module 204 using a common communications circuit.
[0063] One or more of the continuous analyte monitoring system 104, the multiple display devices, the medical device 208, and / or the non-analyte sensors 206 may be configured to communicate together wirelessly using one of a variety of wireless communication technologies (e.g., Wi-Fi, Bluetooth, Near Field Communication (NFC), cellular, etc.). In certain embodiments, a wireless access point (WAP) may be used to couple one or more of the continuous analyte monitoring system 104, the multiple display devices, the medical device 208, and / or the non-analyte sensors 206 to one another. For example, the WAP may provide Wi-Fi, Bluetooth, and / or cellular connectivity between these devices. NFC may also be used between the devices depicted in the diagram 200 of FIG. 2.
[0064] Figure 3 illustrates example inputs and example metrics calculated based on the inputs for use by the decision support system of Figure 1, according to some embodiments disclosed herein. In particular, Figure 3 provides a more detailed illustration of the example inputs and example metrics introduced in Figure 1.
[0065] 3 illustrates exemplary inputs 128 on the left, applications 106 and DAM 116 in the center, and metrics 130 on the right. In certain embodiments, each of the metrics 130 may correspond to one or more values, such as discrete numeric values, ranges, or qualitative values (such as high / medium / low, or stable / unstable). The applications 106 obtain the inputs 128 through one or more channels (e.g., manual user input, sensors, other applications running on the display device 107, an EMR system, etc.). As mentioned above, in certain embodiments, the inputs 128 may be processed by the DAM 116 to output multiple metrics, such as the metrics 130. The inputs 128 and metrics 130 may be used by the training server system 140 and the decision support engine 114 to both train and deploy one or more machine learning models for predicting glycemic events for users engaged in physical activity.
[0066] In certain embodiments, beginning with input 128, physical activity information may include any information surrounding an activity, such as an activity that requires physical exertion by a user. For example, physical activity information may include information regarding physical activities ranging from relatively low intensity physical exertion (e.g., walking, standing, passive stretching, etc.) to relatively high intensity physical exertion (e.g., sprinting, weightlifting, action sports), including aerobic and anaerobic exercise. Such information may be based on continuous analyte sensor data measured by continuous analyte sensor 202, non-analyte sensor data measured by non-analyte sensor 206 (e.g., non-analyte sensor data input from an accelerometer, heart rate monitor, respiration rate sensor, etc.), user statistics stored in user profile 118, etc. For example, in certain embodiments, the physical activity information may be based on glucose data, lactate data, accelerometer data, step data, athletic equipment power meter data, GPS data, heart rate data (e.g., heart rate reserve and heart rate variability (HRV)), electrocardiogram (EKG) data, EMG data, respiration rate data, temperature data, blood pressure data, galvanic skin response data, oxygen intake data, sleep data, impedance data, etc. associated with the user and may be provided by continuous analyte or non-analyte sensors, as described above. In certain embodiments, the physical activity information may be provided by an accelerometer sensor on a wearable device, such as, for example, a watch, a fitness tracker, and / or a patch. In certain embodiments, the physical activity information may also be provided through manual user input.
[0067] In certain embodiments, food consumption is also provided as an input. Food consumption information may include information about one or more of meals, snacks, and / or beverages, such as one or more of amount, content (carbohydrates, fat, protein, etc.), order of consumption, and time of consumption. In certain embodiments, food consumption may be provided by the user through manual input, by providing a photo through an application configured to recognize food type and amount, and / or by scanning a barcode or menu. In various examples, meal amount may be manually entered as one or more of calories, amount ("3 cookies"), menu item ("Royale with Cheese"), and / or food exchange (1 fruit, 1 dairy). In some examples, meal information may be received through a convenient user interface provided by application 106. In further examples, meal amount may be passively determined using continuous analyte sensor data. For example, lactate levels may indicate the amount of a meal (e.g., a larger meal produces a larger spike in lactate levels). Such lactate response data may differ from elevated lactate levels caused by physical activity because lactate spikes caused by food consumption precede glucose changes (e.g., glucose spikes) and, unlike exercise, may be slower to return to baseline.
[0068] In certain embodiments, user statistics such as one or more of age, height, weight, BMI, body composition (e.g., body fat percentage), height, build, or other information may also be provided as input. Other examples of user statistics may include historical athletic data such as race (or other athletic event) results, normal training pace, and historical biomarker responses to physical activity (e.g., heart rate or typical average glucose for running at a given pace or riding at a given power level). In certain embodiments, user statistics may be provided through a user interface, by interfacing with an electronic source such as an electronic medical record, and / or from a measurement device. In certain embodiments, the measurement device may include one or more of a wireless, e.g., Bluetooth-enabled, weight scale, and / or a camera, which may communicate with the display device 107 to provide user data, for example.
[0069] In certain embodiments, treatment / medication information is also provided as an input. The medication information may include information regarding the type, dosage, and / or timing of when one or more medications have been or will be taken by the user. The treatment information may include information regarding different lifestyle habits recommended by the user's physician. For example, the user's physician may recommend that the user drink alcohol in moderation, exercise at least 30 minutes per day, or cut calories by 500-1,000 calories per day to improve general health. In certain embodiments, the treatment / medication information may be provided through manual user input.
[0070] In certain embodiments, analyte sensor data may also be provided as an input, for example, through the continuous analyte monitoring system 104. In certain embodiments, the analyte sensor data may include glucose data (e.g., a user's time-stamped glucose values, trends, patterns, etc.) measured by at least a glucose sensor or multi-analyte sensor in the continuous analyte monitoring system 104. In certain embodiments, the analyte sensor data may include lactate data (e.g., a user's time-stamped lactate values, trends, patterns, etc.) measured by at least one lactate sensor or multi-analyte sensor in the continuous analyte monitoring system 104. In certain embodiments, the analyte sensor data may include ketone data (e.g., a user's time-stamped ketone values, trends, patterns, etc.) measured by at least one ketone sensor or multi-analyte sensor in the continuous analyte monitoring system 104. In certain embodiments, the analyte sensor data may include data extracted from intermittent blood samples by the continuous analyte monitoring system 104.
[0071] In certain embodiments, inputs may also be received from non-analyte sensors, such as the non-analyte sensor 206 described with respect to Figure 2. Inputs from such non-analyte sensors 206 may include information related to a user's heart rate, respiration rate, oxygen saturation, or temperature (e.g., to detect illness, physical activity, etc.), as well as trends and / or patterns thereof. In certain embodiments, electromagnetic sensors may also detect low power radio frequency (RF) fields emitted from an object or a tool in contact with or in proximity to an object, which may provide information about user activity or location.
[0072] In certain embodiments, the input received from the non-analyte sensor may include input regarding the user's insulin delivery. In particular, the input regarding the user's insulin delivery may be received via a wireless connection on the smart pen, via a user input, and / or from an insulin pump. The insulin delivery information may include one or more of insulin amount, delivery time, and the like. Other configurations, such as insulin action time or duration of insulin action, may also be received as input.
[0073] In certain embodiments, a time may also be provided as an input, such as a time of day or a time from a real-time clock. In certain embodiments, one or more of the inputs 128, as well as one or more of the metrics 130, may be time-stamped. For example, in certain embodiments, the input analyte data may be time-stamped to indicate the date and time when the analyte measurement was made for the user.
[0074] User input of any of the above-mentioned inputs 128 may be provided through a user interface, such as the user interface of the display device 107 of Figure 1. As described above, in certain embodiments, the DAM 113 determines or calculates metrics 130 for the user based on the inputs 128. An example list of metrics 130 is shown in Figure 3.
[0075] In certain embodiments, the glucose level may be determined from sensor data (e.g., blood glucose measurements obtained from a continuous glucose sensor of the continuous analyte monitoring system 104). For example, the glucose level may refer to time-stamped glucose measurements or values that are continuously generated and stored over time. In some examples, the glucose level may also be determined based on historical information in a particular situation, for example, for a given combination of food consumption, insulin, and / or physical activity.
[0076] In certain embodiments, glucose trends may be determined based on glucose levels over a particular period of time.
[0077] In certain embodiments, a normal glucose range may be determined from sensor data (e.g., blood glucose measurements obtained from the continuous analyte monitoring system 104). The normal glucose range may indicate a range within which a user does not experience any symptoms (e.g., symptoms associated with hyperglycemia or hypoglycemia) under controlled conditions, such as when the user is fasting, during a given period after a meal, etc. The range may have a minimum and / or maximum value. For example, in certain embodiments, the range may have a minimum normal glucose concentration value. In certain embodiments, the range may have a maximum normal glucose concentration value. A user's normal glucose range may be expected to remain constant or to change gradually over time. Additionally, each user may have a different normal glucose range. In certain embodiments, the normal glucose range may be determined by determining a minimum and / or maximum average value of glucose over a particular amount of time where no variation is expected. In certain embodiments, the normal glucose range for a user may be determined when the user is asleep, sitting in a chair, or over other periods during which the user is sedentary. In certain embodiments, the DAM 116 may continuously or periodically calculate the normal glucose range and a timestamp and store the corresponding information in the user profile 118.
[0078] In yet other embodiments, normal glucose ranges may be determined from population data (e.g., from past data records or samples of diabetic patients). In such embodiments, each user may have personalized, i.e., customized, acceptable minimum and / or maximum glucose values, which may be determined by methods similar to those described above (e.g., determining the ranges under controlled conditions, such as when the user is fasting, for a given period after a meal, and while the user is not experiencing any symptoms).
[0079] In certain embodiments, lactate levels may be determined from sensor data (e.g., lactate measurements obtained from the continuous analyte monitoring system 104). For example, lactate levels may refer to time-stamped lactate measurements or values that are continuously generated and stored over time. In some examples, lactate levels may also be determined based on historical information in a particular situation, e.g., for a given combination of food consumption and / or physical activity.
[0080] In certain embodiments, lactate trend may be determined based on lactate levels over a particular period of time, hi certain embodiments, lactate trend may be determined based on lactate production rate and / or calculated lactate clearance rate over a particular period of time.
[0081] In certain embodiments, the lactate baseline may be determined from sensor data (e.g., lactate measurements obtained from the continuous analyte monitoring system 104). The lactate baseline may indicate a user's lactate values during periods when variation in lactate production / clearance is typically not expected. A user's lactate baseline is typically expected to remain constant or to change gradually over time. Additionally, each user may have a different lactate baseline. In certain embodiments, the lactate baseline may be determined by determining an average of lactate measurements over a particular amount of time when variation is not expected. In certain embodiments, the lactate baseline for a user may be determined while the user is asleep, sitting in a chair, or over other periods when the user is sedentary. In certain embodiments, the DAM 116 may continuously or periodically calculate the lactate baseline and timestamp and store the corresponding information in the user's profile 118.
[0082] In certain embodiments, a "lactate threshold" may be determined from sensor data (e.g., lactate measurements obtained from the continuous analyte monitoring system 104). The lactate threshold may indicate a user's lactate value at which lactate production exceeds lactate clearance. This may be caused by the user engaging in high intensity anaerobic activity. Each user may have a different lactate threshold. In certain embodiments, the lactate threshold may be determined by determining the lowest lactate value for a certain amount of time during which lactate levels increase exponentially (i.e., rapidly). In certain embodiments, the lactate threshold may be determined by determining the highest lactate value before the user's increasing work rate leads to exponentially increasing lactate levels. In certain embodiments, the lactate threshold for a user may be determined over a period of time when engaged in physical activity, such as moderate to high intensity physical activity. In certain embodiments, the DAM 116 may continuously or periodically calculate the lactate threshold and a timestamp and store the corresponding information in the user's profile 118.
[0083] In certain embodiments, insulin sensitivity may be determined using historical data, real-time data, or a combination thereof, and may be based on one or more inputs 128, such as one or more of food consumption information, continuous analyte sensor data, non-analyte sensor data (e.g., insulin delivery information from an insulin device), and the like. Insulin sensitivity refers to how a user's cells respond to insulin, and such information may provide important actionable insights about the user. For example, improving insulin sensitivity for a user may help reduce insulin resistance in the user. In certain embodiments, insulin sensitivity may be utilized to adjust the user's recommended dose of insulin.
[0084] In certain embodiments, insulin sensitivity may be determined based on an active insulin sensitivity assessment. For example, a user may be requested, e.g., by the decision support system 100, to consume a known amount of a substance expected to cause a glucose spike and a known amount of insulin, and then the user's glucose levels and / or trends may be monitored over a predetermined period of time to determine the patient's insulin sensitivity. Such an active insulin sensitivity assessment may be performed at baseline (e.g., at rest), before engaging in physical activity, or after physical activity, using a limited dose of insulin. In some examples, insulin sensitivity may be additionally determined or adjusted based on other considerations, such as adjustments for other analytes or other substances that may be expected to cause an insulin response or affect insulin clearance rate.
[0085] In certain embodiments, insulin sensitivity may be calculated using inputs over an extended period of time and may reflect a long-term average health state. In other embodiments, insulin sensitivity may be estimated frequently using real-time data, historical data, or a combination thereof, and changes in insulin sensitivity may be used to estimate changes in metabolic health. In certain embodiments, insulin resistance may be estimated by monitoring analyte data (e.g., insulin, glucose, lactate, glycerol, ketones, potassium, etc.) in real time, particularly in response to known exercise or food challenges.
[0086] In certain cases, the user may experience hyperlactemia as a result of physical activity, and lactate levels may be slow to return to baseline. In such cases, the user may experience hyperglycemia and acute insulin resistance due to hyperlactemia. In response to normalization of the user's lactate levels, the patient's insulin sensitivity may return to the user's previous insulin sensitivity level. Thus, monitoring lactate levels and / or trends may inform insulin sensitivity measurements. In addition to responding to physical activity and short-term dietary changes, insulin sensitivity also appears to precede metabolic syndrome, type 2 diabetes, and heart disease.
[0087] In certain embodiments, remaining insulin (insulin on board) may be determined using non-analyte sensor data inputs (e.g., insulin delivery information) and / or known or learned (e.g., from user data) insulin time-action profiles that may take into account both basal metabolic rate (e.g., insulin updates to keep the body moving) and insulin usage driven by activity or food consumption.
[0088] In certain embodiments, ketone levels may be determined from sensor data (e.g., ketone measurements obtained from the continuous analyte monitoring system 104). In certain embodiments, ketone levels may be determined from surrogate sensor data, e.g., changes in lactate and / or glucose levels or trends, and / or other non-analyte sensor data, such as changes in heart rate, blood pressure, and / or other non-analyte metrics. In certain embodiments, ketone levels may be expressed as a metric of whether a user is in ketosis. In particular, ketosis is a metabolic state in which high concentrations of ketones are present in the user's blood. Ketosis can be an indicator of low glycogen stores. In diabetic patients, elevated blood ketones indicate relative or absolute insulin deficiency.
[0089] In certain embodiments, the ketone production rate may be determined from sensor data (e.g., ketone measurements obtained from the continuous analyte monitoring system 104). In particular, ketones (chemically known as ketone bodies) are a by-product of the breakdown of fatty acids. Glucose (e.g., blood glucose) is the preferred fuel source for many cells in the body; however, when cells have limited access to glucose, fat may be broken down for fuel, thereby producing ketones as a by-product. In certain embodiments, the ketone production rate may be determined by evaluating the increase in ketone levels over a specified amount of time.
[0090] In certain embodiments, the health and illness metrics may be determined, for example, from a physiological sensor (e.g., temperature), an activity sensor, or a combination thereof, based on one or more of user inputs (e.g., pregnancy information or known illness information). In certain embodiments, based on the values of the health and illness metrics, for example, the user's state may be defined as one or more of healthy, sick, rest, or fatigued.
[0091] In certain embodiments, the meal state metric may indicate the user's state with respect to food consumption. For example, the meal state may indicate whether the user is in one of a fasting state, a pre-meal state, a fed state, a post-meal reaction state, or a stable state. In certain embodiments, the meal state may also indicate remaining nutrients, e.g., meals, snacks, or beverages consumed, which may be determined from food consumption information, mealtime information, and / or digestibility information, which may be correlated, for example, to food type, amount, and / or order (e.g., which food / beverage was eaten first).
[0092] In certain embodiments, the eating habits metric is based on the content and timing of the user's meals. For example, if the eating habits metric is on a scale of 0 to 1, in an example, the better / healthier the diet the user eats, the higher the user's eating habits metric will be relative to 1. A better / healthier diet may be defined as one that does not cause the user's glucose levels to deviate from the user's normal glucose range (e.g., 70-180 mg / dL or the user's desired range). Also, in this example, the more the user's food consumption adheres to a specific time schedule, the closer their eating habits metric will be to 1. In certain embodiments, the eating habits metric may indicate whether the user is consistently engaging in a ketogenic diet (e.g., a low carbohydrate, medium protein, high fat diet) based on the meals, snacks, or beverages consumed by the user over a period of time. In another example, the eating habits metric may reflect the content of the patient's meals, e.g., three numbers may indicate the percentage of carbohydrate, protein, and fat.
[0093] In certain embodiments, medication adherence is measured by one or more metrics that indicate how committed a user is to their medication regimen. In certain embodiments, the medication adherence metric is calculated based on one or more of the timing of when the user takes the medication (e.g., whether the user is on time or on schedule), the type of medication (e.g., whether the user is taking the correct type of medication), and the dosage of the medication (e.g., whether the user is taking the correct dosage).
[0094] In certain embodiments, the fitness metric may indicate a level of fitness of a user. In certain embodiments, the fitness metric may be determined based on one or more inputs 128, such as, for example, one or more of physical activity information, food consumption, user statistics (such as height and weight), continuous analyte sensor data (e.g., lactate threshold), non-analyte sensor data, etc. In certain embodiments, the fitness metric is determined based on inputs from activity sensors or other physiological sensors, and the type, intensity, and / or frequency of physical activity.
[0095] In certain embodiments, the activity intensity level metric may indicate the intensity level at which the user is performing the activity. For example, the activity intensity level metric may include information indicating that the user is engaged in low intensity physical activity, low-moderate intensity physical activity, moderate intensity physical activity, medium-moderate-high intensity physical activity, and / or high intensity physical activity, all of which may affect the user's glucose level. In certain embodiments, the activity intensity level metric may be calculated by the DAM 116 based on one or more of the inputs 128, such as one or more of physical activity information, non-analyte sensor data, time, user statistics, and the like. For example, in certain embodiments, the activity intensity level metric may be determined based on physical activity information, such as input from an activity sensor on a fitness tracker or other physiological sensor. In certain embodiments, the activity intensity level metric may be determined based on input from other non-analyte sensors, such as an accelerometer, an athletic equipment sensor (e.g., a power meter), a GPS device, a heart rate monitor, an EKG device, an EMG device, a respiratory monitor, a temperature monitor, a blood pressure monitor, a pulse oximeter, and the like. In certain embodiments, the activity intensity level metric may be determined based on skin temperature, core body temperature, sweat rate, and / or sweat composition. In certain embodiments, the activity intensity level metric may be determined based on user statistics, such as information stored in user profile 118 or information provided through manual user input. In further embodiments, the activity level metric may be based on continuous analyte sensor data measured by continuous analyte sensor 202, such as lactate (e.g., lactate threshold, lactate production / clearance rate, etc.) and / or glucose levels.
[0096] In certain embodiments, metabolic rate is a metric that may be indicative of or include basal metabolic rate (e.g., energy expended at rest) and / or active metabolism (e.g., energy expended by physical activity such as physical exertion). In some examples, basal metabolic rate and active metabolism may be tracked as separate outcome metrics. In certain embodiments, metabolic rate may be calculated by DAM 116 based on one or more of inputs 128, such as one or more of physical activity information, continuous analyte sensor data, non-analyte sensor data, time, etc.
[0097] In certain embodiments, a body temperature metric may be calculated by the DAM 116 based on the input 128, more specifically, non-analyte sensor data from a temperature sensor. In certain embodiments, a heart rate metric may be calculated by the DAM 116 based on the input 128, more specifically, non-analyte sensor data from a heart rate sensor. In certain embodiments, a respiration metric may be calculated by the DAM 116 based on the input 128, more specifically, non-analyte sensor data from a respiration rate sensor.
[0098] Exemplary methods and systems for predicting glycemic events in patients engaged in physical activity using continuously monitored analyte data 4A and 4B illustrate a flow diagram of an example method 400 for providing decision support using one or more continuous analyte sensors configured to measure or otherwise indicate at least a user's glucose and lactate levels, according to some aspects of the disclosure. For example, the method 400 may be performed to provide decision support to a user using the continuous analyte monitoring system 104, as illustrated in FIGS. 1 and 2. The method 400 may be performed by the decision support system 100, for example, to collect data including the analyte data, patient information, and non-analyte sensor data referred to above to (1) detect (e.g., automatically) a patient engaged in physical activity, (2) determine relevant parameters of the physical activity engaged by the patient (e.g., intensity, duration, and / or type of physical activity), (3) predict current or future physical activity-induced glycemic events based in part on the relevant parameters associated with the physical activity engaged by the patient, and (4) make patient-specific treatment decisions or recommendations for diabetes. In other words, the decision support system presented herein may provide information useful in directing and improving care for a diabetic patient during and / or after engaging in physical activity. Method 400 is described below with reference to Figures 1 and 2 and their components.
[0099] At block 402, the method 400 begins by continuously monitoring multiple analytes of a patient, such as the user 102 illustrated in FIG. 1, during a period of acquiring analyte data, the multiple analytes including at least glucose and lactate. Block 402 may be performed in certain embodiments by the continuous analyte monitoring system 104 illustrated in FIGS. 1 and 2, and more specifically, the continuous analyte sensor 202 illustrated in FIG. 2. For example, the continuous analyte monitoring system 104 may include a continuous glucose sensor 202 for measuring the user's glucose level and a continuous lactate sensor 202 for measuring the user's lactate level. Alternatively, the continuous analyte monitoring system 104 may include a continuous multi-analyte sensor to measure both the user's lactate as well as glucose levels. As previously discussed, in certain circumstances (e.g., periods of anaerobic cellular respiration and high metabolism during moderate to high intensity physical activity), the presence of lactate may be a better indication of the type and intensity of physical activity engaged in by the user than other analytes, such as glucose levels alone. Therefore, in addition to glucose, continuous monitoring of lactate may be required to assess physical activity engaged in by a user in order to accurately predict glycemic events resulting from a user engaging in such physical activity.
[0100] In certain embodiments, continuous monitoring of lactate may be utilized to optimize the performance of a continuous analyte monitoring system, such as continuous analyte monitoring system 104, by enabling more rapid prediction of hyperglycemic events via adjustment of glucose sampling rates. A high rate of change in lactate levels may, at least in certain cases, indicate or predict a high rate of change in glucose levels. For example, a high rate of change in lactate levels may indicate the performance of physical activity or a change in intensity level of physical activity, which may cause a sudden or high rate of change in glucose levels, as described above, and thus signal a prediction of a glycemic event. A high rate of change in lactate levels may be defined in part as a minimum, e.g., a threshold, positive or negative delta in lactate levels. Thus, lactate levels may be utilized to optimize the sampling of glucose during periods of physical activity, for example.
[0101] In certain examples, when the continuous analyte monitoring system 104 determines an abrupt difference in lactate levels, the continuous analyte monitoring system 104 may utilize or adjust a predicted interstitial lag time for interstitial analyte measurements, including glucose and lactate measurements. Generally, changes in interstitial analyte levels are slightly delayed or delayed relative to changes in blood analyte levels due to the time required for the analyte to diffuse from the capillaries to the surrounding tissue. This interstitial lag time may change in length as the volume of distribution shifts when the user engages in physical activity. Thus, detection of an abrupt difference in lactate levels, which may indicate the user's engagement in physical activity and thus a shift in the volume of distribution, may further indicate a change in the interstitial lag time for the analyte measurements. Thus, upon determining such an abrupt change in lactate levels, the continuous analyte monitoring system 104 may adjust a previously predicted interstitial lag time or utilize / determine a new predicted interstitial lag time, which is then incorporated into the continuous measurements and / or predictions of the user's analyte levels. In certain embodiments, the predicted interstitial lag time may be continuously incorporated into the user's sequential analyte measurements, hi certain embodiments, the predicted interstitial lag time may be incorporated into the user's sequential analyte measurements only when an abrupt difference in lactate levels is determined and / or when elevated lactate levels are determined.
[0102] In certain embodiments, the lactate level abruptness utilized to adjust the predicted interstitial lag time is based on the real-time slope of the user's lactate level and / or the rate of change of the user's lactate level. In such embodiments, if the real-time slope and / or rate of change meets or exceeds a predetermined threshold, the continuous analyte monitoring system 104 may then adjust the predicted interstitial lag time for the interstitial analyte measurement.
[0103] In certain examples, if the continuous analyte monitoring system 104 determines an abrupt difference in lactate levels (thus indicating a potentially impending abrupt change in glucose levels), the continuous analyte monitoring system 104 may further increase the sampling rate of the continuous glucose sensor 202 to facilitate a more accurate glucose measurement. By increasing the sampling rate, the number of data points for analysis is increased, thereby facilitating a measurement that is more accurately indicative of the patient's real-time glucose level. For example, if the baseline sampling rate is one sample per minute, if the continuous analyte monitoring system 104 detects a high rate of change in lactate levels, the sampling rate of the continuous glucose sensor 202 may be increased to one sample per 30 seconds, or one sample per 20 seconds, or one sample per 10 seconds, or one sample per 5 seconds, or one sample per 3 seconds, etc.
[0104] In certain examples, when the continuous analyte monitoring system 104 determines an abrupt difference in lactate levels (thus indicating a potential abrupt change in glucose levels), the continuous analyte monitoring system 104 may further increase the data transmission rate from the continuous glucose sensor 202 to a display device, such as the display device 107, to provide the user with a more advanced warning of a current or future glycemic event. Sending sensor readings to the display device 107 more frequently facilitates earlier analysis of measurements and thus earlier prediction of a glycemic event, thus allowing earlier provision of warnings and / or recommendations to the user. For example, if the baseline data transmission rate is one transmission per five minutes, when the continuous analyte monitoring system 104 detects a high rate of change in lactate levels, the data transmission rate of the continuous glucose sensor 202 may be increased to one transmission per three minutes, or one transmission per minute, or one transmission per 30 seconds, etc.
[0105] While the primary analytes for measurements described herein are glucose and lactate, in certain embodiments, other analytes may be considered. In particular, combining glucose and lactate measurements with additional analyte data may help to further inform the analysis regarding prediction of glycemic events due to physical activity. For example, monitoring additional types of analytes, such as ketones measured by the continuous analyte monitoring system 104, may provide additional insight into the type or intensity of physical activity and / or supplemental information used to determine optimal treatments for preventing or mitigating physical activity-induced glycemic events.
[0106] The additional insight gained from using a combination of analytes rather than just glucose and lactate may increase the accuracy of the glycemic event prediction. For example, the probability of accurately predicting a glycemic event may be a function of the number of analytes measured for a user. More specifically, in certain examples, the probability of accurately predicting a glycemic event using only glucose and lactate (in addition to other non-analyte data) may be lower than the probability of accurately predicting a glycemic event using glucose, lactate, and another analyte (in addition to other non-analyte data).
[0107] In certain embodiments described herein, for example, the combination of analytes measured and collected by one or more (e.g., multi-analyte) sensors for glycemic event prediction and diabetes decision support includes glucose, lactate, and ketones, however, other analyte combinations may be considered. For example, in certain embodiments, at block 402, the continuous analyte monitoring system 104 may continuously monitor a user's glucose, lactate, and ketone levels over a period of time. In such embodiments, the measured ketone concentrations may be used to further inform analysis for predicting glycemic events such as hyperglycemia and / or diabetic ketoacidosis (DKA).
[0108] In particular, insulin mediates the precise regulation of glucose metabolism and plasma concentration by promoting glucose uptake by skeletal muscle, liver, and adipose tissue. Thus, when insulin is low, glucose uptake by skeletal muscle, liver, and adipose tissue is restricted and glucose levels rise. Such restricted access to glucose also triggers the breakdown of fat for fuel (e.g., ketogenesis), which produces an acid buildup in the bloodstream called ketones (e.g., ketone bodies). If untreated, increased concentrations of ketones in the blood lead to DKA. Thus, ketone levels, when combined with other analyte and / or non-analyte measurements, may indicate physical activity-induced events such as hyperglycemia and DKA. In certain embodiments, ketone levels and / or the occurrence of ketosis may be utilized to determine the level of glycogen stored in the patient's cells, and thus inform dietary recommendations provided for the patient. For example, the determination of high ketone levels may indicate low glycogen levels in the patient, and thus the patient may need to consume a glucose or carbohydrate-rich meal to normalize glycogen stores.
[0109] In certain embodiments, AI models, such as machine learning models, may be used to provide predictions of physical activity-induced glycemic events and real-time or retrospective diabetes decision support. In certain embodiments, such models may be configured to use inputs from one or more sensors measuring multiple analytes to provide predictions of physical activity-induced glycemic events. Thus, taking into account such comorbidity interactions (e.g., as discussed with respect to the example for a user with DKA and / or hyperglycemia above), parameters and / or thresholds of such algorithms or models may be altered based at least in part on the number of analytes being measured for input to reflect knowledge gained from each of the other analytes being measured or prevalence associated with the additional analytes being measured.
[0110] In addition to continuously monitoring one or more analytes of a user during a period of time to obtain analyte data at block 402, optionally in certain embodiments, method 400 may also include monitoring non-analyte data during the period of time using one or more non-analyte sensors or devices at block 404. Block 404 may be performed in certain embodiments by non-analyte sensor 206 and / or medical device 208 of FIG.
[0111] As mentioned above, the non-analyte sensors and devices may include, but are not limited to, one or more of an insulin pump, a respiration sensor, a heart rate monitor, an electromyogram (EMG) sensor, an accelerometer, an altimeter sensor, a temperature sensor (e.g., a thermometer), a blood pressure monitor, a galvanic skin response (GSR) sensor, a pulse oximeter, a calorie intake monitor, a sensor or device provided by the display device 107 (e.g., an accelerometer, a camera, a global positioning system (GPS), etc.), or any other sensor or device that provides relevant information regarding the user and / or physical activity engaged by the user. In certain embodiments, the non-analyte sensors and / or devices may further include sensors for measuring skin temperature, core body temperature, sweat rate, and / or sweat composition. Such non-analyte sensors and / or devices may be worn or used by the user to aid in detecting periods of physical activity engaged by the user in addition to utilizing analyte data for multiple analytes.
[0112] Metrics, such as the metrics 130 illustrated in Figure 3, may be calculated using the measurement data from each of these non-analyte sensors. As further illustrated in Figure 3, metrics 130 calculated from the non-analyte sensor or device data may include metabolic rate, body temperature, heart rate, respiratory rate, etc. Although not shown, such metrics 130 may include heart rate reserve, heart rate variability (HRV), blood pressure, GSR, oxygen uptake (VO 2), sleep metrics, and the like. In certain embodiments, the metrics may further include skin temperature, core body temperature, sweat rate, and / or sweat composition. In certain embodiments, as described in more detail below, metrics 130 calculated from non-analyte sensor or device data may be used to further inform analyses for predicting physical activity induced glycemic events and for providing real-time or retrospective diabetes decision support.
[0113] In certain embodiments, the non-analyte sensor data may be utilized as a surrogate for the user's continuously monitored lactate levels in block 402. For example, in certain embodiments, the non-analyte sensor data, such as heart rate data (e.g., current or past), may be correlated with the user's continuously monitored lactate data (e.g., current or past). The non-analyte sensor data may then be utilized in place of continuous lactate measurements, thereby eliminating the need for a continuous lactate sensor and / or a continuous multi-analyte sensor configured to measure lactate.
[0114] In certain embodiments, non-analyte sensor data may be utilized to confirm or modify a determination related to physical activity based on the analyte data. For example, the slope of lactate levels and / or other analyte data may be monitored in combination with non-analyte data, such as, for example, GPS data and / or accelerometer data. In such an example, if the slope of lactate levels increases rapidly but the GPS data and / or accelerometer data suggests more moderate physical activity, since the rapid slope of lactate may be caused, for example, by a lack of a warm-up period, the prediction of physical activity may be adjusted to reflect the more moderate physical activity. On the other hand, if the non-analyte data suggests that the user is engaged in more intense physical activity than the lactate slope suggests, the prediction may be adjusted to reflect the more intense physical activity.
[0115] At block 406, the method 400 continues by processing the analyte data from the time period to determine one or more trends for each of the multiple analytes being continuously monitored. For example, as mentioned above, the analyte measurements obtained during the time period, including at least glucose and lactate levels, are processed to determine a directional trend, which may include data such as a minimum measurement of the analyte during the time period, a maximum measurement of the analyte during the time period, a direction of change in the analyte level during the time period (e.g., rising upward or falling downward), a rate of change in the analyte level during the time period, etc. For example, the lactate trend determined at block 406 may include a lactate production rate, a lactate clearance rate, a lactate threshold, etc. over a particular time period. Block 406 may be performed by the decision support engine 114 in certain embodiments.
[0116] At block 408, the decision support engine 114 determines a physiological state of the patient, e.g., a user, during a period of time based on the analyte data for each of the multiple analytes, the trends for each of the multiple analytes, and, in certain embodiments, the non-analyte sensor or device data. The physiological state of the patient refers to the overall or physical condition of the user, including any physical stress on the patient.
[0117] In certain embodiments, determining the user's physiological state includes determining whether the user is engaged in physical activity during the period of time, as shown in block 410 of FIG. 4A. Engagement of physical activity may be determined utilizing analyte data and / or trends for each of the multiple analytes alone, or may be determined using analyte data and / or trends for each of the multiple analytes in combination with non-analyte sensor or device data. For example, a user's lactate level typically increases when the user exercises or engages in other forms of physical activity as a result of reduced availability of oxygen to the user's muscles to break down glucose for energy. Thus, in certain embodiments, the decision support engine 114 may determine engagement of physical activity by detecting elevated or irregular lactate levels, such as lactate spikes, changes in lactate levels, and / or rates of change in lactate levels above a predetermined threshold.
[0118] To rule out elevated or irregular lactate levels and / or trends caused by other events (e.g., eating, emotional stress, or sepsis), lactate levels and / or trends may be mapped against other types of data, including other analyte levels and / or trends, as well as other non-analyte data, as shown in block 412. For example, to rule out lactate changes caused by a user eating a meal, lactate levels and / or trends may be mapped against glucose levels and / or trends, the rise and / or fluctuations of which may indicate consumption of a meal by the user. In another example, to rule out lactate changes caused by emotional stress or other confounding causes, lactate levels and / or trends may be mapped against non-analyte sensor data, for example, from a physical activity sensor, such as a heart rate monitor or accelerometer, that may confirm that the user is engaged in physical activity. Thus, lactate levels and / or lactate trends may be utilized in combination with glucose levels and / or trends and other analyte and / or non-analyte sensor data to determine whether a user is engaged in physical activity during a period of time.
[0119] In certain embodiments, the lactate measurement may further be utilized as a secondary filter to distinguish between glucose data sampling noise and actual glucose level acceleration. For example, because a change in lactate level indicates a physical stress, such as performing physical activity and / or changing the intensity of the physical activity, a change (e.g., an increase) in the lactate level may confirm that the glucose spike is not sampling noise, but an actual glucose spike caused by a physical stress, such as physical activity. Thus, the utilization of lactate may help increase the accuracy of predicting glycemic events by distinguishing between sampling noise and actual changes in glucose level measurements.
[0120] In certain embodiments, the performance of physical activity is determined utilizing data from a non-analyte sensor or device. For example, physical activity may be determined based on the patient's elevated heart rate or increased movement and may be detected by a non-analyte sensor, such as non-analyte sensor 206. Other types of non-analyte sensor data that may be utilized to determine physical activity include accelerometer data, step count data, exercise equipment power meter data, GPS data, heart rate data (e.g., heart rate reserve and HRV), EKG data, EMG data, respiration rate data, temperature data, blood pressure data, galvanic skin response data, oxygen uptake data, sleep data, impedance data, etc.
[0121] If the decision support engine 114 determines that the user engaged in physical activity during the period, then in block 414, the decision support engine 114 determines the intensity (or type) of the physical activity and its duration. Determining the intensity of the physical activity and its duration increases the accuracy of glycemic event prediction, particularly prediction of physical activity-induced glycemic events, since physical activity intensity may indicate changes in insulin sensitivity. For example, physical activity intensity may inform a determination of whether a patient is in an atypical state that may affect, for example, insulin sensitivity. Furthermore, determining whether and / or the intensity of a user performing physical activity, when mapped to time, may further increase the accuracy of predicting nocturnal glycemic events. For example, performing physical activity in the evening (e.g., closer to the time the user normally goes to sleep) may result in a higher likelihood of nocturnal glycemic events, such as hypoglycemia, compared to the morning or early afternoon (e.g., closer to the time the user normally wakes up). Thus, determining when a user engages in physical activity may help inform a prediction of whether the user will experience an overnight glycemic event.
[0122] In particular embodiments, upon determining that the user has engaged in physical activity, the decision support engine 114 may determine whether the user is engaged in low-moderate intensity or moderate-high intensity physical activity.
[0123] During low-moderate physical activity, the increased cellular energy requirements are primarily supplied by fat oxidation. The rates of lactate production and lactate clearance remain fairly balanced with each other such that lactate levels remain relatively constant at or near the user's baseline concentration. Glucose levels initially rise due to the immediate release of glucose from the liver, but eventually begin to decrease as glucose is transported into cells to provide energy for prolonged physical activity. This decrease then continues steadily over the duration of the physical activity. During recovery from low-moderate physical activity, especially during prolonged periods of low-moderate physical activity (e.g., 30 minutes), glucose levels may continue to decrease as insulin activity in vivo or delivered insulin may cause a decrease in glucose production, thereby resulting in hypoglycemic events. Thus, prolonged periods of low-moderate physical activity may expose the user to increased risk of hypoglycemia without any form of intervention or treatment. Low- to moderate-intensity physical activity includes walking, light jogging, light cycling, light swimming, hiking, housework, and gardening.
[0124] During moderate to high intensity physical activity, glucose is the main source of cellular energy, and glucose levels rise as glucose production increases. In the absence of sufficient oxygen supply, increased glucose production and glycolysis leads to increased lactate production. At higher intensity physical activity, such as high intensity physical activity, the rate of lactate production may exceed the rate of lactate clearance at muscle level, thus causing an increase in lactate levels. At a certain point, lactate production may exceed lactate clearance, resulting in a more significant increase in lactate levels. The intensity of physical activity at which lactate production exceeds lactate clearance and a significant increase in lactate levels occurs is referred to as the "lactate threshold", which varies between users depending on the user's fitness level, exercise style and / or other factors. Generally, a higher lactate threshold corresponds to greater fitness and endurance.
[0125] High intensity physical activity is associated with better glucose stability than low-moderate intensity physical activity. During recovery from high intensity physical activity, a user may experience high glucose levels, e.g., hyperglycemia, which may be referred to as "glycemic rebound." Such glycemic rebound may occur as the user's elevated lactate levels normalize due to the user's body consuming excess lactate as fuel. Then, while lactate is utilized as the body's primary energy source, glucose levels may increase, eventually leading to a period of hyperglycemia. This glycemic rebound typically lasts for about an hour. Glucose levels then equilibrate, e.g., reducing the chance of subsequent hypoglycemia compared to low-moderate intensity physical activity. In certain cases, engaging in high intensity physical activity prior to low-moderate intensity physical activity may reduce the glycemic lowering effect of low-moderate intensity physical activity, especially for insulin-resistant diabetic patients. Examples of high-intensity physical activity include sprinting, jumping rope, vigorous cycling, vigorous swimming, circuit training (HIIT), resistance training (weightlifting), etc.
[0126] In certain embodiments, the decision support engine 114 may determine the intensity of physical activity based solely on lactate levels and / or trends. For example, upon determining that the user's lactate level remains relatively constant at or near the baseline concentration over a period of time, the decision support engine 114 may determine that the user is engaged in low-moderate intensity physical activity. In another example, upon determining that the user's lactate level has increased exponentially and / or exceeded the user's lactate threshold, the decision support engine 114 may determine that the user is engaged in moderate-high or high intensity physical activity.
[0127] In certain embodiments, the decision support engine 114 may determine the intensity of the physical activity based on the lactate level and / or trend, in addition to the analyte data and / or trends of one or more other analytes and / or other non-analyte data, as shown in block 416. For example, in certain embodiments, the decision support engine 114 may determine the intensity of the physical activity based on the lactate level and / or trend, the glucose level and / or trend, and a mapping of physical intensity to the non-analyte sensor or device data. For example, the decision support engine 114 may access a reference library having various lactate and glucose related ranges, as well as ranges associated with the non-analyte sensor or device data. The decision support engine 114 may then be configured with various rules to utilize the reference library to determine the intensity of the physical activity in which the user is engaged or has engaged, based on the user's own data. For example, a simplified rule may state, "If the user's lactate level is within X range, the user's glucose level is within Y range, and the user's heart rate is within Z range, then the user is engaged in high intensity exercise."
[0128] Examples of non-analyte sensor or device data include accelerometer data, step count data, athletic equipment power meter data, GPS data, heart rate data (e.g., heart rate reserve and HRV), EKG data, EMG data, respiration rate data, temperature data, blood pressure data, galvanic skin response data, oxygen intake data, sleep data, impedance data, etc., such as provided by the continuous or non-continuous non-analyte sensors described above. Other physiological parameters, such as heart rate, may be good indicators of physical activity intensity, depending on the type of physical activity. For example, heart rate may be a good indicator of aerobic low-intensity physical activity, such as walking and light jogging, aerobic moderate-intensity physical activity, such as swimming laps and cycling, and anaerobic high-intensity physical activity, such as sprinting, jumping, and high intensity interval training (HIIT). Typically, low intensity physical activity results in a user's heart rate hovering between about 40% to about 60% of the user's maximum heart rate (e.g., 220 minus the user's age). Moderate intensity physical activity results in a heart rate of about 50% to about 70% of the user's maximum heart rate. However, heart rate, as well as other non-analyte physiological parameters such as accelerometer data, may not be good indicators of certain anaerobic high-intensity physical activities such as weightlifting. Thus, using a combination of lactate, glucose, and non-analyte sensor or device data may provide a more accurate characterization of the intensity and / or type of physical activity engaged in by the user.
[0129] In block 418, the decision support system 100 generates a glycemic event prediction based on the determined patient's physiological state, the analyte data for the multiple analytes, the trends for each of the multiple analytes, and, in certain embodiments, other non-analyte sensor or device data. Examples of predicted glycemic events include hypoglycemia, hyperglycemia, increased insulin sensitivity, increased insulin resistance, and other metabolic disease events such as hyperketonemia (DKA), hyperlactemia, lactic acidosis, etc. Block 418, in certain embodiments, may be performed by the decision support engine 114 illustrated in FIG. 1.
[0130] In certain embodiments, glycemic event predictions generated by the decision support system 100 are based in part on a determination of the patient's insulin sensitivity during or after the period. As explained above, insulin sensitivity may be informed by a determination of the physical activity engaged in by the patient and / or its intensity. For example, depending on the intensity and duration, physical activity may stimulate either short-term (e.g., immediately after the physical activity and up to 72 hours thereafter) as well as long-term (e.g., beyond 72 hours) changes (increases or decreases) in insulin sensitivity. Thus, when considered along with changes in analyte levels caused by physical activity, insulin sensitivity may facilitate more accurate prediction of physical activity-induced glycemic events.
[0131] In certain embodiments, the change in insulin sensitivity may be based on analyte levels and / or trends. For example, the change in insulin sensitivity may be determined based on measurements of electrolytes such as glucose, ketones, glycerol, sodium and potassium, calculated measurements such as anion gap, and other suitable analytes, including any of the analytes discussed herein. In certain embodiments, the change in insulin sensitivity may be based on lactate levels and / or trends. For example, if a patient's lactate level falls below the patient's lactate baseline after engaging in physical activity, this may indicate that the patient's insulin sensitivity is increasing. Thus, after determining an increase in insulin sensitivity, the decision support system 100 may recommend adjusting / modifying insulin dosage until the patient's lactate level returns to baseline. For example, the decision support system 100 may recommend that the patient adjust their insulin-to-carbohydrate ratio, e.g., reduce their insulin-to-carbohydrate ratio, to avoid hypoglycemia.
[0132] In certain embodiments, the change in insulin sensitivity may be based on lactate levels and / or trends in addition to glucose levels and / or trends. For example, if a patient's lactate levels are decreasing after a lactate spike following physical activity and the patient's glucose levels are simultaneously increasing, this may indicate a glycemic rebound in which the patient's insulin sensitivity may be reduced since lactate is being used as the body's primary fuel source. Thus, the decision support system 100 may recommend to the patient to avoid administering insulin, because doing so may result in insulin "stacking," where administering insulin causes an increase in residual insulin levels but does not immediately affect the patient's increased glucose levels because lactate is still being cleared.
[0133] In certain examples, the decision support system 100 determines a change in the patient's insulin sensitivity for up to 12 hours, up to 24 hours, up to 36 hours, up to 48 hours, up to 60 hours, or up to 72 hours after the patient engages in physical activity. Such a change in insulin sensitivity may be determined and / or confirmed, for example, by a lower patient's fasting lactate level and / or from an insulin pen or pump in communication with the decision support system 100 and by a greater than normal response of the patient to an insulin administration. During this period of insulin sensitivity, the decision support system 100 may adjust one or more glucose thresholds of the user, as well as one or more parameters utilized in determining insulin administration recommendations, meal recommendations, and / or other post-physical activity glycemic treatment recommendations. For example, in certain embodiments, such parameters may include a desired residual insulin level and / or insulin activity. Thus, in certain examples, the adjusted parameters are utilized to modify the insulin administration recommendation. In certain examples, the adjusted parameters are utilized to modify the meal recommendation.
[0134] In certain embodiments, the decision support system 100 generates a prediction of hypoglycemia. Mild to severe hypoglycemia may occur as a result of increased insulin sensitivity and / or administered insulin, as well as decreased glucose levels caused by low to moderate intensity physical activity. Hypoglycemia often occurs within about 45 minutes of prolonged low to moderate intensity physical activity, and in certain cases may occur at night, especially when physical activity is performed in the afternoon or evening. A hypoglycemic event may reduce the effectiveness of counterregulatory responses in subsequent hypoglycemic events, as early as the next day. Thus, with successive days of low to moderate intensity physical activity, the likelihood and / or severity of induced hypoglycemia may increase, and the prediction and / or recommendation may be adjusted to take this into account.
[0135] In certain examples, the decision support system 100 generates a prediction of hyperglycemia. Mild to severe hyperglycemia can occur as a result of increased glucose levels caused by the liver converting lactate to glucose during high-intensity physical activity. During such high-intensity physical activity, the conversion of lactate to glucose by the liver can be resistant to increases in insulin. However, following such high-intensity physical activity, hyperglycemia can persist for approximately one hour before insulin secretion stimulates glucose uptake and equilibrates glucose levels in the body.
[0136] Different methods for generating glycemic event predictions may be used by the decision support engine 114. In certain embodiments, the decision support engine 114 may use rule-based models to provide real-time or retrospective decision support for physical activity-induced glycemic event predictions. Rule-based models involve using a set of rules to manipulate and / or analyze data. These rules are often referred to as "If Statements" because they tend to follow a one-liner: "If X happens, then do Y." In particular, the decision support engine 114 may apply rule statements (e.g., if, then statements) to evaluate the likelihood and / or occurrence of a glycemic event.
[0137] For example, a first rule may be "If a patient's glucose level is between X and Y and the patient is engaged in more than Z minutes of low-to-moderate intensity physical activity, the patient is likely to experience hypoglycemia within a certain amount of time," and a second rule may be "If a patient's glucose level is between Y and Z and the patient is engaged in high intensity activity, the patient is likely to experience hyperglycemia within a certain amount of time." Analyte data for multiple analytes, trends for each of the multiple analytes, and in certain embodiments, other non-analyte sensor or device data may be applied against these predefined rules to predict physical activity induced glycemic events.
[0138] Such rules may be defined based on empirical research and maintained in a reference library by the decision support engine 114. For example, the reference library may maintain ranges of analyte levels, ranges of differences in analyte trends, ranges of other non-analyte sensor and / or device data, etc. that may be mapped to different glycemic events. In certain embodiments, such rules may be determined based on the training server system 140 analyzing historical patient records from the historical record database 112.
[0139] In some cases, the reference library may be very granular. For example, other factors may be used in the reference library to create such "rules". Other factors include gender, age, diet, medical history, fitness level, etc. Increasing granularity may provide more accurate output. As an example, including fitness level in the rules-based approach used by the decision support engine 114 may help to signal differences in lactate threshold or glucose levels such that glycemic event predictions by the decision support engine 114 are more accurate. For example, a patient with higher fitness may have a higher lactate threshold compared to a physically unfit patient's lactate threshold, and thus a patient with a higher level of fitness may be able to engage in longer periods of high intensity efforts without increasing the risk of physical activity-induced glycemic events and / or generally be less likely to experience physical activity-induced glycemic events, and thus physical fitness may be an important factor to analyze in the rules-based approach to better predict physical activity-induced glycemic events.
[0140] In certain embodiments, as an alternative to using rule-based models, AI models such as machine learning models may be used to provide real-time and retrospective decision support for physical activity-induced glycemic event prediction. In certain embodiments, the decision support engine 114 may deploy one or more of these machine learning models to perform prediction of a user's glycemic events.
[0141] In particular, the decision support engine 114 may retrieve information from a user profile 118 associated with the user stored in the user database 110, characterize the information about the user stored in the user profile 118 into one or more features, and use these features as inputs to such models. Alternatively, the information provided by the user's profile 118 may be characterized by another entity, and the features may then be provided to the decision support engine 114 to be used as inputs to the ML model. In machine learning, a feature is an individual measurable characteristic or feature that is useful for analysis. In certain embodiments, the features associated with the user may be used as inputs to one or more of the models to assess the likelihood and / or occurrence of a glycemic event for the user. In certain embodiments, the features associated with the user may be used as inputs to one or more of the models to identify glucose stabilizing actions that may be provided to the user as recommendations for treatment of the predicted glycemic event, as described in block 420 below. Details associated with how one or more machine learning models may be trained to provide real-time and retrospective decision support for glycemic event prediction are further discussed in connection with FIG. 5.
[0142] As mentioned above, in certain embodiments, the analyte data may be used by the decision support engine 114 to generate a glycemic event prediction for the user in block 418. The analyte data, including glucose and lactate data, as well as ketone data and / or any other analytes mentioned above (e.g., from measurements by the continuous analyte monitoring system 104), may be used as inputs to such machine learning models and / or rule-based models to predict physical activity induced glycemic events.
[0143] In some cases, the method 400 continues at block 420 by the decision support engine 114 generating one or more recommendations for treatment based at least in part on the glycemic event prediction at block 418. In particular, the decision support engine 114 makes a glycemic event treatment decision or recommendation for the user. The treatment recommendation may include an alarm and / or an immediate action recommendation, such as a recommendation for administration of a medication or modification of food consumption or physical activity, or a lifestyle change. Examples of recommendations generated by the decision support system 100 include recommending administration of a small or regular insulin dose during or after physical activity, recommending consumption of carbohydrates during or after physical activity, recommending a type of food to consume or avoid after physical activity, recommending a type, duration, and / or intensity of physical activity, recommending an active cool-down activity (e.g., walking) with reduced intensity, recommending a rest period during physical activity, recommending an insulin dose or increased carbohydrate consumption before the next physical activity, and recommending a therapy modification. In certain embodiments, in lieu of a drug administration recommendation, the decision support system 100 may calculate a recommended drug dosage and transmit the recommendation to a drug administration device (e.g., medical device 108) for automatically administering the recommended dosage of the drug to the user.
[0144] The decision support engine 114 may output such recommendations for treatment to the user (e.g., through application 106). In some embodiments, the recommendations may be displayed for viewing by the patient, for example, on the display device 107 illustrated in FIG. 1 and the display devices 210, 220, 230, and 240 illustrated in FIG.
[0145] In certain examples, if hypoglycemia is predicted, the decision support system 100 may generate a recommendation to increase carbohydrate intake and / or reduce insulin dosage to maintain glucose stability, either during or after the current physical activity, or before engaging in the next physical activity, based on the patient's predicted behavior, including, for example, predicted physical activity and / or carbohydrate consumption. Aggressive reductions in insulin dosage or omission of doses may result in hyperglycemia before and during aerobic exercise, and therefore a more moderate reduction in insulin dosage may be recommended.
[0146] In certain examples, if hyperglycemia is predicted, the decision support system 100 may generate recommendations for an additional dose of insulin after physical activity or consumption of a meal before engaging in the next physical activity, for example, based on the patient's predicted behavior. An additional, regular-sized dose of insulin may result in hypoglycemia after physical activity, and therefore a smaller dose may be recommended instead. In certain examples, the decision support system 100 may recommend taking short rest intervals between high-intensity physical activities, as these may amplify hyperglycemia. Thus, longer rest intervals may be recommended instead. In certain embodiments, the decision support system 100 may generate recommendations for an additional dose of insulin during physical activity.
[0147] In certain embodiments, if hyperlactemia is determined and a state of increased insulin resistance is predicted due to hyperlactemia, the decision support system 100 may generate a recommendation for increasing insulin dosage until lactate levels normalize or return to baseline, for example, based on predicted patient behavior.
[0148] In certain embodiments, after administration of the insulin dose, the decision support system 100 may continue to monitor the user's lactate level to determine the lactate response to the insulin dose and then adjust the predicted glycemic event based on the lactate response. For example, little or no change in lactate level (e.g., less than a 50% change in lactate level) may indicate greater insulin sensitivity and thus a more substantial change in glucose level in response to the insulin dose. Alternatively, a large difference in lactate level (e.g., a 50% or greater change in lactate level) may indicate greater insulin resistance and thus little or no change in glucose level in response to the insulin dose.
[0149] In certain embodiments, if it is determined that the user is engaged in physical activity during the period, the decision support system 100 may generate a recommendation in block 410 to perform active recovery following the physical activity. Active recovery or active cool down may include the user performing physical activity that is gradually decreased in intensity to allow the user's body to gradually transition to a resting or near-resting state. Performing active recovery after physical activity may facilitate a more rapid normalization of the user's elevated lactate levels caused by the physical activity, which in turn may result in improved glycemic stability. For example, the user may exhibit elevated lactate levels after performing physical activity. At some point after the physical activity, the user's lactate levels may begin to normalize (e.g., decrease) as the user's body consumes excess lactate for energy. And since glucose is not being consumed while lactate is being utilized as fuel for the user's body, the user's glucose levels may begin to increase and the user's insulin sensitivity may decrease. Thus, the user's glucose may spike, resulting in hyperglycemia in certain embodiments. However, by performing active recovery after physical activity, a user may normalize lactate levels faster, thus preventing or reducing the likelihood of transitioning to a hyperglycemic state and decreasing the patient's period of glucose insensitivity / tolerance. Thus, the decision support system 100, in certain embodiments, may generate a recommendation to perform active recovery if such hyperglycemia is predicted. Examples of active recovery generally include jogging, walking, stretching, etc.
[0150] In embodiments in which a recommendation for active recovery is generated, the decision support system 100 may further provide guidance regarding the type and duration of active recovery. In certain embodiments, the decision support system 100 may provide guidance on how to achieve a desired lactate gradient or absolute lactate level reduction to prevent transition to a hyperglycemic state. In such embodiments, the guidance may be based on a correlation of the user's past lactate data (e.g., the user's past lactate levels and trends) with past non-analyte sensor data, such as the user's past heart rate data and / or past respiration rate data. In certain embodiments, the type and duration of active recovery recommended by the decision support system 100 may be further based on a determination of a change in the patient's insulin sensitivity.
[0151] In particular embodiments, if the decision support engine 114 determines that the user is experiencing glycemic rebound after physical activity, the decision support system 100 may generate recommendations to avoid consumption of foods that would increase lactate levels and worsen the glycemic rebound (e.g., causing a further increase in glucose spikes).
[0152] In certain embodiments, if the decision support engine 114 determines in block 410 that the user is engaged in physical activity during the period and / or if the decision support engine 114 determines in block 414 the intensity, type, and / or duration of the physical activity, the decision support system 100 may generate a recommendation to modify the intensity and / or type of physical activity to promote improved glycemic stability after the physical activity. For example, during short duration, moderate-to-high intensity physical activity, the user may consume glucose at a relatively fast rate but over a short period of time, which may lead to a temporary glycemic rebound after performing the physical activity with a reduced risk of hypoglycemia. However, during long duration, low-to-moderate intensity physical activity, the user may consume glucose at a relatively slower rate but over a longer period of time, which may not cause the same transient post-physical activity glycemic rebound as moderate-to-high intensity activity, thereby leading to an increased risk of hypoglycemia. Thus, in certain embodiments, decision support system 100 may generate recommendations to modify the intensity and / or type of physical activity for which hypoglycemia (or hyperglycemia) was predicted. For example, if a user is performing long duration, low-to-moderate intensity physical activity, decision support system 100 may generate a recommendation for the user to increase the intensity of the activity, for example, toward the end of the period, to promote glycemic rebound after the physical activity.
[0153] 5 is a flow diagram illustrating a method 500 for training a machine learning model to provide predictions of physical activity-induced glycemic events according to certain embodiments of the present disclosure. In certain embodiments, the method 500 is used to train a model to predict current or future glycemic events in a patient engaged in physical activity, such as the user illustrated in FIG.
[0154] The method 500 begins at block 5602 with retrieving data from a medical history record database, such as the medical history record database 112 illustrated in Figure 1, by a training server system, such as the training server system 140 illustrated in Figure 1. As referred to herein, the medical history record database 112 may provide a repository of up-to-date information and medical history information for users of the continuous analyte monitoring system and connected mobile health application, such as users of the continuous analyte monitoring system 104 and application 106 illustrated in Figure 1, as well as data for one or more patients who are not or were not previously users of the continuous analyte monitoring system 104 and / or application 106.
[0155] Retrieval of data from the historical record database 112 by the training server system 140, in block 502, may include retrieval of all or any subset of the information maintained by the historical record database 112. For example, if the historical record database 112 stores information about 100,000 patients (e.g., non-users and users of the continuous analyte monitoring system 104 and application 106), the data retrieved by the training server system 140 to train one or more machine learning models may include information about all 100,000 patients, or only a subset of the data for those patients, e.g., data associated with only 50,000 patients, or only data from the past 10 years.
[0156] As an illustrative example, in block 502, the training server system 140 may retrieve information for 100,000 diabetic patients stored in the medical history record database 112 to train a model for predicting current or future physical activity induced glycemic events in a user. Each of the 100,000 patients may have a corresponding data record, such as the user profile 118 illustrated in FIG. 1, stored in the medical history record database 112. Each data record may include information such as the information discussed with respect to FIG.
[0157] At block 504, the method 500 continues by the training server system 140 selecting one of the historical patient records retrieved by the training server system 140 at block 502. The record contains information associated with the patient, such as information stored in the patient's user profile. Examples of the types of information included in the patient's user profile are provided above. The training server system 140 may use any suitable criteria (e.g., starting with the oldest record, starting with the most recent record, etc.) for the selection of the historical patient record, as the training server system 140 iterates through each historical access record in the training set until all records have been used to train the machine learning model or until the machine learning model accurately predicts physical activity induced glycemic events for each historical patient record input to the model.
[0158] At block 506, the method 500 continues by the training server system 140 extracting one or more features of the selected historical patient record. These features are extracted to be used as input features for the machine learning model. For example, the user profile associated with the patient selected at block 504 may include information regarding at least the patient's age, the patient's gender, the patient's fitness level, the average change (e.g., average difference) in the patient's glucose level during or after low-moderate intensity physical activity, the average change in glucose level for the patient during or after moderate-high intensity physical activity, the average change in lactate level for the patient during or after low-moderate intensity physical activity, the average change in lactate level for the patient during or after moderate-high intensity physical activity, the average duration between low-moderate intensity or moderate-high intensity physical activity and a glycemic event for the patient, etc. The features used to train the machine learning model may vary in different embodiments.
[0159] At block 508, the method 500 continues by the training server system 140 training one or more machine learning models based on the selected historical patient records. In some embodiments, the training server does so by providing the features (e.g., extracted at block 506) as input to the model. The model may be a new model initialized with random weights and parameters, or may be partially or fully pre-trained (e.g., based on a previous training round). Based on the input features, the model in training generates some output. The output may include a prediction of physical activity induced glycemic events, previously recommended treatments, or similar metrics.
[0160] In certain embodiments, the training server system 140 compares this generated output to the actual labels associated with the historical patient record and calculates a loss based on the difference between the actual outcome and the generated outcome. This loss is then used to refine (e.g., via backpropagation) one or more internal weights and parameters of the model so that the model learns to more accurately predict the occurrence of physical activity-induced glycemic events (or their recommended treatments).
[0161] At block 510, the method 500 continues by the training server system 140 determining whether additional training is needed. This may include evaluating whether any additional historical patient records remain in the training dataset. If at block 510, the training server system 140 determines that all training data has been input to the machine learning model, at block 512, the training server system 140 deploys the trained model for physical activity-induced glycemic event prediction during runtime. In some embodiments, this includes transmitting any indication of the trained model (e.g., a weight vector) that can be used to instantiate the model on another device. For example, the training server system 140 may deploy the trained model to the decision support engine 114. The model can then be used to assess in real time the likelihood of a physical activity-induced glycemic event for a user using the application 106.
[0162] If, at block 510, the training server system 140 determines that not all historical patient records of the training data have been input into the model for training, at block 514, the training server system 140 determines whether the model has reached a predetermined minimum accuracy (e.g., 90% accuracy, 95% accuracy, etc.). If the predetermined minimum accuracy has not been met, the training server system 140 determines that additional training remains and the method 500 returns to block 504. Alternatively, if the machine learning model accurately predicts the predetermined minimum accuracy (e.g., accurately predicts 90% or 95% of the time), at block 512, the training server system 140 deploys the trained model for physical activity induced glycemic event prediction during runtime.
[0163] By iteratively processing each data set corresponding to each historical patient, the model may be iteratively refined to generate accurate predictions of the patient's physical activity induced glycemic event predictions.
[0164] FIG. 6 is a block diagram illustrating a computing device 600 configured for (1) predicting current or future physical activity-induced glycemic events and / or (2) providing decision support for managing a patient's diabetes in relation to physical activity, according to certain embodiments disclosed herein. Although depicted as a single physical device, in embodiments, computing device 600 may be implemented using virtual devices and / or across several devices, such as in a cloud environment. As illustrated, computing device 600 includes a processor 605, memory 610, storage 615, a network interface 625, and one or more I / O interfaces 620. In the illustrated embodiment, processor 605 retrieves and executes programming instructions stored in memory 610, as well as stores and retrieves application data present in storage 615. Processor 605 generally represents a single CPU and / or GPU, multiple CPUs and / or GPUs, a single CPU and / or GPU with multiple processing cores, etc. Memory 610 is included to generally represent random access memory (RAM). Storage 615 may be any combination of disk drives, flash-based storage devices, and the like, and may include fixed and / or removable storage devices, such as fixed disk drives, removable memory cards, cache, optical storage, network attached storage (NAS), or a storage area network (SAN).
[0165] In some embodiments, I / O devices 635 (keyboard, monitor, etc.) can be connected via I / O interface 620. Additionally, via network interface 625, computing device 600 can be communicatively coupled to one or more other devices and components, such as user database 110 and / or medical history record database 112. In particular embodiments, computing device 600 is communicatively coupled to other devices via a network, which may include the Internet, a local network, etc. The network may include wired connections, wireless connections, or a combination of wired and wireless connections. As illustrated, processor 605, memory 610, storage 615, network interface 625, and I / O interface 620 are communicatively coupled by one or more interconnects 630. In particular embodiments, computing device 600 represents a display device 107 associated with a user. In particular embodiments, as discussed above, display device 107 may include a user's laptop, computer, smartphone, etc. In another embodiment, computing device 600 is a server running in a cloud environment.
[0166] In the illustrated embodiment, the storage 615 includes a user profile 118. The memory 610 includes a decision support engine 114, which itself includes a DAM 116. The decision support engine 114 is executed by the computing device 600 to perform operations in the workflow 400 of Figures 4A-4B and / or operations of the method 500 of Figure 5 to predict current or future physical activity induced glycemic events and / or to provide decision support for managing a patient's diabetes in relation to physical activity.
[0167] Exemplary embodiments Embodiment 1: In a particular embodiment, a method for generating a glycemic event prediction is provided, the method including continuously monitoring a plurality of analytes of a patient during a time period to obtain analyte data, processing the analyte data from the time period to determine a trend for each of the plurality of analytes, determining a physiological state of the patient based on the trend for each of the plurality of analytes, the determining comprising determining whether the patient is engaging in physical activity, and predicting current or future glycemic events for the patient based on the patient's physiological state, the analyte data, and the trend for each of the plurality of analytes.
[0168] Embodiment 2: The method of embodiment 1, wherein the plurality of analytes comprises at least lactate and glucose.
[0169] Embodiment 3: The method of embodiment 1, wherein determining the patient's physiological state further comprises determining an intensity level of physical activity engaged in by the patient.
[0170] Embodiment 4: The method of embodiment 1, wherein determining whether the patient is engaged in physical activity is based on analyte data and / or trends of a plurality of analytes.
[0171] Embodiment 5: The method of embodiment 1, further comprising generating one or more recommendations for treatment for the patient based at least in part on the patient's current or future glycemic events.
[0172] Embodiment 6: The method of embodiment 5, wherein the one or more recommendations for treatment include at least one of: drug administration recommendations, therapy modification recommendations, food consumption recommendations, and physical activity modification recommendations.
[0173] Embodiment 7: The method of embodiment 1, wherein the plurality of analytes further comprises at least one of ketones, glycerol, potassium, and sodium.
[0174] Embodiment 8: The method of embodiment 1, further comprising monitoring other sensor data of the patient during the period using one or more other non-analyte sensors.
[0175] Embodiment 9: The method of embodiment 8, wherein the one or more other non-analyte sensors comprise at least one of an accelerometer, an impedance sensor, an electrocardiogram (EKG) sensor, a blood pressure sensor, a heart rate monitor, or a respiration sensor.
[0176] Embodiment 10: The method of embodiment 1, wherein the analyte data for lactate is utilized to distinguish between sampling noise and actual analyte data for glucose.
[0177] Embodiment 11: The method of embodiment 1, wherein the glycemic event prediction is generated using a model trained using training data to predict glycemic events induced by the patient's physical activity.
[0178] Additional Considerations The methods disclosed herein include one or more steps or actions for achieving the method. These method steps and / or actions may be interchanged with one another without departing from the scope of the claims. In other words, unless a specific order of steps or actions is specifically specified, the order and / or use of specific steps and / or actions may be modified without departing from the scope of the claims.
[0179] As used herein, a phrase referring to "at least one of" a list of items refers to any combination of those items, including single members. By way of example, "at least one of a, b, or c" is intended to cover a, b, c, ab, ac, bc, and abc, as well as any combination of multiples of the same element (e.g., aa, aaa, aab, aac, abb, acc, bb, bbb, bbc, cc, and ccc, or any other permutation of a, b, and c).
[0180] The foregoing description is provided to enable those skilled in the art to practice the various embodiments described herein. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other embodiments. Thus, the claims are not intended to be limited to the embodiments shown herein, but are to be given the full scope consistent with the claim language, in which reference to an element in the singular is not intended to mean "one and only one" unless specifically stated otherwise, but is intended to mean "one or more". The term "some" refers to one or more, unless specifically stated otherwise. All structural and functional equivalents to the elements of the various embodiments described throughout this disclosure that are now known or that will become known in the future to those skilled in the art are expressly incorporated herein by reference and are intended to be encompassed within the scope of the claims. Furthermore, nothing disclosed herein is intended to be dedicated to the public, regardless of whether such disclosure is expressly recited in the claims. No claim element is to be construed under the provisions of 35 U.S.C. 112(f) unless the element is expressly recited using the phrase "means for," or, in the case of a method claim, the element is recited using the phrase "step for."
[0181] While various examples of the present invention have been described above, it should be understood that the examples are presented merely as examples and not as limitations. Similarly, various figures may depict exemplary architectural or other configurations for the present disclosure, which configurations are made to aid in understanding features and functionality that may be included in the present disclosure. The present disclosure is not limited to the illustrated exemplary architecture or configurations, but may be implemented using various alternative architectures and configurations. Additionally, while the present disclosure has been described above with respect to various exemplary embodiments and aspects, it should be understood that various features and functions described in one or more of the individual embodiments are not limited in their applicability to the particular embodiment in which they are described. On the contrary, those features and functions may be applied, alone or in any combination, to one or more of the other embodiments of the present disclosure, regardless of whether such an embodiment is described and regardless of whether such features are presented as part of a described embodiment. Thus, the breadth and scope of the present disclosure should not be limited by any of the exemplary embodiments described above.
[0182] All references cited herein are incorporated herein by reference in their entirety. To the extent that publications and patents or patent applications incorporated by reference conflict with the present disclosure contained herein, the present specification is intended to supersede and / or take precedence over any such conflicting material.
[0183] Unless otherwise defined, all terms (including technical and scientific terms) are to be given their ordinary and customary meanings to those of ordinary skill in the art and are not to be limited to any special or customized meaning unless expressly defined as such herein.
[0184] Terms and phrases used in this application, and variations thereof, unless expressly stated otherwise, particularly in the appended claims, are to be construed as open ended, as opposed to limiting. As an example of the foregoing, the term "including" should be read to mean "including without limitation," "including, but not limited to," and the like. The term "comprising" as used herein is synonymous with "including," "containing," or "characterized by," and is inclusive or open ended and does not exclude additional, unrecited elements or method steps. The term "having" should be interpreted as "having at least." The term "including" should be interpreted as "including, but not limited to." The term "examples" is used to provide illustrative examples of the items in the discussion, and is not an exhaustive or limiting list of such examples. Adjectives such as "known," "conventional," "standard," and terms of similar import should not be construed as limiting the items being described to a given period of time or to items available at a given time, but rather should be read to embrace known, conventional, or standard technology that may be available or known at any time now or in the future, and words such as "preferably," "preferred," "desired," or "desirable," and similar imports should not be understood to imply that a particular feature is critical, essential, or even essential to the structure or function of the invention, but rather are merely intended to highlight alternative or additional features that may or may not be utilized in a particular example of the invention. Similarly, a group of items connected with the conjunction "and" should not be read as requiring that any and all of those items be present in the group, but rather should be read as "and / or" unless expressly stated otherwise.Similarly, a group of items connected with the conjunction "or" should not be read as requiring mutual exclusivity among the group, but rather as "and / or" unless expressly stated otherwise.
[0185] As used herein, the term "comprising" is synonymous with "including," and "containing" or "characterized by" is inclusive or open-ended and does not exclude additional, unrecited elements or method steps.
[0186] All numerical values expressing amounts of raw materials, reaction conditions, and the like used herein should be understood to be modified in all instances by the term "about." Accordingly, unless otherwise indicated, the numerical parameters set forth herein are approximations that may vary depending upon the desired properties sought to be obtained. At the very least, and not as an attempt to limit the application of the doctrine of equivalents to the scope of any claims in any application claiming priority to this application, each numerical parameter should be construed in light of the number of significant digits and ordinary rounding approaches.
[0187] Moreover, although the foregoing has been described in some detail by way of illustration and example for purposes of clarity and understanding, it will be apparent to those skilled in the art that certain changes and modifications may be practiced. Thus, the description and examples should not be construed as limiting the scope of the invention to the specific examples and embodiments described herein, but rather as covering all modifications and alternatives that come within the true scope and spirit of the invention. [Explanation of symbols]
[0188] 100 Decision Support Systems 102 users 104 Analyte Monitoring Systems 106 Applications 107 Display Devices 108 Medical Devices 110 User Database 112 Medical History Record Database 114 Decision Support Engine 116 Data Analysis Module (DAM) 118 User Profile 120 Demographic Information 122 Disease Information 124 Medication Information 126 Application Data 128 inputs 130 Metric 140 Training Server System 144 Output 202 Continuous Analyte Sensors 204 Sensor Electronics Module 206 Non-analyte sensors 208 Medical Devices 210 Display Device 212 Touch Screen Display 220 Display Device 222 Touch Screen Display 230 Display Device 232 Touch Screen Display 240 Display Devices 600 computing devices 605 Processor 610 Memory 615 Storage 620 I / O Interface 625 Network Interface 630 Interconnection 635 I / O Devices
Claims
1. 1. A method for generating a glycemic event prediction, the method comprising: continuously monitoring a plurality of analytes in a patient over a period of time to obtain analyte data, the plurality of analytes including at least lactate and glucose; processing the analyte data from the time period to determine a trend for each of the plurality of analytes; determining a physiological state of the patient based on the trend of each of the plurality of analytes, the physiological state including determining whether the patient is engaged in physical activity; predicting a current or future glycemic event for the patient based on the physiological condition of the patient, the analyte data, and the trend for each of the plurality of analytes; A method comprising:
2. The method of claim 1 , wherein determining the physiological state of the patient further comprises determining an intensity level of the physical activity engaged in by the patient.
3. The method of claim 1 , wherein determining whether the patient is engaged in physical activity is based on the analyte data and / or trends of the plurality of analytes.
4. 10. The method of claim 1, further comprising generating one or more recommendations for treatment for the patient based at least in part on the current or future glycemic events of the patient.
5. The one or more recommendations for treatment include: Medication recommendations, Recommendations for therapy modification, Food consumption recommendations, or Physical activity modification recommendations, The method of claim 4 , comprising at least one of:
6. The method of claim 1 , wherein the plurality of analytes further comprises at least one of ketones, glycerol, potassium, and sodium.
7. 10. The method of claim 1, further comprising monitoring other sensor data for the patient during the period using one or more other non-analyte sensors.
8. 8. The method of claim 7, wherein the one or more other non-analyte sensors comprise at least one of an accelerometer, an impedance sensor, an electrocardiogram (EKG) sensor, a blood pressure sensor, a heart rate monitor, or a respiration sensor.
9. 10. The method of claim 1, wherein the analyte data for lactate is utilized to distinguish between sampling noise and actual analyte data for glucose.
10. 10. The method of claim 1, wherein the glycemic event prediction is generated using a model trained using training data to predict glycemic events induced by physical activity of the patient.
11. 1. A system for providing glycemic event decision support, the system comprising: one or more continuous analyte sensors configured to continuously monitor a plurality of analytes in a patient over a period of time to obtain analyte data, the plurality of analytes including at least lactate and glucose; one or more memories containing executable instructions; one or more processors in data communication with the one or more memories; processing the analyte data from the time period to determine a trend for each of the plurality of analytes; determining a physiological state of the patient based on the trend of each of the plurality of analytes, the physiological state including determining whether the patient is engaged in physical activity; and predicting a current or future glycemic event for the patient based on the physiological condition of the patient, the analyte data, and the trend for each of the plurality of analytes. A system comprising:
12. The system of claim 11 , wherein determining the physiological state of the patient further comprises determining an intensity level of the physical activity engaged in by the patient.
13. The system of claim 11 , wherein determining whether the patient is engaged in physical activity is based on the analyte data and / or trends of the plurality of analytes.
14. The one or more processors:
12. The system of claim 11, further configured to generate one or more recommendations for a therapy for the patient based at least in part on the current or future glycemic events of the patient.
15. The one or more recommendations for treatment include: Medication recommendations, Recommendations for therapy modification, Food consumption recommendations, or Physical activity modification recommendations, The system of claim 14 , comprising at least one of:
16. 16. The system of claim 15, wherein the plurality of analytes further comprises at least one of ketones, glycerol, potassium, and sodium.
17. 12. The system of claim 11, further comprising one or more non-analyte sensors configured to monitor non-analyte sensor data of the patient during the period of time, the one or more processors further configured to process the non-analyte sensor data from the period of time and determine the physiological condition of the patient based on the processed non-analyte sensor data.
18. 20. The system of claim 17, wherein the one or more non-analyte sensors comprise at least one of an accelerometer, an impedance sensor, an electrocardiogram (EKG) sensor, a blood pressure sensor, a heart rate monitor, or a respiration sensor.
19. 12. The system of claim 11, wherein the one or more processors are further configured to utilize the analyte data for lactate to distinguish between sampling noise and actual analyte data for glucose.
20. 12. The system of claim 11, wherein the one or more processors are further configured to generate a prediction of a glycemic event using a model trained using training data to predict glycemic events induced by physical activity of the patient.
Citation Information
Patent Citations
Glycemic urgency assessment and warning interface
JP2017515520A
Personal health monitoring system, multi-user health monitoring system, and method
JP2020531216A
Multi-analyte continuous glucose monitoring
US20190246962A1
Cited By
system
JP2026140331A
system
JP7927898B2
