A system for predicting undiagnosed prediabetes using a machine learning model and method thereof

Abstract

An AI-based system (100) designed to assess prediabetes risk in individuals. The system (100) integrates seamlessly with the user devices via rest APIs, ensuring secure communication through https. A data collection module (122) collects raw data forwards the data to standardization and pre-processing modules (124) for preparation. The pre-processed data is transmitted to the server (104) through a network (102) using an API integration module (126) and thereafter forwarded to the input module (112) that authenticates and validates patient data, which is transmitted to the processing module (114) that refines the data, while a prediction module (116) employs a deep learning model with an AUC of 0.87 to evaluate prediabetes risk. A clinical pathway (118) and risk prediction response (120) modules provide personalized protocols, risk status, and clinical recommendations. It (100) complements clinician decision-making, adheres to ISO 13485 standards, and serves as a cost-effective community-based prediabetes risk assessment.

Description

4. DESCRIPTION:Field of the invention:

[0001] The present disclosure generally relates to the technical field of healthcare and, in specific, relates to an artificial intelligence (Al)-based system and method to predict the risk ofundiagnosed prediabetes and Type2 Diabetes Mellitus of at least one Individual using at least one machine learning model.Background of the invention:

[0002] Prediabetes is an intermediate stage in the progression of normal blood glucose regulation to Type 2 Diabetes Mellitus (T2DM). As the prevalence of T2DM increases globally, the need for early detection and prevention of prediabetes becomes increasingly crucial.

[0003] The global prevalence of Prediabetes, defined by Impaired glucose tolerance (IGT) or Impaired fasting glucose (IFG), is substantial and growing. The global prevalence of IGT was 9.1% (464 million) and is projected to increase to 10.0% (638 million) in 2045. The global prevalence of IFG was5.8% (298 million) and is projected to increase to 6.5% (414 million).

[0004] The ICMR-INDIAB study found that the overall prevalence of diabetes in India was 11.4%, while Prediabetes was at 15.3%. Existing systems for predicting prediabetes and Type 2 Diabetes Mellitus, such as the CDC Prediabetes Risk Test, the ADA Risk Test, and AUSDRISK, are primarily based on static risk factors and historical data, and they often fail to account for dynamic changes in a person's health status and lifestyle factors. Moreover, these systems generally do not consider the unique lifestyle factors and comprehensive clinical parameters that are specific to the Indian population, thereby limiting their applicability and accuracy.

[0005] Even systems specifically designed for the Indian population, like the Indian Diabetes Risk Score (IDRS), which was developed based on 317 subjects, have their shortcomings. They primarily rely on static demographic factors and baseline clinical parameters without2adequately reflecting the unique lifestyle factors of the Indian population. Most of the systemssuch as the CDC Prediabetes Risk Test, the ADA Risk Test, and AUSDRISK are primarily developed using data from non-lndian populations, which can limit their accuracy and relevance for the Indian population.

[0006] Consequently, existing Systems tend to provide broad categorizations and generic guidelines, lacking the capability to offer personalized risk assessments or address specific patient needs. They also lack a feedback loop and validation in other populations to continuously refine risk predictions based on patient outcomes, further limiting their utility and accuracy.To address these limitations, Apollo Hospitals has developed the Al-Based Prediabetes Risk Assessment system. The study of Development and Validation of a Multivariable Prediction Model to Determine the Risk of Undiagnosed Prediabetes and Type 2 Diabetes Mellitus, harnessed data from 22,560 individuals. However, due to incomplete fields, data from 9194 participants was excluded, resulting in a final development cohort of 13,366 individuals. These individuals were spread across multiple centers in regions like Hyderabad, Chennai, Kolkata, Bangalore, Bhubaneswar, and Bilaspur. The data was gathered from the Apollo Health Check and extracted from the Electronic Medical Records (EMR) of Apollo hospitals. All individuals provided their consent for their data to be used, ensuring ethical compliance.

[0007] The study participants, aged 18 to 79, had no prior diagnosis of diabetes or prediabetes and were not receiving any diabetic treatment or medication. All the collected data was securely stored in a codified format after removing any Patient Identifiable Information. The exclusion criteria for the study were comprehensive. It excluded patients with immediate or urgent conditions such as life-threatening conditions, altered physiological criteria (Triage 1 & 2 of Canadian Triage Acuity Scale), moderate to severe respiratory distress, trauma, pregnancy or the post-partum period, circulatory shock, coma, altered level of consciousness, fever indicative of sepsis or present in immuno-compromised patients, and any form of poisoning. Additionally, the model does not account for patients with uncontrolled and undiagnosed diabetes, gestational diabetes, and Polycystic Ovarian Disease (PCOD), hence those with these conditions were also excluded.

[0008] The study employed a deep learning model, developed on the TensorFlow library, designed specifically for predicting the likelihood of prediabetes development. The structure of this model is comprised of four layers, each with a distinct number of neurons: the initial layer contains 256 neurons, followed by the second and third layers with 128 and 64 neurons respectively, and culminating in a binary classification layer with a single neuron.

[0009] The activation functions 'relu' and 'sigmoid' were employed in the initial three layers and the final layer respectively, effectively tuning the model for the task of binary classification. This led to the identification of 13 predictors that were relevant and correlated with the outcome variable, HbAlc levels. The list of predictors and their corresponding odds ratios (OR) include Age (3.18), Gender (1.08), Alcohol (1.04), BMI (1.57), Family History (0.99), Diet (0.96), Dyslipidemia (6.67), Hypertension (3.85), Physical Activity (0.17), Past Medical History (1.12), and Symptoms (6.22). In addition to these, the model also takes into account changes in body weight over the last six months and the individual's waist circumference. These predictors were then used to predict the risk of Type 2 diabetes mellitus and prediabetes. Notably, the study found that past medical history, especially conditions like dyslipidemia (elevated lipid levels) and hypertension, significantly increased susceptibility. This was evidenced by odds ratios of 6.67 (95% Cl 5.86 - 7.59) and 3.85 (95% Cl 3.58 - 4.14) respectively. Furthermore, symptoms usually associated with prediabetes or diabetes, such as frequent urination and increased thirst, were also identified as predictive indicators. Interestingly, increased physical activity emerged as a considerable protective factor against these conditions, as shown by an odds ratio of 0.17 (95% Cl 0.16 - 0.19). This finding underscores the potential of lifestyle modifications in managing and preventing prediabetes and Type 2 Diabetes Mellitus.

[0010] To mitigate selection bias, the study included various regional subgroups representative of the total population, thereby enhancing the predictability of prediabetes and Type II Diabetes in India. Additionally, the study investigated two more predictors, waist circumference and change in body weight over the past six months, for their impact on the models and the development of the prediabetes risk score.

[0011] The system also features an integrated Clinical Decision Support System (CDSS) that guides clinicians on the next best actions, such as preventive measures and therapeutic interventions, tailored to individual patient's needs. Moreover, the Apollo Prediabetes Scanner uses a deep learning model that has shown an Area Under the Curve (AUC) of 0.87, indicating its Robustness. By offering a dynamic, personalized, and scalable approach to risk assessment and management, the Apollo Prediabetes Scanner represents a significant advancement in the detection of prediabetes and Type 2 Diabetes Mellitus.Objectives of the invention:

[0012] The primary objective of the present invention is to provide a system for predicting the individual risk of undiagnosed prediabetes and Type-2 diabetes mellitus.

[0013] Yet another objective of the present invention is to develop a system that identifies individuals at risk for developing prediabetes and type 2 diabetes mellitus using a simple assessment system, such as a health risk assessment (HRA).

[0014] Another objective of the present invention is to provide a system that provide a clinical decision support system (CDSS) based on stratified risk to guide lab investigations, nutrition and diet plans, physical activity recommendations, referrals, treatment goals, educational materials, and revisit guidelines.

[0015] Yet another objective of the present invention is to provide a system that uses a machine learning model that can analyze complex interactions between numerous data points, like demographics, medical history, and lifestyle factors providing a more comprehensive and personalized risk assessment compared to Existing methods.

[0016] Another objective of the present invention is to Offer a system that can reduce healthcare costs associated with early prevention and management of type 2 diabetes mellitus.

[0017] Another objective of the present invention is to serve as a supplementary system for physicians by providing the individual's risk status, and offering a clinical decision support system based on the stratified risk level assisting physicians in making informed decisions.

[0018] Yet another objective of the present invention is toprovide a system that delivers personalized risk assessment, optimizing resource allocation and eliminating unnecessary tests and tailoring treatments according to stratified risk of individuals.

[0019] Further objective of the present invention is toutilize this system as a communitybased risk system for early prevention and diagnosis.Summary of the invention:

[0020] The present disclosure proposesa system for predicting undiagnosed prediabetes using a machine learning model and method thereof. The following presents a simplified summary in order to provide a basic understanding of some aspects of the claimed subject matter. This summary is not an extensive overview. It is not intended to identify key / critical elements or to delineate the scope of the claimed subject matter. Its sole purpose is to present some concepts in a simplified form as a prelude to the more detailed description that is presented later.

[0021] In order to overcome the above deficiencies of the prior art, the present disclosure is to solve the technical problem to provide a system for predicting the individual's risk of undiagnosed prediabetes and type-2 diabetes mellitus of at least one individual using at least one machine learning model.

[0022] According to one aspect, the invention provides an artificial intelligence (Al)-based system for predicting the individual risk of undiagnosed prediabetes and type-2 diabetes mellitus. The system supports integration with a user device through various approaches, including the use of REST APIs. This integration method ensures seamless operation and generally employs HTTPS for secure communication. The system comprises a server having a processor and a memory, which is connected to a database for efficient data management. The server, in communication with a user device via a network, coordinates interactionsamong various modules within the system. The plurality of modules comprises client modules, an input module, processing module, prediction module, clinical pathway module, and risk prediction response module.

[0023] The client modules client modules comprise a data collection module, a standardization and pre-processing module, and an application programming interface (API) module.

[0024] In one embodiment, the data collection module is configured to collect raw data from various sources such as electronic medical records (EMRs), Data repositories and Databases. In one embodiment herein, the standardization and pre-processing module is configured to standardize and pre-process the collected data to ensure its integrity and compatibility with the system's processing requirements.

[0025] In one embodiment herein, the application programming interface (API) module is configured to transmit the pre-processed data from the data collection module and the standardization and pre-processing module to the server via the network.

[0026] In one embodiment, the input module is configured to enable authentication and validation of input data, ensuring compliance with required formats from various sources. The one or more health parameters of at least one patient include personal variables like age, gender, physical attributes such as height, and weight (used to calculate BMI), alcohol consumption, dietary habits, physical activity levels, a family history of diabetes mellitus, and a history of hypertension and dyslipidemia. It also includes the patient's past medical history of chronic diseases like cardiovascular, renal, liver, thyroid, stroke, transient ischemic attack, hypertension, and frequent infections.

[0027] Further, the input module includes symptoms of diabetes mellitus such as weight gain or loss, polyuria, polydipsia, polyphagia, weakness or fatigue, blurred vision, recent skin or other infections, vulvovaginitis or balanitis, abdominal pain, waist circumference, and changes in body weight over the past six months, among others. The collected data is transmitted to Processing Module.

[0028] In one embodiment, the processing module transforms and standardizes the data of at least one individual from the input module to meet the necessary input criteria for the subsequent analysis. In one embodiment, the prediction module is structured to categorize an individual's risk levels for the occurrence of pre-diabetes and type-2 diabetes mellitus. In one embodiment herein, the machine learning model is a deep learning model with an AUC of 0.87. In one embodiment, the clinical pathway module is designed to provide a recommended protocol of subsequent actions for at least one individual based on the determined risk thresholds.

[0029] In one embodiment, the risk prediction response module presents the individual's risk score along with a clinical algorithm for the next best action based on the stratified risk of prediabetes. The risk prediction response module provides an overview including the patient's risk level (e.g., high, medium, or low), an Al-predicted score (ranging from 1 to 10) for prediabetes, and a clinical algorithm with recommendations for lab investigation, nutrition and diet, activity, referrals, treatment goals, educational materials, and revisit guidelines.

[0030] According to another aspect, the invention provides a method for predicting the risk of undiagnosed prediabetes and type 2 diabetes mellitus using the system. At one step, the data collection module collects the raw data obtained from various sources and the standardization and pre-processing module processes the raw data to ensure system's processing requirements.

[0031] At another step, the pre-processed data is transmitted to the server via the network through an API integration module and then transmitted to the input module for further validation via the server. Atanother step, the input module initiates the authentication and validation of at least one individual's data, including personal parameters, medical history, and lifestyle factors.

[0032] At another step, the processing module analyzes the input data of the at least one individual to obtain health data using at least one machine learning model, therebypredicting the risk of prediabetes and type 2 diabetes mellitus.At other step, the Prediction Module uses the deep learning model to categorize an individual's risk levels of having prediabetes and type-2 diabetes mellitus based on the input data.At one step, the clinical pathway module generates a recommended protocol of next-based actions for at least one individual based on the determined risk thresholds from the prediction module.

[0033] Further, at other step, the risk prediction response moduledisplays the patient's risk status and score along with a clinical algorithm for the next best action based on the stratified risk of prediabetes.

[0034] Further, objects and advantages of the present invention will be apparent from a study of the following portion of the specification, the claims, and the attached drawings.Detailed description of drawings:

[0035] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate an embodiment of the invention, and, together with the description, explain the principles of the invention.

[0036] FIG. 1 illustrates a block diagram of a system for predicting the risk of undiagnosed prediabetes and type-2 diabetes mellitus, in accordance to an exemplary embodiment of the invention.

[0037] FIG. 2 illustrates a schematic flow diagram of the system for predicting undiagnosed prediabetes and type-2 diabetes mellitus, in accordance to an exemplary embodiment of the invention.

[0038] FIG. 3 illustrates a flowchart of a method for predicting undiagnosed prediabetes and type-2 diabetes mellitus using the system, in accordance to an exemplary embodiment of the invention.

[0039] FIG. 4 illustrates anarchitecture diagram of a method for developing and deploying a multivariable Al-basedsystemfor predicting undiagnosed prediabetes and type-2 diabetes mellitus using the system, in accordance to an exemplary embodiment of the invention.Detailed invention disclosure:

[0040] Various embodiments of the present invention will be described in reference to the accompanying drawings. Wherever possible, same or similar reference numerals are used in the drawings and the description to refer to the same or like parts or steps.

[0041] The present disclosure has been made with a view towards solving the problem with the prior art described above, and it is an object of the present invention to provide a system for predicting the individual risk of undiagnosed prediabetes and Type-2 Diabetes Mellitus of at least individualusing at least one machine learning model.

[0042] According to one exemplary embodiment of the invention, FIG. 1 refers to a block diagram of a system 100 for predicting the risk of undiagnosed prediabetes and type-2 diabetes mellitus. In one embodiment herein, the system 100 is configured to predict the risk of type 2 diabetes mellitus and undiagnosed prediabetes using a machine learning model, which allows potential prevention of type 2 diabetes mellitus development. The machine learning model uses deep learning algorithms to analyze complex interactions between numerous data points, like demographics, medical history, and lifestyle factors of patients, to provide a more comprehensive and personalized risk assessment compared to traditional systems.

[0043] In one embodiment, the system 100 supports integration with a user device through various approaches, one of the methods being the use of REST APIs. This integration method ensures seamless operation and generally employs HTTPS for secure communication. The system comprises a server 104 having a processor 106 and a memory 108, which is10connected to a database 110 for efficient data management. The server 104 is in communication with a user device via a network 102. Importantly, the server 104 is interconnected with the processor 106 and the memory 108, which is connected to a database 110 for efficient data management.

[0044] In one embodiment, the server 104 communicates with aplurality of modules, which is configured to predict a risk of clinical deterioration and mortality of at least one patient. The plurality of modules comprises client modules, an input module 112, a processing module 114, a prediction module 116, a clinical pathway module 118, and a risk prediction responsemodule 120. In one embodiment herein, the client modules are configured to perform multiple functions for identifying patient clinical deterioration. The client modules comprise a data collection module 122, a standardization and pre-processing module 124, and an application programming interface (API) module 126.

[0045] In one embodiment herein, the data collection module 122 is configured to collect raw data from various sources such as electronic medical records (EMRs), Data repositories and Databases. In one embodiment herein, the standardization and pre-processing module 124 is configured to standardize and pre-process the collected data to ensure its integrity and compatibility with the system's processing requirements.

[0046] In one embodiment herein, the application programming interface (API) module 126 is configured to transmit the pre-processed data from the data collection module 122 and the standardization and pre-processing module 124 to the server 104 via the network 102.

[0047] In one embodiment, the input module 112 is configured to receive the transmitted data from the server 104. In one embodiment, the input module 112 initiates authentication and validation of input data, ensuring compliance with required formats. It allows authorized users to facilitate the entry of health parameters for at least one individual, including personal variables like age, gender, physical attributes such as height and weight (used to calculate BMI), alcohol consumption, dietary habits, physical activity levels, family history of diabetes mellitus, and history of hypertension and dyslipidemia. It can also includethe patient's past medical history of chronic diseases like cardiovascular, renal, liver, thyroid, stroke, transient ischemic attack, hypertension, and frequent infections.

[0048] Further, the input module 112 facilitates the entry of symptoms of diabetes mellitus such as weight gain or loss, polyuria, polydipsia, polyphagia, weakness or fatigue, blurred vision, recent skin or other infections, vulvovaginitis or balanitis, abdominal pain, waist circumference, and changes in body weight over the past six months, among others.

[0049] In one embodiment, the processing module 114 transforms and standardizes data from the input module 112 to meet the necessary input criteria for the subsequent analysis. In one embodiment, the prediction module 116 is structured to categorize an individual's risk of prediabetes. The score is predicted using a machine learning model, notably the Deep Learning Model, which has shown an AUC of 0.87. In one embodiment, the clinical pathway module 118 is designed to provide a personalized protocol of subsequent actions for at least one patient, based on the determined risk thresholds.

[0050] In one embodiment, the risk prediction response module 120 presents the patient's risk score along with a clinical algorithm for the next best action based on the stratified risk of prediabetes. The risk prediction response module 120 provides an overview, including the patient's risk status (e.g., high, medium, or low), an Al-predicted score (e.g., 1 to 10), and a clinical algorithm with recommendations for lab investigation, nutrition, and diet, activity, referrals, treatment goals, educational materials, and revisit guidelines.

[0051] According to another exemplary embodiment of the invention, FIG. 2 refers to a schematic flow diagram 200 of the system 100 for predicting undiagnosed prediabetes and type-2 diabetes mellitus. At step 202, deanonymized and ethically sourced health data on individuals is gathered from six regions of Apollo Hospitals. At step 204, the collected data is cross-referenced with the discharge summaries of adult patients from previous hospitalizations to eliminate cases that meet the exclusion criteria. This includes HbAlc levels above 7.5%, uncontrolled and undiagnosed diabetes, gestational diabetes, patients with life-threatening or urgent conditions, and cases with altered physiological criteria. At step 206, the data hygiene for clinical features of health data is obtained according to patient discharge summaries.

[0052] At step 208, the health data collected is optimized using correlation matrices, multivariate odds ratios, and the deep learning model for feature selection, engineering, and importance ranking in their prediabetes risk prediction. The non-contributing clinical features are excluded. At step 210, the health data is processed by at least one deep learning model, which has demonstrated an Area Under the Curve (AUC) of 0.87. This model is then compared with the Korean national health and nutrition examination survey (KNHANES) model for prediabetes and diabetes prediction in the development cohort. At step 212, the risk factors and multivariate odds ratio are clinically validated.

[0053] The KNHANES stands for Korean National Health and Nutrition Examination Survey model for predicting prediabetes and diabetes. The KNHANES model exhibits an area under curve (AUC) of 0.712, 0.751 in KNHANES 2010, 2011 external validation. At step 214 involves the validation of the internal health data of 36k API calls from six regions of India. At step 216, these test cohort results are compared with the KNHANES model. The deep learning models, both with and without dyslipidemia, have consistently outperformed the KNHANES prediabetes prediction study (AUC 0.734) in terms of AUC in internal validation sets, recording AUCs of 0.77 and 0.76, respectively. In summary, the deep learning models, both with and without dyslipidemia, have shown superior performance when compared to the KNHANES model.

[0054] According to another exemplary embodiment of the invention, FIG. 3 refers to a flowchart 300 of a method for predicting the risk of undiagnosed prediabetes and type-2 diabetes mellitus using system 100. At step 302, the data collection module 122 collects the raw data obtained from various sources and the standardization and pre-processing module 124 processes the raw data to ensure system's processing requirements. At step 304, the pre-processed data is transmitted to the server 104 via the network 102 through an API integration module 126 and then transmitted to the input module 112 for further validation via the server 104.

[0055] At step 306, the input module 112 authenticates requests and validates input data as per the required format for at least one patient through the Application programminginterface (API) or through web user interface on a user device.At step 308, the processing module 114 transforms and standardizes data to meet the required input format for the model. At step 310, the prediction module 116 predicts the prediabetes risk of at least one individual through a machine learning model based on the patient's data analyzed through the processing module 114. At step 312, the clinical pathway module 118 provides clinical decision support for the next best action based on determined risk levels. Further, at step 314, the risk prediction response module 120 displays the risk status, i.e., high, medium, or low risk of prediabetes, and Al predicted score and provides a recommended protocol based on the risk threshold or risk status.

[0056] In accordance with another exemplary embodiment of the present invention, FIG. 4 provides a schematic depiction of the system architecture 400 for a method devised to develop and deploy the Al-based Prediabetes Risk Assessment System. The diagram highlights a sequence of operations that are integral to the system's development and deployment. As depicted in step 402, we have the data sourcing phase, where information is collected from various sources, including individual health checks from Electronic Medical Records (EMR) servers 414. Additional sources include clinical knowledge bases 416, like Standard Operating Protocols (SOPs), literature from research papers, journals, and consultations with clinicians 418.

[0057] In step 404, we move to the data ingestion phase. Here, data is collected and put into a centralized database 420. This data then traverses through a data pipeline 422 and is then stored in a data repository 424 in structured formats. This phase is crucial for organizing the data and preparing it for further pre-processing stages. In step 406, the data pre-processing phase encompasses metadata management 426, ETL processes 428, and data transformation and harmonization 430. These processes improve data quality and consistency, setting the stage for subsequent analysis and model training. In step 408, for data analysis related to prediabetes risk identification, we utilize a variety of statistical systems 408. Propensity matching 432 reveals hidden patterns, while descriptive statistics 434 provide a summarized overview of the data. Correlation coefficients 436 establish relationships between various risk factors, and odds / hazard ratios 438 compare the possibility of outcomes. KM plots and survival charts 440 visualize time-to-event data, andvisualization systems 442 simplify complex data representation. These systems, in concert, contribute to the feature selection and model building of the prediabetes risk assessment system.

[0058] In step 410, we move towards model development, employing a neural network model created with the TensorFlow library, specifically designed for predicting the likelihood of prediabetes development. By training the model with domain insights, data insights, and observing various iterations of the model's performance, we have arrived at the optimal performance. The model's structure is built of four layers, each with a specific number of neurons. The first layer has 256 neurons, followed by the second and third layers with 128 and 64 neurons, respectively, and finally a binary classification layer with one neuron. The activation functions'relu' and'sigmoid' are used in the first three layers and the last layer, respectively, effectively tuning the model for binary classification. In step 412, the deployment of the model for integration and usage through REST API 452 protocols involves using an API Management Service 444 and an application service resource 450 for model inference.

[0059] The model inference code is developed in Python programming language 454. All the resources are hosted on a virtual private network 448. The API service 446 acts as the interface for the REST API, facilitating communication between the API Management Service and the web application. The API Management Service includes storage capabilities 458 for securely managing and storing model artifacts, data inputs, and results. In one such application of the integration, this model inference API is integrated into a web application code 456 to be deployed and used as a web application 460. This web application serves as a user-friendly platform for users or clinicians, enabling them to input data and obtain prediabetes risk scores. Therefore, this phase makes the model operational and accessible, facilitating the prediction of prediabetes risk.

[0060] Numerous advantages of the present disclosure may be apparent from the discussion above. In accordance with the present disclosure, a system 100 for predicting the risk of undiagnosed prediabetes and type 2 diabetes mellitus is disclosed. The proposed invention provides a system 100 for predicting the individual risk of undiagnosedprediabetes and type-2 diabetes mellitus in at least one individual using at least one machine learning model.

[0061] The proposed invention provides the Al-based Prediabetes Risk Prediction System 100, which provides comprehensive and holistic risk assessment. The system 100 identifies individuals at risk for developing prediabetes and type 2 diabetes mellitus using a simple assessment system, such as a health risk assessment (HRA). The system 100 provides a clinical decision support system (CDSS) based on stratified risk to guide lab investigations, nutrition and diet plans, physical activity recommendations, referrals, treatment goals, educational materials, and revisit guidelines. The system 100 uses a machine learning model that can analyze complex interactions between numerous data points, like demographics, medical history, and lifestyle factors, providing a more comprehensive and personalized risk assessment compared to existing methods.

[0062] The system 100 can reduce healthcare costs associated with early prevention and management of type 2 diabetes mellitus. The system 100 delivers personalized risk assessment, optimizing resource allocation, eliminating unnecessary tests, and tailoring treatments according to the stratified risk of individuals. The system 100 is utilized as a community-based risk system for early prevention and diagnosis.

[0063] It will readily be apparent that numerous modifications and alterations can be made to the processes described in the foregoing examples without departing from the principles underlying the invention, and all such modifications and alterations are intended to be embraced by this application.

Claims

5. CLAIMS: l / We Claim:

1. An artificial intelligence (Al)-based system (100) designed to predict the risk of undiagnosed prediabetes and type-2 diabetes mellitus, comprising: a server (104) having a processor (106) and a memory (108), which is connected to a database (110) for efficient data management, wherein the server (104) is in communication with a user device via a network (102), the server (104) communicates with a plurality of modules, which is configured to predict the risk of clinical deterioration and mortality of at least one patient, wherein the plurality of modules comprises: client modules, which are configured to perform multiple functions for collecting and standardizing and transmitting the data forpredicting patient clinical deterioration, wherein the client modules comprises: a data collection module (122) configured to collect raw data from various sources such as electronic medical records (EMRs), Data repositories and Databases; a standardization and pre-processing module (124) configured to standardize and pre-process the collected data to ensure its integrity and compatibility with the system's processing requirements; and an application programming interface (API) module (126) configured to transmit the pre-processed data from the data collection module (122) and the standardization and pre-processing module (124) to the server (104) via the network (102); an input module (112) configured to enable authentication and validation of input data, ensuring compliance with required formats from various sources, wherein the input module (112) allows users to facilitate data related to age, gender, height, and weight to calculate body mass index (BMI), alcohol consumption, dietary pattern, physical activity, family history of diabetes mellitus, history of hypertension and dyslipidemia, past medical history, symptoms of diabetes mellitus, waist circumference,and body weight change over the previous six months, thereby collecting thedata of a patient, wherein the data collected by the input module (112) is transmitted to processing module; a processing module (114) configured to transform and standardize data from the input module (112) to meet the necessary input criteria for the subsequent analysis; a prediction module (116) configured to categorize an individual's risk of prediabetes, the score is predicted using a machine learning model, notably the deep learning model, which has shown an AUC of 0.87; a clinical pathway module (118) configured to provide a personalized protocol of subsequent actions for at least one individual, based on the determined risk thresholds; and a risk prediction response module (120) configured to provide a patient's risk score and a clinical algorithm for next steps based on their prediabetes risk level, wherein the risk prediction response module (120) is configured to: offer a comprehensive view of the patient's risk status, stratified as low, medium, and high risk, and predicted Al score, provide a clinical algorithm that suggests lab investigation, nutrition and diet, activity, referrals, treatment goals, educational materials, and revisit guidelines, thereby acting as a complementary system for clinicians, it enhances informed decision-making without replacing clinicians or diagnostic tests, and the artificial intelligence (Al)-based system (100) adheres to ISO 13485 standards, thereby ensuring patient safety and reliability. Functioning as a Software as a Medical Device (SaMD) and certified by ISO 13485, the artificial intelligence (Al)-based system (100) integrates algorithms for assessing the risk of prediabetes in individuals, thereby promoting early prevention and supporting clinicians,whereby the Al-based system (100) predicts the risk of prediabetes in individuals through the machine learning model.

2. The system (100) as claimed in claim 1, wherein the past medical history includes chronic diseases like cardiovascular, renal, liver, thyroid disease, stroke, transient ischemic attack, hypertension, and frequent infections.

3. The system (100) as claimed in claim 1, wherein the symptoms of diabetes mellitus such as weight gain or loss, polyuria, polydipsia, polyphagia, weakness or fatigue, blurred vision, recent skin or other infections, vulvovaginitis or balanitis, and abdominal pain.

4. The system (100) as claimed in claim 1, wherein the at least one machine learning model is used to predict the occurrence of prediabetes and type-2 diabetes mellitus, is a deep learning model, which shows an AUC of 0.87.

5. The Al-based system (100) for assessing prediabetes risk in a patient as claimed in claim 1, wherein the at least one risk category includes a low risk category, a moderate risk category, and a high risk category, and an Al-predicted score.

6. The Al-based system (100) for assessing prediabetes risk in a patient as claimed in claim 1, wherein the Al-based system (100) comprises a clinical pathway module (118) configured to provide a recommended protocol of next-based actions such as lab investigation, nutrition and diet, activity, referrals, treatment goals, educational materials, and revisit guidelines for at least one individual based on the determined risk thresholds.

7. The Al-based system (100) assessing prediabetes risk in a patient as claimed in claim 1, wherein the processor (106) is in communication with an application server (104) through a network (102).

8. A method for predicting undiagnosed prediabetes and type-2 diabetes mellitus using a system (100), comprising: collecting, by a data collection module (122) the raw data obtained from various sources, and processing the raw data by a standardization and pre-processing module (124) to ensure system's processing requirements;transmitting, the pre-processed data to a server (104) via a network (102) through an API integration module (126), and then transmitted to an input module (112) for further validation via the server (104); enabling, by an input module (112), the system to initiate authentication and validation of at least one individual's data, including personal parameters, medical history, and lifestyle factors; processing, by a processing module (114) to standardize the collected data to fit the necessary input criteria for subsequent analysis; predicting by a prediction module (116) and applying the deep learning model to categorize the patient's risk of having prediabetes; enabling a clinical pathway module (118) to provide clinical decision support upon determining the risk and enabling the creation of a personalized protocol of subsequent actions; and displaying, by a risk prediction response module (120) presents the patient's risk status, an Al predicted score, and a clinical algorithm suggesting next steps based on the stratified risk of prediabetes.

6. DATE AND SIGNATURE:Dated this 28thday of August, 2024PATENT AGENT NAME: VARALAKSHMI VANAMINPA - 328720

Images