Description:BACKGROUND
Field of Invention
[001] Embodiments of the present invention generally relate to a machine learning system and particularly to a federated learning system for progression prediction relating to diabetic retinopathy (DR).
Description of Related Art
[002] Diabetic Retinopathy (DR) is a leading cause of vision impairment worldwide, primarily affecting individuals with long-term diabetes. The disease progresses silently, with early stages often remaining undetected until significant damage has occurred. Timely and accurate diagnosis is critical for preventing severe complications, yet current screening methods are either resource-intensive or inaccessible to large populations. Conventional diagnostic approaches rely heavily on retinal imaging, requiring trained specialists for evaluation. However, access to such expertise is limited in many regions, leading to delayed diagnoses and suboptimal patient outcomes. Additionally, the integration of electronic health records (EHRs) with diagnostic tools remains a challenge, restricting the ability to track long-term disease progression effectively.
[003] Artificial intelligence (AI) has emerged as a promising solution for automated DR detection, leveraging deep learning techniques to analyze retinal images. Several AI-based systems, such as IDx-DR, EyeArt, and Google's ARDA, have demonstrated high accuracy in identifying DR from retinal scans. Additionally, telemedicine-based solutions like RetinaVue have enabled remote screening, expanding access to care. However, most of these systems operate as standalone models that primarily rely on static image-based analysis. They fail to incorporate a patient's longitudinal health history, such as HbA1c levels, blood pressure trends, and kidney function over time, which are crucial indicators of DR progression. Furthermore, existing AI models often require centralized data collection, raising privacy concerns and compliance issues in healthcare institutions.
[004] Federated learning has been introduced to address data privacy challenges by allowing decentralized model training across multiple healthcare facilities. Solutions like NVIDIA Clara FL have made strides in this area, yet their application to DR prediction remains in its early stages. Current federated learning models lack adaptability, failing to update dynamically with evolving patient data and medical advancements. This limitation reduces predictive accuracy and restricts personalized treatment recommendations. The need for a more adaptive, privacy-preserving AI system that integrates longitudinal EHR data with retinal imaging remains a crucial gap in existing DR prediction methodologies.
[005] There is thus a need for an improved and advanced federated learning system for progression prediction relating to diabetic retinopathy (DR) that can administer the aforementioned limitations in a more efficient manner.
SUMMARY
[006] Embodiments in accordance with the present invention provide a federated learning system for progression prediction relating to diabetic retinopathy (DR). The system comprising a local hospital training node adapted to receive the retinal images and longitudinal electronic health records (EHR) data from a computing device. The system further comprising a processing unit in communication with the local hospital training node. The processing unit is configured to normalize, align, and handle, using a preprocessing engine, missing values in the retinal images and the EHR data; federate learning, using the local hospital training node, to train deep learning models locally without transferring the retinal images and the EHR data; aggregate encrypted model, using a central federated server, updates from multiple hospital nodes to improve prediction accuracy while preserving privacy; integrate, using an EHR-Image fusion engine, the retinal images and the EHR data for adaptive learning and disease progression prediction; and provide, using an output engine, explainable risk classification and predictive insights to medical practitioners.
[007] Embodiments in accordance with the present invention further provide a method for providing progression prediction relating to diabetic retinopathy (DR) in retinal images. The method comprising steps of receiving the retinal images and longitudinal electronic health records (EHR) data from a computing device; normalizing, aligning, and handling, using a preprocessing engine, missing values in the retinal images and the EHR data; federating learning, using a local hospital training node, to train deep learning models locally without transferring the retinal images and the EHR data; aggregating encrypted model, using a central federated server, updates from multiple hospital nodes to improve prediction accuracy while preserving privacy; integrating, using an EHR-Image fusion engine, the retinal images and the EHR data for adaptive learning and disease progression prediction; and providing, using an output engine, explainable risk classification and predictive insights to medical practitioners.
[008] Embodiments of the present invention may provide a number of advantages depending on their particular configuration. First, embodiments of the present application may provide a federated learning system for progression prediction relating to diabetic retinopathy (DR)
[009] Next, embodiments of the present application may provide a system for providing prediction for diabetic retinopathy (DR) that leverages federated learning to train AI models locally without sharing raw patient data, ensuring compliance with healthcare privacy regulations and protecting sensitive medical information.
[0010] Next, embodiments of the present application may provide a system for providing prediction for diabetic retinopathy (DR) that incorporates long-term patient health records, such as HbA1c levels, blood pressure trends, and kidney function, leading to more accurate disease progression prediction.
[0011] Next, embodiments of the present application may provide a system for providing prediction for diabetic retinopathy (DR) that dynamically updates its AI models with new patient data and evolving medical knowledge, improving prediction accuracy over time and adapting to changes in disease patterns.
[0012] Next, embodiments of the present application may provide a system for providing prediction for diabetic retinopathy (DR) that provides early warnings for high-risk individuals, enabling timely medical interventions and reducing the risk of vision loss.
[0013] Next, embodiments of the present application may provide a system for providing prediction for diabetic retinopathy (DR) that is more efficient and scalable for deployment across multiple healthcare facilities without extensive infrastructure costs.
[0014] These and other advantages will be apparent from the present application of the embodiments described herein.
[0015] The preceding is a simplified summary to provide an understanding of some embodiments of the present invention. This summary is neither an extensive nor exhaustive overview of the present invention and its various embodiments. The summary presents selected concepts of the embodiments of the present invention in a simplified form as an introduction to the more detailed description presented below. As will be appreciated, other embodiments of the present invention are possible utilizing, alone or in combination, one or more of the features set forth above or described in detail below.
BRIEF DESCRIPTION OF THE DRAWINGS
[0016] The above and still further features and advantages of embodiments of the present invention will become apparent upon consideration of the following detailed description of embodiments thereof, especially when taken in conjunction with the accompanying drawings, and wherein:
[0017] FIG. 1 illustrates a schematic block diagram of a federated learning system for progression prediction relating to diabetic retinopathy (DR), according to an embodiment of the present invention;
[0018] FIG. 2 illustrates a block diagram of a processing unit of the federated learning system for progression prediction relating to diabetic retinopathy (DR), according to an embodiment of the present invention; and
[0019] FIG. 3 depicts a flowchart of a method for providing progression prediction relating to diabetic retinopathy (DR) in retinal images, according to an embodiment of the present invention.
[0020] The headings used herein are for organizational purposes only and are not meant to be used to limit the scope of the description or the claims. As used throughout this application, the word "may" is used in a permissive sense (i.e., meaning having the potential to), rather than the mandatory sense (i.e., meaning must). Similarly, the words “include”, “including”, and “includes” mean including but not limited to. To facilitate understanding, like reference numerals have been used, where possible, to designate like elements common to the figures. Optional portions of the figures may be illustrated using dashed or dotted lines, unless the context of usage indicates otherwise.
DETAILED DESCRIPTION
[0021] The following description includes the preferred best mode of one embodiment of the present invention. It will be clear from this description of the invention that the invention is not limited to these illustrated embodiments but that the invention also includes a variety of modifications and embodiments thereto. Therefore, the present description should be seen as illustrative and not limiting. While the invention is susceptible to various modifications and alternative constructions, it should be understood, that there is no intention to limit the invention to the specific form disclosed, but, on the contrary, the invention is to cover all modifications, alternative constructions, and equivalents falling within the scope of the invention as defined in the claims.
[0022] In any embodiment described herein, the open-ended terms "comprising", "comprises”, and the like (which are synonymous with "including", "having” and "characterized by") may be replaced by the respective partially closed phrases "consisting essentially of", “consists essentially of", and the like or the respective closed phrases "consisting of", "consists of”, the like.
[0023] As used herein, the singular forms “a”, “an”, and “the” designate both the singular and the plural, unless expressly stated to designate the singular only.
[0024] FIG. 1 illustrates a schematic block diagram of a federated learning system 100 (hereinafter referred to as the system 100) for progression prediction relating to diabetic retinopathy (DR), according to an embodiment of the present invention. The system 100 may be adapted to receive the retinal images and longitudinal electronic health records (EHR) data. Further, the system 100 may be adapted to detect a presence of the diabetic retinopathy (DR) in the received retinal images. Moreover, the system 100 may train an artificially computable model for adaptive learning and disease progression prediction. Further, the training may be driven by real-time updates based on emerging medical trends and patient health records. The system 100 may utilize advanced feature extraction techniques to analyze progressive changes in a retina in correlation with patient metabolic health data.
[0025] According to the embodiments of the present invention, the system 100 may incorporate non-limiting hardware components to enhance the processing speed and efficiency such as the system 100 may comprise a computing device 102, a local hospital training node 104, a processing unit 106, a preprocessing engine 108, a central federated server 110, an EHR-Image fusion engine 112, an output engine 114, deep learning models 116, and an encrypted model 118. In an embodiment of the present invention, the hardware components of the system 100 may be integrated with computer-executable instructions for overcoming the challenges and the limitations of the existing systems.
[0026] In an embodiment of the present invention, the computing device 102 may be adapted to upload the retinal images and the EHR data to the system 100. The EHR data may be, but not limited to, blood glucose levels, blood pressure, kidney function data, and so forth. Embodiments of the present invention are intended to include or otherwise cover any physiological data that may be comparted in the EHR data, including known, related art, and/or later developed technologies. The computing device 102 may be, but not limited to, a laptop, a mobile, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the computing device 102, including known, related art, and/or later developed technologies.
[0027] In an embodiment of the present invention, the local hospital training node 104 may be adapted to receive the retinal images and the EHR data from the computing device 102. In an embodiment of the present invention, the local hospital training node 104 may be installed singularly and centralized in an hospital. In another embodiment of the present invention, the local hospital training node 104 may be installed in plurality and distributed in the hospital. Further, the local hospital training node 104 may be trained using deep learning techniques without requiring data sharing, ensuring compliance with data protection regulations.
[0028] In an embodiment of the present invention, the processing unit 106 may be in communication with the local hospital training node 104. The processing unit 106 may further be configured to execute computer-executable instructions to generate an output relating to the system 100. According to embodiments of the present invention, the processing unit 106 may be, but not limited to, a Programmable Logic Control (PLC) unit, a microprocessor, a development board, and so forth. Embodiments of the present invention are intended to include or otherwise cover any type of the processing unit 106 including known, related art, and/or later developed technologies. In an embodiment of the present invention, the processing unit 106 may further be explained in conjunction with FIG. 2.
[0029] FIG. 2 illustrates a block diagram of the processing unit 106 of the system 100, according to an embodiment of the present invention. The processing unit 106 may comprise the computer-executable instructions in form of programming modules such as a data alignment module 200, a data learning module 202, a data aggregating module 204, a data integration module 206, and a data output module 208.
[0030] In an embodiment of the present invention, the data alignment module 200 may be configured to activate the preprocessing engine 108 to normalize, align, and handle missing values in the retinal images and the EHR data. The preprocessing engine 108 may be configured to correct image distortions and normalize EHR records to enhance feature extraction for diabetic retinopathy progression prediction. The data alignment module 200 may be configured to transmit a first activation signal to the data learning module 202.
[0031] The data learning module 202 may be activated upon receipt of the first activation signal from the data alignment module 200. The data learning module 202 may be configured to activate the local hospital training node 104 to federate learning. The federated learning may train the deep learning models 116 locally without transferring the retinal images and the EHR data. The federated learning using the local hospital training node 104 may be configured to enable privacy preservation by transmitting the encrypted model 118 updates instead of raw patient data. The data learning module 202 may be configured to transmit a second activation signal to the data aggregating module 204.
[0032] The data aggregating module 204 may be activated upon receipt of the second activation signal from the data learning module 202. The data aggregating module 204 may be configured to activate the central federated server 110 to aggregate the encrypted model 118. The central federated server 110 may continuously update a global AI model by aggregating local model parameters from multiple hospitals. The encrypted model 118 may get updates from multiple hospital nodes to improve prediction accuracy while preserving privacy. The data aggregating module 204 may be configured to integrate retinal imaging trends with blood sugar fluctuations, HbA1c levels, and kidney function deterioration to enhance prediction accuracy. The data aggregating module 204 may be configured to transmit a third activation signal to the data integration module 206.
[0033] The data integration module 206 may be activated upon receipt of the third activation signal from the data aggregating module 204. The data integration module 206 may be configured to activate the EHR-Image fusion engine 112 to integrate the retinal images and the EHR data for the adaptive learning and the disease progression prediction. The data integration module 206 may be configured to transmit a fourth activation signal to the data output module 208.
[0034] The data output module 208 may be activated upon receipt of the fourth activation signal from the data integration module 206. The data output module 208 may be configured to activate the output engine 114 to provide explainable risk classification and predictive insights to medical practitioners. The output engine 114 may provide an explainable AI-based risk classification that helps doctors take timely preventive actions against blindness.
[0035] In an exemplary embodiment of the present invention, Hospital X, a leading multi-specialty hospital, may integrate the federated learning system 100 to enhance predictive capabilities for diabetic retinopathy (DR). When a diabetic patient undergoes a routine retinal screening, high-resolution retinal images may be captured, and relevant EHR data such as blood glucose levels, HbA1c levels, blood pressure, and kidney function data may be collected. These may be uploaded via the computing device 102 to the system 100. The local hospital training node 104 may process the retinal images and EHR data, with the preprocessing engine 108 normalizing and aligning the data to ensure accuracy.
[0036] The data learning module 202 may activate federated learning to allow AI training locally without transferring sensitive patient data. Instead, updates from the encrypted model 118 may be shared with the central federated server 110, which may aggregate updates from multiple hospitals to improve predictive accuracy. The EHR-Image fusion engine 112 may integrate retinal imaging trends with metabolic health data, enabling adaptive learning and disease progression prediction. The output engine 114 may generate risk classifications and predictive insights for informing medical practitioners about high-risk DR progression and suggesting timely interventions. This AI-driven approach may enhance early detection, ensure data privacy, and may aid in timely medical decision-making for significantly improving patient outcomes at Hospital X.
[0037] FIG. 3 depicts a flowchart of a method 300 for providing progression prediction relating to diabetic retinopathy (DR) in the retinal images, according to an embodiment of the present invention.
[0038] At step 302, the system 100 may receive the retinal images and the EHR data from the computing device 102.
[0039] At step 304, the system 100 may normalize, align, and handle the missing values in the retinal images and the EHR data.
[0040] At step 306, the system 100 may federate learning to train the deep learning models 116 locally without transferring the retinal images and the EHR data.
[0041] At step 308, the system 100 may aggregate the encrypted model 118 updates from the multiple hospital nodes to improve the prediction accuracy while preserving privacy.
[0042] At step 310, the system 100 may integrate the retinal images and the EHR data for adaptive learning and the disease progression prediction.
[0043] At step 312, the system 100 may provide the explainable risk classification and the predictive insights to the medical practitioners
[0044] While the invention has been described in connection with what is presently considered to be the most practical and various embodiments, it is to be understood that the invention is not to be limited to the disclosed embodiments, but on the contrary, is intended to cover various modifications and equivalent arrangements included within the scope of the appended claims.
[0045] This written description uses examples to disclose the invention, including the best mode, and also to enable any person skilled in the art to practice the invention, including making and using any devices or systems and performing any incorporated methods. The patentable scope of the invention is defined in the claims, and may include other examples that occur to those skilled in the art. Such other examples are intended to be within the scope of the claims if they have structural elements that do not differ from the literal language of the claims, or if they include equivalent structural elements within substantial differences from the literal languages of the claims. , Claims:CLAIMS
I/We Claim:
1. A federated learning system (100) for progression prediction relating to diabetic retinopathy (DR), characterized in that the system (100) comprising:
a local hospital training node (104) adapted to receive the retinal images and longitudinal electronic health records (EHR) data from a computing device (102); and
a processing unit (106) in communication with the local hospital training node (104), wherein the processing unit (106) is configured to:
normalize, align, and handle, using a preprocessing engine (108), missing values in the retinal images and the EHR data;
federate learning, using the local hospital training node (104), to train deep learning models (116) locally without transferring the retinal images and the EHR data;
aggregate encrypted model (118), using a central federated server (110), updates from multiple hospital nodes to improve prediction accuracy while preserving privacy;
integrate, using an EHR-Image fusion engine (112), the retinal images and the EHR data for adaptive learning and disease progression prediction; and
provide, using an output engine (114), explainable risk classification and predictive insights to medical practitioners.
2. The system (100) as claimed in claim 1, wherein the federated learning using the local hospital training node (104) is configured to enable privacy preservation by transmitting encrypted model (118) updates instead of raw patient data.
3. The system (100) as claimed in claim 1, wherein the preprocessing engine (108) is configured to correct image distortions and normalize EHR records to enhance feature extraction for diabetic retinopathy progression prediction.
4. The system (100) as claimed in claim 1, wherein the local hospital training node (104) is trained using deep learning techniques without requiring data sharing, ensuring compliance with data protection regulations.
5. The system (100) as claimed in claim 1, the central federated server (110) continuously updates a global AI model by aggregating local model parameters from multiple hospitals.
6. The system (100) as claimed in claim 1, wherein the processing unit (106) is configured to integrate retinal imaging trends with blood sugar fluctuations, HbA1c levels, and kidney function deterioration to enhance prediction accuracy.
7. The system (100) as claimed in claim 1, wherein the EHR data is selected from blood glucose levels, blood pressure, kidney function data, or a combination thereof.
8. The system (100) as claimed in claim 1, wherein the processing unit (106) is configured to allow real-time updates based on emerging medical trends and patient health records.
9. The system (100) as claimed in claim 1, wherein the processing unit (106) is configured to utilize advanced feature extraction techniques to analyze progressive changes in a retina in correlation with patient metabolic health data.
10. A method (300) for providing progression prediction relating to diabetic retinopathy (DR) in retinal images, the method (300) is characterized by steps of:
receiving the retinal images and longitudinal electronic health records (EHR) data from a computing device (102);
normalizing, aligning, and handling, using a preprocessing engine (108), missing values in the retinal images and the EHR data;
federating learning, using a local hospital training node (104), to train deep learning models (116) locally without transferring the retinal images and the EHR data;
aggregating encrypted model (118), using a central federated server (110), updates from multiple hospital nodes to improve prediction accuracy while preserving privacy;
integrating, using an EHR-Image fusion engine (112), the retinal images and the EHR data for adaptive learning and disease progression prediction; and
providing, using an output engine (114), explainable risk classification and predictive insights to medical practitioners.
Date: March 26, 2025
Place: Noida

Nainsi Rastogi
Patent Agent (IN/PA-2372)
Agent for the Applicant