Fundus Image Analysis for Cardiovascular Aging Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current CVD risk prediction methods using retinal images are inaccurate due to reliance on indirect measures and lack of identification of major contributors, leading to many false positives and negatives, and existing AI models fail to provide precise cardiovascular aging indicators.
Innovation Solution
A deep learning-based method for analyzing fundus images to determine relative cardiovascular aging (Cardiac BioAge) by comparing an individual's predicted CVD risk with a set of individuals in the same age group, using convolutional neural networks (CNNs) to identify risk factors like blood pressure, cholesterol, and glycaemic control, and providing personalized recommendations for management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional statistical methods with regression models are used for CVD risk prediction, then the method is simple and easy to implement, but the accuracy is low resulting in many false positives and false negatives
Solution Approach 1:
The patent replaces conventional statistical regression models with deep learning-based artificial intelligence models that directly analyze retinal fundus images. This substitution enables the system to capture complex non-linear relationships and patterns in retinal vasculature that traditional statistical methods cannot detect, thereby significantly improving CVD risk prediction accuracy while reducing false positives and negatives.
Solution Approach 2:
The patent introduces retinal fundus images as an intermediary biomarker that directly reflects cardiovascular health status. By using retinal vasculature characteristics as a mediator between observable features and CVD risk assessment, the system achieves more accurate predictions compared to conventional indirect statistical measures, while the deep learning model handles the complexity of image analysis automatically.
2Measurement precision
If AI deep learning models are used to estimate CVD risk from retinal images, then prediction accuracy improves, but the model complexity increases and requires extensive computational resources
Solution Approach 1:
The patent segments the CVD risk assessment process into multiple specialized deep learning models, each trained to detect specific cardiovascular risk factors from retinal images. This segmentation allows the system to handle different aspects of CVD risk (e.g., hypertension, hyperlipidemia, diabetes) separately, improving overall accuracy while making the computational task more manageable through modular architecture.
Solution Approach 2:
The patent performs preliminary processing and preprocessing of retinal fundus images before feeding them into the deep learning models. This includes image quality assessment, normalization, and feature extraction steps that prepare the data in advance, reducing the computational burden on the main prediction models and enabling more efficient processing while maintaining high accuracy.
3Loss of information
If conventional CVD risk equations are used, then the method is widely applicable and easy to operate, but it fails to identify major contributors to CVD risk such as blood pressure, cholesterol, and glycaemic control
Solution Approach 1:
The patent implements a feedback mechanism where the deep learning model not only predicts overall CVD risk but also provides breakdown information about contributions from different risk factors. The system analyzes retinal vasculature patterns associated with specific conditions (hypertension, hyperlipidemia, diabetes) and feeds this detailed information back to clinicians, enabling targeted interventions while maintaining ease of operation through automated analysis.
Solution Approach 2:
The patent utilizes color and texture variations in retinal fundus images as visual indicators of different cardiovascular risk factors. By analyzing subtle color changes and vascular patterns in the retinal image, the system can identify and quantify contributions from blood pressure, cholesterol, and glycaemic control issues, providing rich diagnostic information without requiring complex manual assessment procedures.
4Adaptability or versatility
If retinal images are analyzed for CVD risk prediction, then non-invasive assessment is achieved, but the ability to provide personalized health management recommendations is limited
Solution Approach 1:
The patent applies local quality analysis by examining specific regions and features within retinal fundus images that correspond to different cardiovascular risk factors. The deep learning model identifies localized vascular patterns, vessel caliber changes, and regional abnormalities that indicate specific conditions (e.g., hypertensive retinopathy in certain vascular territories), enabling personalized risk factor identification and tailored health management recommendations for each individual patient.
Data Source
AI summary
Systems and methods for determining one or more recommendations for management of wellbeing of an individual are disclosed. An indication of relative cardiovascular aging of the individual is determined based at least in part on determining an indication of relative cardiovascular aging of the first individual, based at least in part on a predicted risk of cardiovascular disease (CVD) of the individual determined by a deep learning model based on one or more fundus images; and determining the at least one recommendation for management of the individual's wellbeing based at least in part on the determined indication of relative cardiovascular aging of the first individual.