Fundus Image Embedding for Faster Neurodevelopmental Disorder Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional diagnostic methods for neurodevelopmental disorders like autism spectrum disorder (ASD) and attention-deficit/hyperactivity disorder (ADHD) are labor-intensive and subjective, necessitating a more efficient and objective approach for early and precise diagnosis.
Innovation Solution
A machine learning model trained on fundus images to classify retinal features, using embedding vectors and similarity measures to predict the presence of neurodevelopmental disorders, utilizing a processor and memory unit to analyze fundus images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional diagnostic methods (behavioral assessments, clinical observations, structured interviews) are used for neurodevelopmental disorders, then diagnostic accuracy can be achieved, but the process becomes labor-intensive and time-consuming
Solution Approach 1:
The patent replaces manual clinical assessment methods with an automated machine learning system that processes fundus images. The mechanical system of behavioral observations and structured interviews is substituted with an algorithmic system that extracts features from retinal images and generates diagnostic predictions, thereby eliminating time loss while maintaining measurement precision.
Solution Approach 2:
The patent introduces fundus images as an intermediary medium for diagnosis. Instead of directly observing behavioral characteristics, the system uses retinal images as an intermediate representation that contains biomarkers correlated with neurodevelopmental disorders. This intermediary enables automated analysis while preserving diagnostic accuracy.
2Measurement precision
If conventional diagnostic methods are used for neurodevelopmental disorders, then comprehensive assessment can be performed, but the process becomes labor-intensive and subject to judgment variations
Solution Approach 1:
The patent replaces subjective human judgment with an objective machine learning system. The complex process of clinical observation and interpretation by practitioners is substituted with automated feature extraction and classification algorithms that consistently apply the same diagnostic criteria, eliminating judgment variations while managing system complexity through modular architecture.
Solution Approach 2:
The patent transforms the diagnostic process from qualitative behavioral assessment to quantitative image analysis. By converting clinical observations into measurable parameters from fundus images (such as vascular patterns, optic nerve characteristics), the system achieves objective diagnosis while the complexity is managed through standardized parameter extraction procedures.
3Measurement precision
If machine learning models are trained on diverse fundus images from multiple subject groups, then classification accuracy improves, but training data requirements and processing complexity increase
Solution Approach 1:
The patent creates a universal machine learning model that can diagnose multiple neurodevelopmental disorders (autism spectrum disorder, attention-deficit/hhyperactivity disorder, and other conditions) using a single trained system. The model learns universal features from diverse fundus images that are applicable across different disorder types, reducing the need for separate training datasets for each condition while maintaining high classification accuracy.
Data Source
AI summary
Provided are apparatuses, a non-transitory computer-readable medium or media, for predicting of neurodevelopmental disorder using a fundus image of a subject. In certain aspects, disclosed a method including the steps of: receiving the fundus image; processing the fundus image to classify one or more features contained in the fundus image using a machine learning model; and predicting, based on a classification, whether the fundus image is indicative of presence of neurodevelopmental disorder in the subject, wherein the machine learning model has been trained on stored fundus images obtained from at least two group of subjects who have previously been diagnosed as having the neurodevelopmental disorder, the machine learning model is trained by processes comprising; extracting a first embedding vector and a second embedding vector from a first fundus image and a second fundus image, respectively; and maximizing a similarity between the first embedding vector and the second embedding vector.