User-Trained Health Monitoring Model Using Skin Pattern Reflection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing portable devices for monitoring human health require adjustment of hardware and software for each use case, leading to inflexibility and a lack of personalization, as they rely on data-driven models trained with historic datasets that do not account for individual user deviations.
Innovation Solution
A method involving a device that uses a projector to illuminate a body part with a light pattern containing pattern features, a camera to capture images, and a processor to determine skin pattern features and train a data-driven model using user-selected parameter values, allowing for personalized monitoring and analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data-driven models are trained with historic datasets according to a specific use case, then the model can determine health parameters, but the model works identically for all users and does not account for personal deviations
Solution Approach 1:
The system enables users to train their own personalized data-driven models by collecting and processing their own image data. Users actively participate in the model training process by providing images of their body parts under various lighting conditions, allowing the system to adapt to their individual characteristics such as skin color, age, and health conditions without requiring manual configuration by developers
Solution Approach 2:
The system performs preliminary data collection and model training for each user before actual health monitoring begins. Users complete an onboarding process where they provide training images in advance, allowing personalized models to be prepared and ready for accurate health parameter determination from the start of usage
2Adaptability or versatility
If the provider collects historic datasets and trains models for each use case, then monitoring functionality is available, but hardware and software must be adjusted for each use case reducing flexibility
Solution Approach 1:
The system uses a universal image processing approach that can determine multiple different health parameters (blood oxygen saturation, blood pressure, heart rate, etc.) from the same type of image data. A single camera and lighting setup can support various health monitoring use cases by training different models on the same hardware platform, eliminating the need for use case-specific hardware adjustments
Solution Approach 2:
The system replaces physical hardware adjustments with software-based model training. Instead of modifying hardware components for different health parameters, the invention uses software models that can be trained and retrained to recognize different physiological indicators from the same image input, providing flexibility through computational methods rather than mechanical changes
3Quantity of substance
If all pattern features from images are used for training, then comprehensive data is collected, but disturbances like glasses or hats require dedicated training datasets
Solution Approach 1:
The system extracts and isolates only the relevant skin pattern features from images by filtering out unrelated elements. Using light reflection analysis, the system identifies and extracts pattern features that are specifically reflected by skin tissue, automatically excluding disturbances such as glasses, hats, or other non-skin objects without requiring separate training datasets for each disturbance type
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach provides greater flexibility and personalization in monitoring human health, as it allows users to set up their own monitoring use cases and accounts for individual characteristics, such as skin color and health conditions, improving reliability and accuracy.
Implementation Method 1
receiving an image of a body part while the body part is illuminated by a light pattern containing at least one pattern feature
Implementation Method 2
determining skin pattern features from the image, wherein a skin pattern feature is a pattern feature which has been reflected by skin
Data Source
AI summary
Disclosed herein is a method for training a data-driven model for determining a user-selected parameter value related to the condition of a human including:(a) receiving an image of a body part of the human while the body part is illuminated by a light pattern containing at least one pattern feature,(b) determining skin pattern features from the image, where a skin pattern feature is a pattern feature which has been reflected by skin,(c) receiving from a user interface a user-selected parameter value related to the condition of the human, and(d) training a data-driven model with a training dataset comprising the skin pattern features and the user-selected parameter value.