Mobile Medication Identification via Image Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current medication tracking systems are inefficient and inaccessible, particularly for patients taking multiple medications, as they require specialized sensors for each type of medication application device, and cannot account for all types of medications, including inhalers and traditional pill forms.
Innovation Solution
A mobile device with an image sensor that captures medication images, uses a machine learning engine to identify the medication, and records dosage inputs, allowing for efficient and accessible tracking without the need for specialized sensors, and can transmit data to a server for patient record updates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If specialized sensors are used for each type of medication application device, then medication tracking accuracy is improved, but device complexity and ease of operation deteriorate
Solution Approach 1:
The patent applies universality by using a single image sensor and machine learning system to identify and track multiple types of medications (inhalers, pills, other medicament devices) without requiring specialized sensors for each device type. The machine learning model is trained on diverse medication images to enable universal recognition across different medication forms.
Solution Approach 2:
The patent replaces the mechanical/sensor-based tracking system with an optical system (image sensor/camera) combined with machine learning. Instead of using specialized sensors that attach to each medication device, the system uses image capture and computational analysis to identify and track medication usage across all device types.
2Reliability
If specialized sensors are required for each medication type, then tracking reliability is improved, but ease of operation and adaptability worsen
Solution Approach 1:
The system provides a single unified application that can track all medication types through image recognition, eliminating the need for users to select specific sensors or download different applications for different medications. The machine learning model automatically identifies the medication type from the captured image.
Solution Approach 2:
The system performs automatic medication identification and tracking without requiring user intervention to select sensor types or configure device-specific settings. The machine learning model automatically processes the captured image to identify the medication and record the dosage event.
3Adaptability or versatility
If comprehensive medication coverage is achieved, then adaptability is improved, but device complexity increases
Solution Approach 1:
The patent achieves comprehensive medication coverage by designing a universal image recognition system that can identify inhalers, oral medications, and other medicament devices. The machine learning model is trained on diverse medication images to enable recognition across all these different medication forms using a single system.
Solution Approach 2:
The patent introduces a machine learning model as an intermediary between the image sensor and the tracking system. This intermediary processes the captured images and automatically identifies medication types, bridging the gap between a single sensor type and multiple medication forms without requiring complex device-specific configurations.
4Measurement precision
If manual dosage recording is required, then measurement precision is improved, but ease of operation and time efficiency worsen
Solution Approach 1:
The system performs automatic dosage recording by capturing an image of the medication and using the machine learning model to identify it. The system then automatically records the dosage event with timestamp and medication details, eliminating the need for manual entry while maintaining accuracy through the structured data collection process.
Data Source
AI summary
An application for the identification of medication and the recording of a dosage of such medication is disclosed. An image of an unidentified medicine, such as self-administered medication or a medicament device, is captured from an image sensor on a mobile computing device. The type of medicine is determined based on the captured image. An input interface is displayed to accept a dosage input of the determined medicine. The dosage input of the determined medicine is recorded.