Multimodal Activity Tracking via Bagged Formal Concept Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current activity tracking systems are inadequate in automatically characterizing daily activities from diverse and semantically rich multimodal data streams, leading to inaccurate and tedious user-initiated food journaling, which results in low adherence and poor health monitoring.
Innovation Solution
The system employs Bagging Formal Concept Analysis (BFCA) to synchronize and segment multimodal inputs from various devices, recognizing activities through classifiers trained on random attribute subsets, and uses event-triggered mechanisms to prompt users for food journal entries via voice commands, leveraging heart rate and activity patterns to identify eating moments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If user-initiated food journaling is used, then food consumption tracking can be performed, but user adherence is low due to tedious manual intervention
Solution Approach 1:
The system automatically tracks food consumption by analyzing sensor data from wearables (heart rate, activity levels) and smartphone data (location, time) without requiring users to manually log meals. The system serves itself by autonomously inferring eating events from multimodal data patterns, eliminating the need for user-initiated journaling while maintaining tracking accuracy
Solution Approach 2:
The patent replaces the mechanical system of manual food logging with an automated computational system that uses machine learning algorithms to analyze sensor data patterns. The system substitutes physical user actions (writing down meals) with automated data processing and pattern recognition, transforming the tracking mechanism from user-dependent to system-autonomous
2Quantity of substance
If manual food journaling is implemented, then food tracking data can be collected, but the process is tedious and time-consuming
Solution Approach 1:
The system continuously collects and pre-processes sensor data in the background before eating events occur. By maintaining a ready state with pre-processed activity, heart rate, and location data, the system can immediately infer food consumption events without requiring users to spend time manually logging meals after eating
Solution Approach 2:
The system performs continuous automated tracking of food consumption by constantly analyzing sensor data streams. Instead of requiring discrete user actions at specific times, the system maintains continuous monitoring and automatically processes eating events as they occur, eliminating time loss while maintaining comprehensive data collection
3Extent of automation
If automated activity tracking is implemented, then user intervention is reduced, but accuracy in recognizing diverse activities may deteriorate
Solution Approach 1:
The system segments the complex task of activity recognition into multiple specialized classifiers, each trained on specific subsets of sensor data (accelerometer, gyroscope, heart rate, location). By dividing the recognition problem into manageable segments and combining their results, the system achieves high automation while maintaining accuracy across diverse activity types
Solution Approach 2:
The patent combines multiple data sources (sensor readings, location data, time information) into a composite feature vector for activity recognition. This composite approach integrates information from different modalities, allowing the automated system to achieve high accuracy by leveraging the complementary strengths of each data source rather than relying on a single sensor
Data Source
AI summary
Systems and methods for tracking activities from a plurality of multimodal inputs are described. Activity tracking can include receiving a plurality of multimodal inputs, synchronizing the plurality of multimodal inputs, generating segments from the synchronized multimodal inputs, recognizing activities associated with each generated segment by performing a bagged formal concept analysis (BFCA), and recording the recognized activities in a storage. Tracking of activities can include the detection of moments (e.g., eating moments), during which an activity tracking application can prompt a user for information (e.g., a food journal).