Improving Biomedical Signal Search Results in Big Data Case-Based Reasoning Environments

Area of Science:

• Healthcare informatics
• Biomedical data analysis
• Time series analysis

Background:

• Time series subsequence matching is crucial for healthcare informatics tasks like diagnosis and trend discovery.
• Current medical systems rarely use subsequence matching due to high computational and memory demands.
• Existing methods face challenges in balancing efficiency with comprehensive search capabilities.

Purpose of the Study:

• To propose a novel randomized Monte Carlo sampling method for time series subsequence matching.
• To address the computational and memory complexities associated with traditional subsequence matching algorithms.
• To enhance the efficiency and effectiveness of subsequence matching in healthcare informatics.

Main Methods:

• Implemented a randomized Monte Carlo sampling approach.
• Integrated the method with R-NN indexing to broaden search criteria.
• Evaluated the method's performance in terms of computational and memory complexity.

Main Results:

• The proposed method broadens search criteria with minimal increases in computational and memory complexities compared to R-NN indexing.
• Information gain is improved, and result sets approximate the theoretical result space.
• Query results increased by several orders of magnitude, with improved recall and no significant degradation in precision.

Conclusions:

• The randomized Monte Carlo sampling method offers an efficient and effective solution for time series subsequence matching in healthcare.
• This approach significantly enhances the utility of subsequence matching in medical systems.
• The method provides a scalable and precise tool for analyzing patient data and identifying trends.