Regressive Machine Learning for Real-Time Monitoring of Bed-Based Patients

• DOI: 10.20944/preprints202409.1279.v2
• Publication Date: 2024-10-18T01:24:21Z
• Authors: Paul Joseph; Husnain Ali; Daniel Matthew; Anvin Thomas; Rejath Jose; Jonathan Mayer; Molly Bekbolatova; Timothy Devine; Milan Toma
• Keywords: Technology; model; QH301-705.5; automated detection; T; Physics; QC1-999; ensemble; Engineering (General). Civil engineering (General); Chemistry; machine learning; regressor; patient safety; TA1-2040; Biology (General); QD1-999

This study introduces an ensemble model designed for real-time monitoring of bedridden patients.The was developed using a unique dataset, specifically acquired this study, that captures six typical movements. The dataset balanced the Synthetic Minority Over-sampling Technique, resulting in diverse distribution movement types. Three models were evaluated: Decision Tree Regressor, Gradient Boosting and Bagging Regressor. Regressor achieved accuracy 0.892 R2 score 1.0 on training 0.939 test dataset. 0.908 0.99 0.943 selected due to its superior performance trade-offs such as computational cost scalability. It 0.950, 0.996 data, 0.959 data. also employs K-Fold cross-validation learning curves validate robustness model. proposed system addresses practical implementation challenges monitoring, data latency false positives/negatives, is seamless integration with hospital IT infrastructure. research demonstrates potential machine enhance patient safety healthcare settings.