Hybrid Adaptive Threshold-Based Wearable System for Patient Vital Sign Monitoring and Motion Artifact Mitigation

Motion artifact from patient movement remains a critical barrier to reliable wearable photoplethysmography (PPG) monitoring in nursing practice, causing false physiological readings and alarm fatigue that compromise patient safety in remote and ambulatory care settings. This study aimed to develop and validate a wearable edge computing system integrating a Hybrid Adaptive Threshold algorithm for personalized motion artifact mitigation and a partial National Early Warning Score 2 (NEWS-2) module for automated nursing early warning across heterogeneous patient motion profiles. An experimental quantitative study was conducted on five healthy participants exposed to seven predefined motion scenarios (L0–L6). The prototype integrated a MAX30102 PPG sensor at the finger and an MPU-6050 IMU with an ESP32 microcontroller at the wrist. Performance was evaluated across four dimensions: static vital sign accuracy (MAPE against a commercial pulse oximeter), adaptive calibration under normal and tremor-simulated conditions, Data Gating effectiveness comparing Phase 1 (unfiltered) versus Phase 2 (filter active), and NEWS-2 scoring with MQTT transmission integrity. Static accuracy yielded a mean heart rate MAPE of 0.71% and SpO₂ MAPE of 0.31%. The Hybrid Adaptive Threshold algorithm successfully generated personalized gating boundaries for all participants, including tremor-simulated subjects. Data Gating reduced heart rate output standard deviation by 99.4% during aggressive arm swinging and 92.6% during finger tapping. NEWS-2 scoring executed with zero logical error across all six injection test cases, and MQTT payload integrity reached 100%. The proposed system demonstrated clinically acceptable accuracy, effective personalized motion gating, and reliable automated early warning capability, confirming its potential as a wearable nursing surveillance tool for ambulatory patient populations with heterogeneous motion profiles. Keywords : Photoplethysmography, Motion Artifact, Adaptive Threshold, Edge Computing, Wearable Sensor, NEWS-2.