An Al-Enabled Health Monitoring System for Early Diagnosis and Real-Time Vital Tracking
Keywords:
Health Monitoring, Early Diagnosis, Vital Tracking, Healthcare, Artificial IntelligenceAbstract
The increasing demand for timely and accurate healthcare delivery has driven the adoption of intelligent digital technologies for patient monitoring and medical diagnosis. This study develops an AI-enabled health monitoring system for early diagnosis and real-time vital tracking using a CNN-LSTM hybrid model. The methodology combined primary sensor data from 120 participants with public datasets (MIMIC-III, PhysioNet, Kaggle), totalling 76,495 patient episodes. The model was trained on multivariate time-series data (heart rate, blood pressure, temperature, SpO₂) to detect physiological anomalies. Five-fold cross‑validation yielded accuracy of 94.6%, precision of 92.3%, recall of 93.8%, F1‑score of 93.0%, and AUC‑ROC of 97.2%, significantly outperforming standalone CNN, LSTM, Random Forest, XGBoost, and Transformer models (p < 0.05). The system was deployed in a web environment with real-time alerts and visualisation, achieving a 1.2‑second average response time. The findings confirm that AI‑based monitoring enhances early diagnosis and continuous care, with future work focusing on wearable IoT integration and cloud analytics.
