Design of Data Analytics Model for Health Care Surveillance Systems Using Machine Learning Techniques

Live healthcare monitoring has become a necessity to swiftly identify problematic conditions, track the outbreak, and conduct continuous evaluation of patients. The traditional monitoring systems usually have a setback of delayed reporting, manual interventions as well as disjointed data sources leading to slow clinical decision-making. The paper includes a single real-time data analytics model that combines the tools of machine learning (ML) and deep learning (DL) into the analysis of continuous health data streams, with high accuracy and minimum time lag. The proposed framework works with vital signs, symptom patterns and logs of events in a hospital by running them through a structured pipeline comprising of preprocessing, feature extraction, supervised ML classification and LSTM-based time-series prediction. The model was tested using a synthetic yet realistic dataset of 10,000 samples of patients. Empirical evidence shows that the ML model like the Random Forest and XGBoost have a high baseline performance, but the LSTM model has higher accuracy (97.8%) and ability to understand the time (temporal understanding). The real-time streaming architecture also facilitates fast inference of an average latency of 120 ms allowing immediate alerting of high-risk cases. The suggested system offers a scalable and cloud-edge deployable system that can be used in hospitals, population health surveillance, and telemedicine systems. This work explains that the application of ML/DL models with real-time analytics has the potential to empower health surveillance frameworks and aid in response to clinical decisions.