HEALTHGUARD: An Unsupervised Learning Framework for Early Detection of Anomalies in Healthcare Data

This research presents HEALTHGUARD, an unsupervised machine learning framework developed for the early detection of anomalies in healthcare data. The system integrates Isolation Forest and Autoencoder models to identify abnormal patient patterns without requiring labeled datasets. The framework analyzes healthcare parameters and categorizes potential risks, supporting proactive medical decision-making. The project also incorporates disease risk prediction and clustering techniques to provide comprehensive patient health analysis. Experimental evaluation demonstrates strong anomaly detection performance and highlights the potential of machine learning in modern healthcare monitoring systems. Authors: Vedant Raut, Kunal Sambhare, and Anju Gupta Affiliation: Department of Biomedical Engineering, Ramdeobaba University, Nagpur, India.