A Systematic Literature Review of Machine Learning Methods in Healthcare

Abstract

The integration of Machine Learning (ML) into healthcare has brought transformative potential across clinical decision-making, diagnostics, imaging, and personalized treatment. This review presents a comprehensive synthesis of recent advances in ML applications within healthcare, focusing on supervised, unsupervised, and deep learning methods. Through a systematic review of 102 peer-reviewed studies published between 2015 and 2025, we explore how ML has enhanced diagnostic accuracy, predicted disease progression, analyzed medical imaging, and enabled precision medicine. The review also highlights the increasing use of ML in resource-constrained settings like India, where mobile imaging tools and low-cost AI models are being adopted for public health screening and rural diagnostics. Supervised learning methods, including Random Forests, Support Vector Machines, and Logistic Regression, dominate current applications, particularly in disease diagnosis and risk stratification. Deep learning, especially Convolutional Neural Networks, has revolutionized medical imaging analysis, while unsupervised learning contributes to phenotype discovery and patient clustering. Despite the progress, challenges remain in model interpretability, data quality, regulatory compliance, and real-world deployment. Limited external validation and underrepresentation of diverse populations, especially from developing countries, are also noted. This review not only maps existing research trends but also underscores the need for context-specific innovations, ethical considerations, and robust clinical integration strategies to realize the full promise of ML in healthcare systems globally and within the Indian context.