AI-Enabled Smart Healthcare Systems: Innovative Approaches to Disease Prediction, Accurate Diagnosis, and Effective Public Health Management

Abstract

Artificial intelligence (AI) has moved from a peripheral research interest to a central component of contemporary healthcare delivery, with deep learning systems now matching or exceeding specialist-level performance on discrete diagnostic tasks spanning dermatology, ophthalmology, radiology, and oncology. This paper reviews the technical foundations and applied evidence base for AI-driven smart healthcare systems across three domains: disease prediction from structured and unstructured clinical data, image-based diagnostic classification, and population-level public health management, including the accelerated adoption of AI tools during the COVID-19 pandemic. The review synthesizes landmark diagnostic-performance studies, including dermatologist-level skin cancer classification, diabetic retinopathy detection, pneumonia detection from chest radiographs, and international breast cancer screening evaluation, alongside the clinical machine learning literature addressing implementation, validation, and algorithmic bias. Comparative tables summarize reported diagnostic performance metrics, data modalities, and clinical domains across the reviewed systems. The paper concludes that AI-driven diagnostic systems have achieved genuine, reproducible performance parity with human specialists on narrow, well-defined tasks, while broader clinical deployment remains constrained by validation, generalizability, and algorithmic-bias challenges that the reviewed literature has only begun to resolve, and identifies prospective real-world validation and equity-focused model auditing as the central future research prospects.