Artificial intelligence in healthcare management: Clinical applications, evidence, governance, and implementation challenges

Artificial intelligence (AI) has evolved from experimental computer science into a practical component of modern healthcare systems. AI is increasingly applied in medical imaging, oncology, dermatology, ophthalmology, electronic health record analytics, clinical decision support, drug discovery, patient communication, and hospital management. However, the effectiveness of AI depends not only on algorithmic performance but also on safety, equity, explainability, workflow integration, regulatory oversight, and public trust. This paper presents an integrative narrative review of AI in healthcare based on evidence from peer-reviewed literature indexed in major databases, particularly Scopus and Web of Science. The review focuses on five key themes: diagnostic support, predictive analytics, treatment personalization, generative AI, and responsible implementation. Evidence indicates that AI can achieve specialist-level performance in tasks such as skin cancer classification, diabetic retinopathy detection, breast cancer screening, and medical image interpretation. AI systems using electronic health records can also predict deterioration, mortality, readmission, and acute kidney injury. Recent advances in large language models demonstrate potential for medical question answering, documentation assistance, and patient communication. Despite these benefits, many AI systems remain inadequately validated in real-world settings. Major concerns include algorithmic bias, lack of transparency, privacy risks, automation bias, and weak external validity. Safe AI adoption therefore requires rigorous clinical validation, continuous monitoring, transparent governance, and human-centered implementation that supports rather than replaces professional clinical judgment.