A systematic literature review approach is employed in this paper to investigate the use of artificial intelligence algorithms in medical diagnostics. The search identifies studies published between 2015 and January 2025, using rigorous inclusion and exclusion criteria. Thirty-three articles were selected to review the evolution of artificial intelligence, identify commonly used algorithms, and evaluate their validation methods. The results show a significant increase in artificial intelligence-related publications over the last decade, with convolutional neural networks becoming the most widely used algorithm. Validation methods mainly rely on performance measures and cross-validation. Artificial intelligence applications primarily focus on chronic disease diagnosis, medical outcome prediction, and cancer screening, with limited exploration of personalized medicine. The study identifies gaps in evaluation methods and highlights the lack of large-scale clinical validations in low-resource settings. Addressing these challenges requires standardized evaluation frameworks, improved data diversity, extensive clinical validations, and expanded research in underexplored areas to enhance scalability, equity, and innovation in health applications.