AI Applications in Diagnosing Rare Genetic Disorders

Abstract

Rare diseases (RDs) represent a broad category of conditions that individually affect a relatively small proportion of the population but collectively impact millions of people worldwide. Because of their clinical heterogeneity and frequently nonspecific symptoms, obtaining an accurate diagnosis can be a prolonged process, with many patients experiencing years of uncertainty before receiving a definitive diagnosis. Approximately 80% of rare diseases have a genetic basis, and the majority are chronic conditions with limited treatment options. In recent years, Artificial Intelligence (AI) has emerged as a valuable tool for addressing these challenges. Using Machine Learning (ML) and Deep Learning (DL) techniques, AI systems can process and analyze large volumes of genomic, phenotypic, and clinical data, improving both the speed and accuracy of diagnosis. AI-based platforms such as Phenomizer and Face2Gene assist clinicians in recognizing disease- associated phenotypic patterns and prioritizing potential diagnoses. In addition, AI facilitates patient screening using Electronic Health Records (EHRs) and contributes to treatment optimization as well as disease management. Within neonatal medicine, AI is increasingly applied to facilitate the diagnosis of rare diseases through rapid genomic data interpretation, syndrome recognition based on facial features, and screening for inherited metabolic disorders. Despite these advances, several limitations remain, including restricted data availability, concerns regarding data privacy, and the need for greater transparency and trust in AI-driven healthcare applications.