Artificial intelligence applications in thrombotic thrombocytopenic Purpura: A systematic review of diagnostic, risk stratification, and prognostic models

Thrombotic thrombocytopenic purpura (TTP) is an uncommon but life-threatening thrombotic microangiopathy (TMA) caused by severe ADAMTS13 deficiency. Early diagnosis and prompt initiation of plasma exchange are critical; however, access to rapid ADAMTS13 testing remains variable across institutions. Clinical tools such as the PLASMIC and French scores assist risk assessment but have limited specificity. Artificial intelligence (AI) has emerged as a potential adjunctive approach to improve diagnostic and prognostic evaluation in TTP. This systematic review evaluated current evidence on AI applications in the diagnosis, risk stratification, and prognostic prediction of TTP. Following PRISMA guidelines (PROSPERO: CRD420261323545), PubMed, Scopus, Embase, and Web of Science were searched through January 9, 2026. Eight studies met inclusion criteria. Reported area under the curve (AUC) values ranged from 0.77 to 0.94. However, all studies were retrospective and relied on internal validation without external or prospective evaluation. Although several models demonstrated good discriminative performance, heterogeneity in study design and target populations limits direct comparison, evidence remains limited, and further externally validated studies are required before clinical implementation.