Beyond the platelet count: A systematic review of artificial intelligence in immune thrombocytopenia

Immune thrombocytopenia (ITP) is an acquired autoimmune bleeding disorder with variable disease course, bleeding risk, and treatment response. Artificial intelligence (AI) and machine learning (ML) approaches may address diagnostic and therapeutic challenges by identifying complex patterns across heterogeneous clinical and biological data. A systematic search of PubMed, Embase, Web of Science and Scopus from inception to 13 August 2026 was conducted for studies applying ML to patient-level prediction in primary or disease-associated secondary ITP. Twelve studies were included, eleven from China, applying a range of machine learning approaches. Reported areas under the curve ranged from 0.693 to 0.937, but only two studies reported estimates from independent cohorts. Risk of bias was assessed using PROBAST+AI and eleven of the 12 studies were at high risk of bias. Further development with external validation and prospective evaluation in real-world clinical workflows is needed before these tools can be integrated into clinical practice.