Abstract
Thrombotic thrombocytopenic purpura (TTP) is a rare, life threatening thrombotic microangiopathy that requires prompt diagnosis to reduce mortality. However, its early identification is often hindered by delayed ADAMTS13 testing, particularly in low resource settings. In this study, we developed a machine learning-based model using readily available inflammatory markers, including systemic immune inflammation index (SII), platelet to lymphocyte ratio (PLR), and platelet neutrophil product (PPN), to distinguish TTP from immune thrombocytopenia (ITP). A retrospective analysis of 196 hospitalized patients was conducted, and eight machine learning models were trained and compared. Logistic regression achieved the best performance (AUC = 0.78), with SII identified as the most influential predictor. While the PLASMIC score remains a widely used tool with higher diagnostic accuracy (AUC = 0.92), our model relies only on routine blood tests and offers a fast, accessible alternative for early risk stratification. These findings suggest that composite inflammatory markers combined with machine learning can assist in the rapid triage of suspected TTP cases, especially in emergency or resource-limited environments.