Advantages and application of a decision tree algorithm combined with the Harris-Boyd criterion in the construction of dynamic reference intervals for coagulation indicators during pregnancy

INTRODUCTION: Reference intervals (RIs) based on fixed gestationals may not accurately reflect the dynamic changes of the hypercoagulable state. This study aimed to evaluate the rationality of integrating a decision tree algorithm (DTA) with the Harris-Boyd criterion (DTAHBC) to establish dynamic RIs for coagulation indicators during pregnancy. MATERIALS AND METHODS: Dynamic data of prothrombin time (PT), international normalized ratio (INR), activated partial thromboplastin time (APTT), thrombin time (TT), fibrinogen (Fib), D-dimer (D-d), and fibrin degradation product (FDP) were retrospectively analyzed in 4,698 healthy pregnant women. Scatter plots were drawn to observe changes regarding each indicator during pregnancy. DTAs were used to identify dynamic inflection points, which were validated using the Harris-Boyd criterion. Moreover, 2.5th and 97.5th percentiles and their 90% confidence intervals were calculated. New RIs were compared to traditional trimester-based RIs and validated by coincidence rates within the cohort. RESULTS: The study showed that PT, INR, APTT, and TT values were stable across gestation weeks. However, Fib, D-d, and FDP values increased rapidly. The DTA analysis found inflection points misaligned with traditional trimester-based nodes. Compared to traditional methods, the new method identified the initial stage of coagulation activation in the second trimester (FDP exceeding non-pregnant ranges at 14-21 weeks) and the aggravation stage of hypercoagulability in the late trimester (median FDP exceeding non-pregnant upper limits at 34-40 weeks). Validation in the same cohort revealed compliance rates >90% for new RIs, whereas traditional TT (82.11%) and D-d (87.89%) measurements obtained at 14-27 weeks failed validation. CONCLUSION: DTAHBC-based RIs do not follow the traditional trimester division, can effectively reflect dynamic changes in coagulation indicators, and provide a new solution for coagulation-function evaluation during pregnancy.