Abstract
Background/Objectives: We define severe postpartum hemorrhage (PPH) with macroscopic hematuria as clinical disseminated intravascular coagulation (DIC), a life-threatening condition. We also report a methodology using machine learning, a subtype of artificial intelligence, for developing the boundary criterion for predicting hematuria on the fibrinogen-fibrin/fibrinogen degradation product (FDP) plane. A positive FDP-fibrinogen/3-60 (mg/dL) value indicates hematuria; otherwise, non-hematuria is observed. We aimed to validate this criterion using severe placental abruption (PA), and to examine the activation of the coagulation-fibrinolytic system in clinical DIC. Methods: Of 17,285 deliveries across nine perinatal centers in Japan between 2020 and 2024, 13 had severe PA without hematuria, 18 had severe PPH without hematuria, and 3 had severe PPH with hematuria, i.e., clinical DIC. We calculated the values of the criterion formula for 13 cases of severe PA to validate the boundary criterion and compared the laboratory tests for coagulation-fibrinolytic activation among the three groups. Results: The calculated values using the criterion for the 13 PA without hematuria ranged from -108.91 to -5.87 and all were negative. In cases of clinical DIC, fibrinogen levels (median, 62 mg/dL) were lower (p < 0.05), while levels of FDP (96 mg/dL), the thrombin-antithrombin complex (120 ng/mL), and the plasmin-α(2)-plasmin inhibitor complex (28.4 μg/mL) were significantly higher than in the other two groups. Conclusions: This study demonstrated the validity of the boundary criterion for predicting hematuria using severe PA. The coagulation-fibrinolytic test results suggested that PPH cases with hematuria were assumed to have clinical DIC, indicating that this criterion may be considered for diagnosing DIC during delivery. However, further additional patient data are needed to confirm the usefulness of this criterion because of the very low number of hematuria cases.