Explainable machine learning for predicting venous thromboembolism in septic shock patients - Takeaways - MDSpire

Utilizing Explainable Machine Learning to Forecast Venous Thromboembolism in Patients Experiencing Septic Shock

  • By

  • Yuanyuan Li

  • Qi Xin

  • Yizhao Lu

  • Xiaoyuan Yu

  • Chunyu Gu

  • July 20, 2026

Share

  • 1

    The study developed an explainable machine learning framework to predict venous thromboembolism in patients with septic shock.

  • 2

    A total of 733 patients were included in the internal development cohort, with 257 patients in the external validation cohort.

  • 3

    Six predictors were identified: fibrin degradation products, prothrombin time, white blood cells, activated partial thromboplastin time, D-dimer, and C-reactive protein.

  • 4

    The Random Forest algorithm achieved an AUC of 0.9718 and an F1 score of 0.7917 in the external validation cohort.

  • 5

    SHAP analysis indicated that elevated thrombo-inflammatory markers and coagulation intervals significantly influenced VTE risk.

Original Source(s)

Explainable machine learning for predicting venous thromboembolism in septic shock patients

  • by Yuanyuan Li, Qi Xin, Yizhao Lu, Xiaoyuan Yu, Chunyu Gu

  • July 20, 2026

Related Content

Pre-transplant measurable residual disease by flow cytometry is an independent prognostic factor in pediatric acute myeloid leukemia undergoing allogeneic hematopoietic stem cell transplantation

Ponatinib Safety Profile: An Analysis of 10 Years of Real-World Experience

Rivaroxaban plus antiplatelet therapy for coronary artery ectasia: 36-month outcomes and risk prediction from a retrospective cohort study