Prediction model for major bleeding in anticoagulated patients with cancer‑associated venous thromboembolism using machine learning and natural language processing

• Andrés J. Muñoz Martín ^[1] ; Ramón Lecumberri ^[2] ; Juan Carlos Souto ^[3] ; Berta Obispo ^[4] ; Antonio Sanchez ^[5] ; Jorge Aparicio ^[6] ; Cristina Aguayo ^[7] ; David Gutierrez ^[8] ; Andrés García Palomo ^[9] ; Diego Benavent ^[10] ; Miren Taberna ^[10] ; María Carmen Viñuela Benéitez ^[11] ; Daniel Arumi ^[12] ; Miguel Ángel Hernández Presa ^[12]
  1. [1] Hospital General Universitario Gregorio Marañón

     Hospital General Universitario Gregorio Marañón

     Madrid, España

     
  2. [2] Clínica Universitaria de Navarra

     Clínica Universitaria de Navarra

     Pamplona, España

     
  3. [3] Hematology Department, Santa Creu i Sant Pau Hospital, Barcelona, Spain
  4. [4] Oncology Department, Infanta Leonor Hospital, Madrid, Spain
  5. [5] Oncology Department, Puerta de Hierro Hospital, Madrid, Spain
  6. [6] Oncology Department, Polytechnic and University Hospital of La Fé, Valencia, Spain
  7. [7] Oncology Department, Infanta Sofía Hospital, Madrid, Spain
  8. [8] Oncology Department, Fuenlabrada Hospital, Madrid, Spain
  9. [9] Oncology Department, University Hospital of León, León, Spain
  10. [10] Savana Research, Madrid, Spain
  11. [11] Medicine Department, Facultad de Medicina, Universidad Complutense, Madrid, Spain
  12. [12] Pfzer S.L.U. Medical Department, Madrid, Spain

  Mostrar afiliaciones +
• Localización: Clinical & translational oncology, ISSN 1699-048X, Vol. 27, Nº. 4, 2025, págs. 1816-1825
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • Purpose We developed a predictive model to assess the risk of major bleeding (MB) within 6 months of primary venous thromboembolism (VTE) in cancer patients receiving anticoagulant treatment. We also sought to describe the prevalence and incidence of VTE in cancer patients, and to describe clinical characteristics at baseline and bleeding events during follow-up in patients receiving anticoagulants.

    Methods This observational, retrospective, and multicenter study used natural language processing and machine learning (ML), to analyze unstructured clinical data from electronic health records from nine Spanish hospitals between 2014 and 2018. All adult cancer patients with VTE receiving anticoagulants were included. Both clinically- and ML-driven feature selection was performed to identify MB predictors. Logistic regression (LR), decision tree (DT), and random forest (RF) algorithms were used to train predictive models, which were validated in a hold-out dataset and compared to the previously developed CAT-BLEED score.

    Results Of the 2,893,108 cancer patients screened, in-hospital VTE prevalence was 5.8% and the annual incidence ranged from 2.7 to 3.9%. We identifed 21,227 patients with active cancer and VTE receiving anticoagulants (53.9% men, median age of 70 years). MB events after VTE diagnosis occurred in 10.9% of patients within the frst six months. MB predictors included: hemoglobin, metastasis, age, platelets, leukocytes, and serum creatinine. The LR, DT, and RF models had AUCROC (95% confdence interval) values of 0.60 (0.55, 0.65), 0.60 (0.55, 0.65), and 0.61 (0.56, 0.66), respectively. These models outperformed the CAT-BLEED score with values of 0.53 (0.48, 0.59).

    Conclusions Our study shows encouraging results in identifying anticoagulated patients with cancer-associated VTE who are at high risk of MB.