Machine Learning Model Predicts In-Hospital Mortality in Cancer Patients with Acute Pulmonary Embolism

Strategic Insights -

Research conducted at Peking University People's Hospital in Beijing, People's Republic of China, has developed and validated a machine learning model using XGBoost to predict in-hospital mortality in cancer patients with acute pulmonary embolism (APE). The study analyzed a retrospective cohort of 448 cancer patients with APE, divided into a training set and an internal validation set. The model achieved an area under the ROC curve (AUC) of 0.806 in the internal validation set and 0.724 in the external validation set. The model's performance was found to be accurate and reliable, with a high clinical benefit demonstrated through decision curve analysis (DCA).

Key Takeaways:

  • A machine learning model using XGBoost was developed and validated to predict in-hospital mortality in cancer patients with APE.
  • The model was trained on a retrospective cohort of 448 cancer patients with APE, divided into a training set and an internal validation set.
  • The model's performance was evaluated using the area under the ROC curve (AUC), which was found to be 0.806 in the internal validation set and 0.724 in the external validation set.
  • The model's top 10 predictors of in-hospital mortality included Glasgow Coma Scale (GCS) score, albumin, platelet count, age, serum creatinine, hemoglobin, presence of metastasis, lactate, creatine kinase (CK), and types of cancer.
  • The model's performance was found to be accurate and reliable, with a high clinical benefit demonstrated through decision curve analysis (DCA).
  • The model has the potential to improve patient outcomes through early intervention and personalized treatment strategies.
  • Further validation in diverse clinical settings is warranted to confirm the model's generalizability.
  • Yu-Juan Xue and Qi-Wei Xue led the research team, consisting of Zhen-Nan Yuan, Hai-Jun Wang, Shi-Ning Qu, Chu-Lin Huang, Hao Wang, Hao Zhang, Min-Ze Zhang, and Xue-Zhong Xing.

Statistics:

  • 448 cancer patients with APE were analyzed in the study.
  • The model was trained on a training set of 314 patients (70% of the total) and an internal validation set of 134 patients (30% of the total).
  • The external validation cohort consisted of 56 patients.
  • The model's performance in the internal validation set was found to be AUC = 0.806 (95% CI: 0.717-0.896).
  • The model's performance in the external validation set was found to be AUC = 0.724 (95% CI: 0.686-0.901).
  • The model's top 10 predictors of in-hospital mortality included Glasgow Coma Scale (GCS) score, albumin, platelet count, age, serum creatinine, hemoglobin, presence of metastasis, lactate, creatine kinase (CK), and types of cancer.

Sources:

  • A predictive model for hospital death in cancer patients with acute pulmonary embolism using XGBoost machine learning and SHAP interpretation. Scientific Reports, 2025;15(1):18268.
  • Nature Publishing Group - www.nature.com/
  • Scientific Reports - www.nature.com/srep/