Machine Learning Model Tool for Identifying Postoperative Bleeding Effective

Strategic Insights -

A new study from the Department of Technology at Shanghai Palan DataRx Co. Ltd. has developed a machine learning model tool for identifying postoperative patients with major bleeding based on electronic medical records. The researchers used data from a database of 2,000 patients who underwent in-hospital tumor resection surgery between January 2018 and December 2021. The study aimed to determine whether machine learning can efficiently process large volumes of medical text to identify postoperative bleeding effectively. The research found that the logistic regression (LR) and convolutional neural network (CNN) models demonstrated good performance in identifying major bleeding in patients with postoperative malignant tumors, exhibiting high sensitivity and specificity.

Key Takeaways:

  • The study developed a machine learning model tool for identifying postoperative patients with major bleeding based on electronic medical records.
  • The researchers used data from a database of 2,000 patients who underwent in-hospital tumor resection surgery between January 2018 and December 2021.
  • The study found that the logistic regression (LR) method achieved an accuracy of 0.8275, a sensitivity of 0.8947, and a specificity of 0.8241 in the test set.
  • The convolutional neural network (CNN) method demonstrated an accuracy of 0.8900, a sensitivity of 0.8421, and a specificity of 0.8924 in the test set.
  • Both the LR and CNN methods exhibited high sensitivity and specificity in identifying major bleeding in patients with postoperative malignant tumors.
  • The researchers concluded that both models hold promise for practical application, depending on specific clinical priorities.
  • The study highlights the potential of machine learning in processing large volumes of medical text to identify postoperative bleeding effectively.

Statistics:

  • 2,000 patients were selected from the database for the study.
  • 4.31% (69/1600) of the training set and 4.75% (19/400) of the test set had major bleeding.
  • The LR method achieved an accuracy of 0.8275 and a sensitivity of 0.8947 in the test set.
  • The CNN method demonstrated an accuracy of 0.8900 and a sensitivity of 0.8421 in the test set.
  • The C-statistic for the LR method was 0.9018 and for the CNN method was 0.8830.

Sources:

  • NewsRx. Department of Technology Reports Findings in Machine Learning (Identification of Major Bleeding Events in Postoperative Patients With Malignant Tumors in Chinese Electronic Medical Records: Algorithm Development and Validation). Journal of Mathematics. May 20, 2025; p 135.
  • Identification of Major Bleeding Events in Postoperative Patients With Malignant Tumors in Chinese Electronic Medical Records: Algorithm Development and Validation. JMIR Formative Research, 2025;9.