Sign in Subscribe
Cancer

Machine Learning Framework for Gastric Cancer Clinical Trial Enrollment Boosts Efficiency

Strategic Insights

— 1 min read

Researchers from Kyungpook National University have developed a machine learning (ML)-based framework using clinical laboratory parameters to identify eligible participants for gastric cancer clinical trials. The framework, which was tested using electronic medical records from 11,592 patients with gastric cancer, achieved a high accuracy rate, reducing the workload by 57% and efficiently identifying 150 valid patients from a pool of 209. This research provides valuable insights into the potential of ML-based frameworks for participant selection in clinical trials, enabling faster enrollment and improving trial efficiency.

Key Takeaways:

  • The ML-based framework was developed using 8 clinical laboratory parameters, including complete blood count, liver, and kidney function tests, and was able to accurately identify patients eligible for clinical trials with an F1-score above 0.8 and an AUC value exceeding 0.8.
  • The framework reduced the workload by 57%, efficiently identifying 150 valid patients from a pool of 209, compared to 485 patients required by random selection.
  • The research highlighted the importance of participant enrollment in clinical trial failure, with insufficient enrollment being a major factor contributing to trial failure.
  • The ML-based framework was compared with a random selection method, showing its ability to accurately identify valid clinical trial candidates while minimizing misclassification.
  • The proposed model's sensitivity was high, further enhancing its efficiency in prioritizing patients for screening.
  • The framework was validated using two datasets: a training dataset to design the ML-based candidate selection method and a test dataset to evaluate its performance.
  • The research was peer-reviewed and published in the Journal of Engineering.

Statistics:

  • 11,592 patients with gastric cancer were used to develop and validate the ML-based framework.
  • 8 clinical laboratory parameters were used in the framework, including complete blood count, liver, and kidney function tests.
  • The framework achieved an F1-score above 0.8 and an AUC value exceeding 0.8.
  • The workload was reduced by 57% using the proposed ML-based framework.
  • 150 valid patients were identified from a pool of 209 using the ML-based framework, compared to 485 patients required by random selection.

Sources:

  • Clinical Laboratory Parameter-Driven Machine Learning for Participant Selection in Bioequivalence Studies Among Patients With Gastric Cancer: Framework Development and Validation Study. JMIR AI, 2025;4.
  • NewsRx. Kyungpook National University Reports Findings in Gastric Cancer (Clinical Laboratory Parameter-Driven Machine Learning for Participant Selection in Bioequivalence Studies Among Patients With Gastric Cancer: Framework Development and Validation ...). Journal of Engineering. July 14, 2025; p 1677.

Read more