Predicting Phase 1 Lymphoma Clinical Trial Durations Using Machine Learning: An In-Depth Analysis and Broad
Bowen Long1, Shao-Wen Lai2, Jiawen Wu1
1Department of Analytics, Harrisburg University of Science and Technology, Harrisburg, PA 17101, USA.
Clinics and Practice
|January 22, 2024
Summary
A new machine learning model accurately predicts Phase 1 clinical trial durations for lymphoma and lung cancer. This tool aids in clinical research planning and may improve patient outcomes in oncology.
Area of Science:
- Oncology
- Clinical Research Methodology
Background:
- Lymphoma diagnoses are substantial in the US, emphasizing the need for efficient clinical trial processes.
- Phase 1 clinical trials are critical for developing innovative oncology treatments.
Purpose of the Study:
- To develop and validate a predictive model for assessing adherence to expected average durations of Phase 1 clinical trials.
- To evaluate the model's efficacy and versatility across different cancer types.
Main Methods:
- Analysis of 1089 completed Phase 1 lymphoma trials from clinicaltrials.gov.
- Development of a binary predictive model using the Random Forest machine learning algorithm.
- Statistical validation including accuracy, ROC-AUC, and confidence intervals.
Main Results:
- The Random Forest model achieved an accuracy of 0.7248 and an ROC-AUC of 0.7677 for Phase 1 lymphoma trials.
- The model demonstrated significant accuracy compared to alternative models.
- The model showed versatility with an ROC-AUC of 0.7701 on lung cancer trials.
- Higher predicted probabilities correlated with extended trial durations.
Conclusions:
- The developed Random Forest model effectively predicts Phase 1 clinical trial durations.
- The model's performance and versatility suggest utility in enhancing clinical research planning.
- Improved trial planning has the potential to positively impact patient outcomes in oncology.
