Survival Period Prediction in Cervical Cancer Patients Using the Selective Stacking Technique

Area of Science:

• Oncology
• Machine Learning
• Biostatistics

Background:

• Cervical cancer remains a significant global health challenge.
• Accurate survival prediction is crucial for effective treatment planning.
• Existing prediction models may lack the precision needed for personalized care.

Purpose of the Study:

• To develop an advanced survival prediction model for cervical cancer using ensemble machine learning.
• To enhance the effectiveness of cervical cancer treatment through improved predictive accuracy.
• To introduce and validate the selective stacking technique for clinical application.

Main Methods:

• Utilized patient data from Chiang Mai University, Thailand, for real-world validation.
• Implemented a two-stage selective stacking framework with meta-level learning.
• Applied local interpretable model-agnostic explanations for feature importance analysis.

Main Results:

• The selective stacking model achieved 91.41% accuracy in classification.
• The regression model demonstrated a root mean square error of 18.92 and an r-value of 0.669.
• Side effect status involving surrounding organs was identified as the most influential predictor.

Conclusions:

• The selective stacking model significantly outperformed individual base machine learning models.
• This ensemble approach shows promise for cervical cancer survival prediction.
• The findings support the development of personalized treatment strategies for cervical cancer patients.