Utilizing patient data: A tutorial on predicting second cancer with machine learning models.

Area of Science:

• Oncology
• Medical Physics
• Data Science

Background:

• Radiation therapy (RT) can increase the risk of secondary cancer (SC).
• Current risk assessment models have limitations in predicting SC.
• Novel modeling techniques are needed to mitigate SC risk.

Purpose of the Study:

• To develop a practical framework for forecasting SC occurrence using patient data.
• To leverage machine learning (ML) for personalized risk stratification.
• To identify key factors influencing SC development post-RT.

Main Methods:

• Utilized machine learning (ML) models, specifically decision trees.
• Employed patient data to train and assess ML models for SC prediction.
• Established a framework for categorizing patients into high-risk or low-risk groups.

Main Results:

• The developed framework aids in personalized treatment planning.
• Identified factors like radiation dosage, patient age, and genetic predisposition influence SC risk.
• Highlighted limitations of current models in accounting for complex variables.

Conclusions:

• Enhanced understanding and prediction of SC following RT.
• Facilitation of personalized treatment approaches for cancer patients.
• Established a framework for utilizing patient data in ML models for improved oncological care.