Machine Learning Prediction of Financial Toxicity in Patients with Resected Lung Cancer
Nathaniel Deboever1, Qasem Al-Tashi2, Michael Eisenberg1
1From the Departments of Thoracic and Cardiovascular Surgery (Deboever, Eisenberg, Antonoff, Hofstetter, Mehran, Rice, Roth, Swisher, Vaporciyan, Walsh, Rajaram), University of Texas MD Anderson Cancer Center, Houston, TX.
Journal of the American College of Surgeons
|March 3, 2025
Summary
Machine learning accurately predicts financial toxicity (FT) in lung cancer patients. Identifying at-risk individuals preoperatively allows for interventions to mitigate financial stress and improve quality of life.
Area of Science:
- Oncology
- Health Economics
- Data Science
Background:
- Financial toxicity (FT) significantly impacts lung cancer (LC) patients' quality of life due to treatment costs.
- Identifying patients at risk for major FT preoperatively is crucial for timely intervention.
Purpose of the Study:
- To develop and validate machine learning (ML) models for predicting major FT in patients undergoing resected LC.
- To identify key preoperative characteristics associated with FT in LC survivors.
Main Methods:
- Survey data on demographics, finances, and FT were collected from 462 LC resection patients.
- Clinicopathologic variables were extracted, and patients were split into training/testing sets.
- Four ML algorithms were trained and ensembled, optimizing predictive performance.
Main Results:
- 10.0% of surveyed patients experienced major FT.
- Key predictors included age, race, income, credit score, employment status, and lung function (FEV1).
- The ensemble ML model achieved high accuracy (0.86), precision (0.93), and F1 score (0.88).
Conclusions:
- ML models reliably identify LC patients at high risk of major FT.
- Preoperative risk stratification enables proactive interventions to manage financial toxicity.
- Early identification can improve patient outcomes and reduce financial burden.
