Development of a Machine Learning Algorithm-Based Predictive Model for Physical Activity Levels in Lung Cancer...

Area of Science:

• Oncology
• Rehabilitation Medicine
• Data Science

Background:

• Lung cancer survivors often exhibit low physical activity levels, impacting their quality of life.
• Physical activity is a crucial non-pharmacological intervention in cancer rehabilitation.
• Identifying factors influencing physical activity is key for targeted interventions.

Purpose of the Study:

• To investigate physical activity levels in lung cancer survivors.
• To analyze factors influencing these activity levels.
• To develop a machine learning-based predictive model for physical activity.

Main Methods:

• A cross-sectional study surveyed 2231 lung cancer survivors across 14 hospitals in China.
• Data on demographic, disease, health, physical, and psychosocial factors were collected.
• Four machine learning models were evaluated, with Random Forest selected for its superior performance (AUC-ROC 0.86).

Main Results:

• 30% of survivors had low physical activity levels.
• 15 independent factors were identified, including grip strength, MDASI score, and depression score.
• A Random Forest model achieved an AUC-ROC of 0.86, leading to an online predictive tool.

Conclusions:

• A machine learning model, specifically Random Forest, accurately predicts physical activity in lung cancer survivors.
• The developed tool aids in early identification of low-activity survivors.
• Facilitates timely, personalized rehabilitation and health management for improved quality of life.