Improving Early Identification of Significant Weight Loss Using Clinical Decision Support System in Lung Cancer...

Area of Science:

• Oncology
• Radiotherapy
• Medical Informatics

Background:

• Early identification of patients at risk of significant weight loss (SWL) is crucial for effective intervention during lung cancer radiotherapy (RT).
• Predicting SWL aids in timely clinical management and can improve patient outcomes.

Purpose of the Study:

• To implement and assess a clinical decision support system (CDSS) for predicting SWL in lung cancer patients within a routine clinical workflow.
• To evaluate the performance of a machine learning-based CDSS against physician predictions.

Main Methods:

• A prospective cohort of 37 lung cancer patients undergoing definitive RT was enrolled.
• A CDSS integrated radiomics and dosiomics features with a machine learning model for SWL prediction.
• Physicians' SWL predictions were compared before and after utilizing the CDSS.

Main Results:

• The CDSS achieved significantly higher prediction accuracy (0.73) and specificity (0.81) than physicians (0.54 accuracy, 0.50 specificity).
• Physicians' predictions improved in accuracy, sensitivity, and specificity after reviewing CDSS outputs.
• The CDSS correctly predicted SWL in all cases where physician predictions were altered.

Conclusions:

• A machine learning-based CDSS shows potential for enhancing SWL prediction accuracy in lung cancer RT.
• Further validation on larger patient cohorts is necessary to confirm the clinical utility and benefits of the CDSS in decision-making.