Predicting Depression in Patients With Knee Osteoarthritis Using Machine Learning: Model Development and Validation...

Area of Science:

• Computational medicine
• Artificial intelligence in healthcare
• Osteoarthritis research

Background:

• Knee osteoarthritis (OA) is a leading cause of global disability and pain.
• Depression is a common comorbidity in knee OA patients, impacting treatment outcomes.
• Current tools for identifying at-risk patients are lacking.

Purpose of the Study:

• To develop and validate machine learning (ML) models for predicting depression in knee OA patients.
• To identify key clinical features predictive of depression development.
• To assess model performance using internal and external datasets.

Main Methods:

• Utilized Osteoarthritis Initiative (OAI) and Multicenter Osteoarthritis Study (MOST) datasets.
• Trained six ML classification models on 42 demographic and clinical features.
• Evaluated models using Area Under the Receiver Operating Characteristic Curve (AUC) and F1 score.

Main Results:

• The Least Absolute Shrinkage and Selection Operator (LASSO) model demonstrated the highest predictive performance.
• On external validation, models achieved an AUC ranging from 0.720 to 0.876.
• Blood pressure, baseline depression score, knee pain/stiffness, and quality of life were identified as most predictive.

Conclusions:

• ML models can predict depression in knee OA patients with clinically acceptable performance (AUC > 0.7) using routinely collected data.
• This study is the first to apply ML for predicting depression in knee OA.
• Further research is needed to address data imbalance and evaluate clinical utility for early intervention.