JMIR Publications

ABSTRACT

Background:

Chronic pain and functional loss secondary to knee osteoarthritis puts patients at risk of developing depression, which can also impair their treatment response. However, no tools exist to assist clinicians in identifying at risk patients. We investigated whether ML models could predict the development of depression in patients with knee osteoarthritis, and examined which features are most predictive.

Objective:

This study aims to develop, test, and externally validate an ML prediction model for depression in patients with knee osteoarthritis.

Methods:

Osteoarthritis Initiative Study (OAI) data was used for model development and external validation was conducted using Multicenter Osteoarthritis Study (MOST) data. Forty-two features denoting routinely collected demographic and clinical data were used to train common ML classification models to predict the presence of depression at two years following enrolment. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC) and F1 score, and the most important features were extracted from the best performing model on external validation.

Results:

5,947 patients were included in this study, with 2,969 in the training set, 742 in the test set, and 2,236 in the external validation set. For the test set, AUC ranged from 0.673 (95%CI:0.604-0.742) to 0.869 (95%CI:0.824-0.913), with F1 score of 0.435 to 0.490. On external validation, the AUC varied from 0.720 (95%CI:0.685-0.755) to 0.876 (95%CI:0.853-0.899), with F1 score of 0.456 to 0.563. LASSO modelling offered the highest predictive performance. Blood pressure, baseline depression score, knee pain and stiffness, and quality of life were the most predictive features

Conclusions:

Machine learning models can deliver a clinically acceptable level of performance (AUC >0.7) in predicting the development of depression in knee osteoarthritis patients using routinely available demographic and clinical data. Further work is required to address the class imbalance in the training data, and to evaluate their clinical utility in facilitating early intervention and improved outcomes.