Exploratory application of machine learning methods on patient reported data in the development of supervised models...

Area of Science:

• Computational medicine
• Health informatics
• Machine learning in healthcare

Background:

• Patient-reported outcome measurements (PROMs) are vital for clinical decision-making.
• Limited research exists on machine learning for predicting PROMs outcomes.

Purpose of the Study:

• To evaluate machine learning methods for predicting patient outcomes using PROMs data.
• To assess the predictive capabilities of machine learning on neck and low back pain patient datasets.

Main Methods:

• Utilized two PROMs datasets from primary care and rehabilitation settings.
• Applied data preprocessing techniques and evaluated regression and classification models.
• Focused on predicting patient outcomes for non-specific neck and low back pain.

Main Results:

• Machine learning shows potential for predicting and classifying PROMs outcomes.
• Prediction models using baseline measurements performed well, with potential for reduced predictors.
• Classification accuracy was limited by the available predictors in one dataset.

Conclusions:

• PROMs hold significant potential for predicting short-term patient outcomes.
• Machine learning can effectively utilize the predictive power of PROMs to support clinical decisions.
• Generalizable machine learning pipelines were developed, adaptable to other PROMs datasets.