Applying Machine Learning on Patient-Reported Data to Model the Selection of Appropriate Treatments for Low Back Pain: A Pilot Study

Abstract

The objective of this pilot study was to determine whether machine learning can be applied on patient-reported data to model decision-making on treatments for low back pain (LBP). We used a database of a university spine centre containing patient-reported data from 1546 patients with LBP. From this dataset, a training dataset with 354 features (input data) was labelled on treatments (output data) received by these patients. For this pilot study, we focused on two treatments: pain rehabilitation and surgery. Classification algorithms in WEKA were trained, and the resulting models were validated during 10-fold cross validation. Next to this, a test dataset was constructed -containing 50 cases judged on treatments by 4 master physician assistants (MPAs) -to test the models with data not used for training. We used prediction accuracy and average area under curve (AUC) as performance measures. The interrater agreement among the 4 MPAs was substantial (Fleiss Kappa 0.67). The AUC values indicated small to medium (machine) learning effects, meaning that machine learning on patient-reported data to model decision-making processes on treatments for LBP seems possible. However, model performances must be improved before these models can be used in real practice.

Natural Language Processing of Referral Letters for Machine Learning–Based Triaging of Patients With Low Back Pain to the Most Appropriate Intervention: Retrospective Study

“…Health care professionals have been able to triage patients to surgical, anesthesiologic, or rehabilitation interventions with moderate agreement [ 7 ]. To progress in individualizing treatment and referrals, artificial intelligence systems, particularly machine learning (ML), are a relatively new method that can identify patient patterns based on more input than can be incorporated into traditional models.…”

“…To progress in individualizing treatment and referrals, artificial intelligence systems, particularly machine learning (ML), are a relatively new method that can identify patient patterns based on more input than can be incorporated into traditional models. A few examples have appeared on LBP previously, showing that models could be applied to the data [ 7 ] and support the decision of who should get low back surgery [ 8 ]. A clinical decision support system (CDSS) may lead to an increase in triaging patients to the best-possible or targeted treatment and increase the cost-effectiveness of current care for LBP.…”

“…A clinical decision support system (CDSS) may lead to an increase in triaging patients to the best-possible or targeted treatment and increase the cost-effectiveness of current care for LBP. These models, however, are limited to self-report questionnaires as input and may consequently still have low model fit values, leading to nonvaluable support tools [ 7 ].…”