Pain Prediction Via Machine Learning: Electrical Thresholds and Interoception

The aim of this study is to investigate the relationship between musculoskeletal pain, electrical threshold values, and interoception in healthy university students, and to evaluate the predictive accuracy of machine learning models for both retrospective and prospective musculoskeletal pain.

Designed as an observational, analytical, and cross-sectional study with a prospective follow-up component, this study will include approximately 550 healthy, right-hand dominant students aged 18-25 who provide written informed consent.

Participants' demographic data and healthy lifestyle behaviors (physical activity, sitting time, nutrition, and tobacco/alcohol/substance use via ASSIST) will be recorded. Musculoskeletal complaints will be evaluated using the Nordic Musculoskeletal Questionnaire, and interoception will be assessed using the Multidimensional Assessment of Interoceptive Awareness-II. Finally, electrical sensory, motor, and pain thresholds will be measured using a combined electrotherapy device with electrodes placed over the flexor muscles of the forearm. One week later, participants will be contacted again to re-evaluate their pain complaints during the preceding 7 days.

In statistical analyses, in addition to correlation and regression analyses among parameters, the performance of three machine learning models-Logistic Regression, Random Forest, and XGBoost-in predicting participants' retrospective and prospective pain will be evaluated.

The research results will demonstrate several key outcomes. First, they will elucidate the relationship between musculoskeletal pain, electrical threshold values, and interoception in university students, while testing the utility of electrical threshold values as an objective tool in musculoskeletal pain assessment. Furthermore, the study will evaluate the accuracy of machine learning models in predicting both retrospective and prospective pain within this population. By leveraging objective measurements obtained from widely accessible clinical devices, the findings will contribute to digital health approaches and applications that guide researchers and clinicians throughout pain evaluation and treatment processes. Ultimately, this work will provide critical insights into the efficacy of the proposed research methodology before its implementation in diverse patient populations.