Utilizing machine learning to predict post-treatment outcomes in chronic non-specific neck pain patients undergoing cervical extension traction

Ibrahim M Moustafa; Dilber Uzun Ozsahin; Mubarak Taiwo Mustapha; Amal Ahbouch; Paul A Oakley; Deed E Harrison

doi:10.1038/s41598-024-62812-7

Utilizing machine learning to predict post-treatment outcomes in chronic non-specific neck pain patients undergoing cervical extension traction

Sci Rep. 2024 May 23;14(1):11781. doi: 10.1038/s41598-024-62812-7.

Authors

Ibrahim M Moustafa^{1

2

3}, Dilber Uzun Ozsahin^{4

5

6}, Mubarak Taiwo Mustapha^{5

6

7}, Amal Ahbouch^{1

2}, Paul A Oakley^{8

9

10}, Deed E Harrison¹¹

Affiliations

¹ Department of Physiotherapy, College of Health Sciences, University of Sharjah, 27272, Sharjah, United Arab Emirates.
² Neuromusculoskeletal Rehabilitation Research Group, RIMHS-Research Institute of Medical and Health Sciences, University of Sharjah, 27272, Sharjah, United Arab Emirates.
³ Faculty of Physical Therapy, Cairo University, Giza, 12613, Egypt.
⁴ Department of Medical Diagnostic Imaging, College of Health Science, University of Sharjah, Sharjah, United Arab Emirates.
⁵ Operational Research Centre in Healthcare, Near East University, TRNC Mersin 10, 99138, Nicosia, Turkey.
⁶ Research Institute for Medical and Health Sciences, University of Sharjah, Sharjah, United Arab Emirates.
⁷ Department of Biomedical Engineering, Near East University, Nicosia, Mersin 10, Turkey.
⁸ CBP Nonprofit (a Spine Research Foundation), Eagle, ID, 83616, USA.
⁹ Private Practice, Newmarket, ON, L3Y 8Y8, Canada.
¹⁰ Kinesiology and Health Science, York University, Toronto, ON, M3J 1P3, Canada.
¹¹ CBP Nonprofit (a Spine Research Foundation), Eagle, ID, 83616, USA. drdeedharrison@gmail.com.

Abstract

This study explored the application of machine learning in predicting post-treatment outcomes for chronic neck pain patients undergoing a multimodal program featuring cervical extension traction (CET). Pre-treatment demographic and clinical variables were used to develop predictive models capable of anticipating modifications in cervical lordotic angle (CLA), pain and disability of 570 patients treated between 2014 and 2020. Linear regression models used pre-treatment variables of age, body mass index, CLA, anterior head translation, disability index, pain score, treatment frequency, duration and compliance. These models used the sci-kit-learn machine learning library within Python for implementing linear regression algorithms. The linear regression models demonstrated high precision and accuracy, and effectively explained 30-55% of the variability in post-treatment outcomes, the highest for the CLA. This pioneering study integrates machine learning into spinal rehabilitation. The developed models offer valuable information to customize interventions, set realistic expectations, and optimize treatment strategies based on individual patient characteristics as treated conservatively with rehabilitation programs using CET as part of multimodal care.

Keywords: Cervical spine; Disability; Lordosis; Machine learning; Neck pain; Prediction; Traction.

MeSH terms

Adult
Cervical Vertebrae
Chronic Pain* / therapy
Female
Humans
Machine Learning*
Male
Middle Aged
Neck Pain* / therapy
Traction* / methods
Treatment Outcome