Subgrouping non-specific low back pain based on spinal marker trajectory data: an unsupervised machine learning approach

Key Points

1. Movement patterns in non-specific low back pain patients are highly heterogeneous.

All key points available for members only

Start free trial

Or log in here

BACKGROUND & OBJECTIVE

A total of 90% of LBP lacks identifiable pathology and is termed ‘non-specific’ low back pain (NSLBP) (1). NSLBP is a heterogeneous condition with multiple contributary factors: physical, psychological, behavioural, lifestyle and societal (2). There has been significant research into NSLBP movement alterations being a contributary factor via repetitive tissue irritation and the fear of movement, loss of function, physical deconditioning and the subsequent disability loop (3,4).

Clinically, designing a customized movement re-education program for the NSLBP patient is challenging given the varied underlying mechanisms and diverse movement alterations.

This study aimed to investigate whether clinic based smart phone assessments could identify distinct movement- based subgroups using thoraco-lumbo-pelvic marker trajectories and a novel machine learning K-means clustering approach.

A total of 90% of LBP lacks identifiable pathology and is termed ‘non-specific’ low back pain.

This study suggests the future of treatments of back pain should be tailored to subgrouping patients to improve treatment effect, and AI/smart phone technology may assist with this.

METHODS

Public service office workers were recruited: NSLBP n= 115 (15 m, 100 f) and Controls n=57 (8 m, 49 f). Kinematic data was gathered via a smart phone-based video recording system. Open- source software extracted X and Y axis displacements

Start free trial

to unlock full access to this review and 1229 more

Or log in here