An artificial intelligence platform for movement analysis and rehabilitation: Clinical applications of stepsense to complex pain and Long COVID

Adar Pelah*, Viswadeep Sarangi, Elsje de Villiers, Nicholas Shenker, Thomas Stone, Peter Estiebeiro, Elan Barenholtz, Ximena Levy, Gregg Fields

*Corresponding author for this work

Research output: Contribution to journalConference articlepeer-review

Abstract

Purpose: The study describes two exemplar clinical applications of the StepSense platform to virtual reality (VR) biofeedback rehabilitation in complex pain and to assessment of gait and balance in post-acute Covid-19 (‘Long COVID’). Methods: StepSense is a low-cost, high performance gait and movement analysis system that uses machine learning (ML) to extract 3-dimensional (x–y–z) skeletal joints during patient movement from plain 2-dimensional video. The captured musculoskeletal (MSK) movements animate a rendered avatar seen within a VR environment to provide a gamified biofeedback display in real-time. Examples will be given for each of the platform's three components: StepSense Clinic, deployed for MSK or neurological assessment and VR-rehabilitation in hospital or clinical settings; StepSense Home, a smartphone-based app for assessment and real-time VR rehabilitation in a person's home or in care-home facility; and StepSense Lab, a cloud-centred telemedicine infrastructure that provides clinicians with browser-based data visualisation, clinical evaluation and ML analytics to assist with decisions on diagnostics and therapy. Results: Chronic pain conditions such as fibromyalgia, lower back pain (LBP) and complex regional pain syndrome (CRPS) share common features of motor neglect and affect hundreds of thousands of people in the UK that incur high costs to the economy and the National Health Service. In a single-site clinical trial using StepSense technology (Brain Sci. 2021, 11, 4), 10 participants with one or more of the above conditions were randomly allocated to intervention (VR biofeedback) or control (no VR biofeedback) groups and underwent a treadmill task three times per week for two weeks. Primary outcomes of distance walked (at baseline compared to the final 5-minute cycle of week 2) and the Lower Extremity Functional Index (LEFI) questionnaire were evaluated. Conclusion(s): In the complex pain study, distance walked was significantly higher in the intervention group (p < 0.05) with 33% (2/6) reporting clinically improved LEFI improvement at week 2 compared to 0% (0/4) in the control group. The intervention group received significantly higher satisfaction scores that controls on follow-up at week 24. Additional findings are reported on gait, balance and cognition effects in persons with Long COVID compared with healthy controls. Impact: While signs of the UK emerging from the coronavirus pandemic are promising, a growing number of people continue to suffer from post-acute manifestations of the disease. According to the Office of National Statistics, over a million people report having symptoms after four weeks while 1 in 7 still report symptoms 12 weeks later; 20% of people report that symptoms such as chronic fatigue, joint pain and ‘brain fog’ are adversely affecting their daily lives. The growing prevalence of ‘Long COVID’ sufferers has prompted NHS England to launch about 70 specialist clinics, yet with over 20 distinct problems being observed the condition is poorly understood and treatments are illusive. The StepSense platform is an AI-driven technology that provides an end-to-end solution for the measurement, clinical evaluation, treatment and investigative analysis of gait and balance in conditions such as Long COVID as well as other manifestations of health and disease that cut across healthcare.

 Funding acknowledgements: Funded in part by a grant to AP under the Grow MedTech Proof of Feasibility programme, supported by UKRI Research England's Connecting Capability Fund [project code: CCF11-7795].
Original languageEnglish
Pages (from-to)E74-E75
Number of pages2
JournalPhysiotherapy
Volume114
DOIs
Publication statusPublished - 16 Feb 2022

Keywords

  • Long COVID
  • Chronic Pain/therapy
  • Gait analysis
  • Artificial intelligence
  • Machine learning
  • Complex pain

Cite this