Roles of artificial intelligence in the physical therapy practice: a scoping review / [author]: Arcenal, Lyka Shayne P. Arroyo, Denise Emmanuelle V. Bias, Dan Amiel D. Llosala, Raphael A. Medina, Hannah Danielle B. Mostacho, Nissos Ardel A [co author]: Marc Edward Garrido

Includes Appendix

ABSTRACT:
Introduction. Technological advances have now sprouted in different fields and continue its advancement in the healthcare system, including Artificial Intelligence (Al). Despite fear of job replacement, the technology, nowadays, undeniably
makes all the work efficient and accessible. Artificial Intelligence are devices made to replicate human intelligence with the goal of alleviating quality of life. As part of healthcare innovation, Al is now starting to be integrated in the physical therapy
practice. Physical therapy is a health specialty focused on re attainment and improvement of the function and well-being of people with disabilities. It generally involves examinations, assessments, evaluations, and exercises that vary from
different physiotherapist’s approaches to another. The lack of consistency and objectivity in the practice may undermine the overall patient care and outcomes. Thus, this study seeks to conduct a comprehensive research of the available literature on the roles of Artificial Intelligence (Al) in the Physical Therapy practice.
Methods. This study follows a Scoping Review Design according to the frameworks by Arksey and O’Malley (2005) and Joanna Briggs Institute (JBI) Approach to Conducting Scoping Review. In this review, 1,379 literatures were gathered from online databases, with specific search terms, such as: PubMed Cochrane, Embase, UpToDate, ClinicalTrials.gov, PEDro. The screening for eligibility has 2 levels: 1st is the screening for relevance to pass the inclusion criteria, specifically the titles, abstracts, and keywords, leaving 220 studies to undergo the 2nd screening. Next is the screening for relevance, where full text copies of the studies are scanned, leaving only 10 literatures to be appraised.
Specific details collected from each of the 10 studies were gathered using a tablethe obtained data are as follows: (1) Authors, Year, Country, (2) Title, (3) Artificial Intelligence Technology, (4) Research Design, (5) Research Methods, (6)
Application to PT Practice, (7) Results. The collected data was further examined and segregated based on their use in the (1) Assessment, (2) Treatment Part, and their (3) Limitations and Gaps.
Results and Discussion. Majority of the ten (10) obtained studies were (6) Randomized Controlled Trials, followed by (2) Meta Analysis, (1) Systematic Review and (1) Systematic Review and Meta Analysis. Al tools and equipment that were used in the clinic were OptiTrack, Al exercise mobile application, Al machine learning, Robot-Assisted Gait Training (RAGT) with Hybrid-Assistive Locomotion/Limb (HAL), Smart digital wristband, Overground (OG) Robotic Exoskeleton, wearable activity tracker, Honda Walking Assist (HWA), and mobile application. All Al devices were found to be helpful in the assessment and exercise
implementation in physical therapy, but should still be thoroughly studied to solve the problem in its applicability.
Conclusion. This study was able to obtain more knowledge about the clinical usage of Artificial Intelligence (Al) in the physical therapy practice, proving how its clinical use can improve efficiency in conducting assessments, monitoring status
and progress of patients during their rehabilitation journey, and in the application of enhanced techniques in treating patients. Hence, resulting in better patient care. Despite the benefits Al could provide, it also has its disadvantages as it could not
adapt to immediate changes unlike in conventional practice. The study came to a conclusion that this emerging technology still could not replace the modern physical therapy practice, but could only be a catalyst in amplifying overall competence of physical therapy rehabilitation.

School of Physical Therapy Thesis - School of Physical Therapy