|Year : 2021 | Volume
| Issue : 2 | Page : 55-57
Artificial intelligence in physiotherapy
Rajeev Aggarwal1, Suvarna Shyam Ganvir2
1 Senior Physiotherapist, Neuro-Physiotherapy Unit, Department of Neurology, All India Institute of Medical Sciences, New Delhi, India
2 Department of Neurophysiotherapy, DVVPF's College of Physiotherapy, Ahmednagar, Maharashtra, India
|Date of Submission||29-Jan-2022|
|Date of Acceptance||29-Jan-2022|
|Date of Web Publication||15-Feb-2022|
Dr. Rajeev Aggarwal
Senior Physiotherapist, Neuro-Physiotherapy Unit, Department of Neurology, All India Institute of Medical Sciences, New Delhi
Source of Support: None, Conflict of Interest: None
|How to cite this article:|
Aggarwal R, Ganvir SS. Artificial intelligence in physiotherapy. Physiother - J Indian Assoc Physiother 2021;15:55-7
Computers have become an integral part of our lives. Research and development (R&D) is heading to more and more automation, minimizing the need and interference of humans to run the system. Corona epidemic has brought humans closer to computers and machines. Due to this epidemic, the health sector has gotten unprecedented attention and focus of policymakers. Advancement in health sector is converging to more evidence-based practice, improving efficiency, minimizing errors, and resolving emergent challenges. Usage of computers in physiotherapy practice and research has been instrumental. Recordkeeping, data search, data analysis, algorithm-based flowcharts, norm-based data, billing, appointment system, etc., are widely used applications of computers. The need for artificial intelligence (AI) in the health sector has arisen due to the high cost of training of professionals, limitation of specialists due to specific set of skills, and lack of transparency in their work.
New-generation computers are now manufactured with “machine learning” that is capacity to learn and think without explicit programming. This learning and thinking capacity of the computer is known as AI. Internet connectivity and renewable energy are considered the fourth industrial revolution by many after coal, gas, electronic, and nuclear revolutions. The vast usage of internet in our day-to-day lives has led to the development of AI and digitalization. We have witnessed the impact of digitalization in our lives. Education, banking, trading, and archive of documents are few such examples of digitalization. The data so collected by computers are paving the way to a new set of skills, professionals, and industry.
Computers may learn in a controlled and supervised manner where labeling is done to a data set. Physiotherapists use this feature to diagnose based on present criteria that may include presenting symptoms, demographic details of the client, relevant history, clinical examination findings, special tests, and various investigations. Computers collect data from various sources and may lead to unsupervised learning based on the analysis of “big data.” Classification, analysis, differentiation, summarization, and extraction from big data may explain some unanswered questions in the past. This type of learning by computers is considered descriptive learning. Physiotherapists may get answers to various clinical presentations in their area based on morphological characteristics, anthropological characteristics, genetic predisposition, food habits, and living standards of the population. For example, why low back pain is more common or why early osteoarthritis is seen in any particular population? Predictive learning model is used by computers to predict an outcome based on big data.
There is 2.1% chance that Physical Therapists will be replaced by robots, which is negligible. However, AI has become an integral part of physiotherapy assessment and treatment even though there is a scarcity of good-quality researches investigating effectiveness of AI.
AI has been extensively used in Physiotherapy assessment, the common example of which can be gait analysis. Recent progress in video analysis driven by machine learning has shown that computers are able to automate the diagnosis of gait abnormalities and underlying pathology, for example, in patients with Parkinson's disease and osteoarthritis. Patients at increased risk of falling can be identified early based on constant gait analysis, and changes in gait patterns (indicators of changes in pathology) can be highlighted to relevant staff. Another aspect is virtual personal assistants which are being increasingly embedded within smartphones and watches, which help track the changes in bodily parameters. However, extensive scientific research is needed to establish its validity. Natural language processing (NLP) is another feather in the cap of AI where computers can understand, intercept, analyze, process, and reproduce the data in a lucid, semantic format. Using NLP, patient–clinician conversation may be used to automatically produce robust reports as patient information sheets, clinical summaries, treatment plans, and prognostication charts.
AI in physiotherapy education may cause a whole paradigm shift. The use of technology to understand anatomy, physiology, clinical signs, clinical tests, investigations, therapeutic interventions, and electrical modalities will increase the physiotherapists' knowledge exponentially. The use of robotics and automated mannequins will curtail the errors and mishandling of patients by physiotherapy students and learners. The permutations and combinations of various therapeutic interventions such as electrotherapy, actinotherapy, mobilization, therapeutic exercise, and their effects will help the learner infer a customized or hypothetical treatment regimen. The use of AI in elucidating radiological imaging may be a leap in therapists' knowledge and skill. As intelligent algorithm, machine learning and NLP are smarter than most of us in certain limited domain and successful clinical practice in the future will depend on our understanding of these technologies, so inclusion of AI in the physiotherapy course curriculum will be a futuristic move. Physiotherapists should be smart to make use of AI in their clinical practice, develop wisdom about accepting or rejecting the advice of AI, and provide new data set to computers for improvising the AI in physiotherapy.
In physiotherapy treatment area, Dextrous or soft robot hands have been used in providing simple mobilization in patients with musculoskeletal dysfunction. However, its long-term efficacy has not been studied. Storing, access, and analysis of everything about the pathologies related to human illness help to provide clinical decision support to enhance decision-making.
A study has used machine learning because it has the potential to involve physiotherapy practice through specific diagnostic, decision-making, and measurement for physiotherapy interventions. X-rays, computed tomography scans, and magnetic resonance imaging reading can be done with great precision using AI. Home-based physiotherapy by digital therapist – real-time observation, advice, and tracking remotely by therapist – is being used successfully. AI enabled robotics that can guide the patient for correct movement as well as assist the patient to do the movements will be easily available in the near future.
In India, National Research Foundation, an autonomous body under the new National Education Policy 2020, has been established to boost research across segments, including AI. This will strengthen the governance structure of the research-related institutions and will improve linkages between R&D, academia, and industry. Centre for Artificial Intelligence and Robotics, a laboratory of the Defense Research Development Organization, in 2014 for R&D in AI, robotics, command and control, networking, information, and communication security shoots for the development of mission-critical products for battlefield communication and management systems. Many high-end technologies in health sector are initially researched and developed in defense sector. We envisage that physiotherapists will have better opportunities for incorporating the AI technology in routine clinical practice with different types of apps.
Every coin has two sides, and so does AI. It has its own limitations in the form of requirement of accessory technological support, lack of human touch, and restricted scope of interaction with patients, thus hindering communication. Potential threats to its application include unwarranted dependency on technology by therapist as well as patient.
Ethical consideration has been a debatable issue in the application of AI in routine clinical setup. Potential for biasing to be built into AI technologies, data storage and access, and delegating the responsibility in case of any adverse event are the primary ethical issues. AI technologies developed on algorithm or data input may not be sensitive about cultural or societal differences. The output of collected data by algorithms used by the AI technology is not clearly understood. To defend the decisions taken by AI technology may not be explicable. Biasing of AI technology provider by using the data and directing the system to interpret a set of data may be a conflict of interest. Weightage may be added by biased game players to promote a specific approach-like conservative or surgical management of a particular condition. Privacy and confidentiality while storing the data can be a concern for patients. Lack of specific training in delivering patient care can be an ethical issue as it may cause harm to some patients. Patient's trust in technology may affect the output of care given. However, technology can and should be used as a supplement to the traditional treatment approach to provide maximum possible quality healthcare.
Change is the only constant in life, and resisting the change leads to stagnation. Constant learning, accepting the change, and moving ahead are the ways of life. Albert Einstein had quoted, “Learn the rules of the game and then play better than anyone else.” Accepting and adopting AI in physiotherapy practice and learning is the next-generation movement. It is imperative to understand the intricacies of AI, its strengths, its limitations, and its utility to improvise patient care and maintain professional dignity.
| References|| |
Kidziński Ł, Delp S, Schwartz M. Automatic real-time gait event detection in children using deep neural networks. PLoS One 2019;14:e0211466.
Pedersen I, Iliadis A. Embodied Computing: Wearables, Implantables, Embeddables, Ingestibles. Cambridge: MIT Press; 2020.
Choi C, Schwarting W, DelPreto J, Rus D. Learning object grasping for soft robot hands. IEEE Robot Autom Lett 2018;3:2370-7.
Shortliffe EH, Sepúlveda MJ. Clinical decision support in the era of artificial intelligence. JAMA 2018;320:2199-200.
Godse SP, Singh S, Khule S, Yadav V, Wakhare S. Musculoskeletal physiotherapy using artificial intelligence and machine learning. Int J Innov Sci Res Technol 2019;4:592-8.
Rowe M, Nicholls DA, Shaw J. How to replace a physiotherapist: Artificial intelligence and the redistribution of expertise. Physiother Theory Pract 2021;3:1-9. doi: 10.1080/09593985.2021.1934924. Epub ahead of print. PMID: 34081573.