Using artificial intelligence for exercise prescription in personalised health promotion: A critical evaluation of OpenAI's GPT-4 model

Ismail Dergaa; Helmi Ben Saad; Abdelfatteh El Omri; Jordan M Glenn; Cain C T Clark; Jad Adrian Washif; Noomen Guelmami; Omar Hammouda; Ramzi A Al-Horani; Luis Felipe Reynoso-Sánchez; Mohamed Romdhani; Laisa Liane Paineiras-Domingos; Rodrigo L Vancini; Morteza Taheri; Leonardo Jose Mataruna-Dos-Santos; Khaled Trabelsi; Hamdi Chtourou; Makram Zghibi; Özgür Eken; Sarya Swed; Mohamed Ben Aissa; Hossam H Shawki; Hesham R El-Seedi; Iñigo Mujika; Stephen Seiler; Piotr Zmijewski; David B Pyne; Beat Knechtle; Irfan M Asif; Jonathan A Drezner; Øyvind Sandbakk; Karim Chamari

doi:10.5114/biolsport.2024.133661

Using artificial intelligence for exercise prescription in personalised health promotion: A critical evaluation of OpenAI's GPT-4 model

Biol Sport. 2024 Mar;41(2):221-241. doi: 10.5114/biolsport.2024.133661. Epub 2023 Dec 13.

Authors

Ismail Dergaa^{1

2

3}, Helmi Ben Saad^{4

5}, Abdelfatteh El Omri⁶, Jordan M Glenn⁷, Cain C T Clark^{8

9}, Jad Adrian Washif¹⁰, Noomen Guelmami^{3

11}, Omar Hammouda^{12

13}, Ramzi A Al-Horani¹⁴, Luis Felipe Reynoso-Sánchez¹⁵, Mohamed Romdhani¹², Laisa Liane Paineiras-Domingos¹⁶, Rodrigo L Vancini¹⁷, Morteza Taheri¹⁸, Leonardo Jose Mataruna-Dos-Santos¹⁹, Khaled Trabelsi², Hamdi Chtourou², Makram Zghibi³, Özgür Eken²⁰, Sarya Swed²¹, Mohamed Ben Aissa¹¹, Hossam H Shawki²², Hesham R El-Seedi^{23

24

24}, Iñigo Mujika^{25

26}, Stephen Seiler²⁷, Piotr Zmijewski²⁸, David B Pyne²⁹, Beat Knechtle³⁰, Irfan M Asif³¹, Jonathan A Drezner³², Øyvind Sandbakk³³, Karim Chamari³⁴

Affiliations

¹ Primary Health Care Corporation (PHCC), Doha, Qatar.
² Research Laboratory Education, Motricité, Sport et Santé (EM2S) LR19JS01, High Institute of Sport and Physical Education of Sfax, University of Sfax, Sfax 3000, Tunisia.
³ High Institute of Sport and Physical Education of Kef, Jendouba, Kef, Tunisia.
⁴ University of Sousse, Farhat HACHED hospital, Research Laboratory LR12SP09 «Heart Failure», Sousse, Tunisia.
⁵ University of Sousse, Faculty of Medicine of Sousse, laboratory of Physiology, Sousse, Tunisia.
⁶ Surgical Research Section, Department of Surgery, Hamad Medical Corporation, Doha 3050, Qatar.
⁷ Neurotrack Technologies, Redwood City CA, USA.
⁸ College of Life Sciences, Birmingham City University, Birmingham, B15 3TN, UK.
⁹ Institute for Health and Wellbeing, Coventry University, Coventry, CV1 5FB, UK.
¹⁰ Sports Performance Division, National Sports Institute of Malaysia, Kuala Lumpur, Malaysia.
¹¹ Postgraduate School of Public Health, Department of Health Sciences (DISSAL), University of Genoa, Genoa, Italy.
¹² Interdisciplinary Laboratory in Neurosciences, Physiology and Psychology: Physical Activity, Health and Learning (LINP2), UFR STAPS (Faculty of Sport Sciences), UPL, Paris Nanterre University, Nanterre, France.
¹³ Research Laboratory, Molecular Bases of Human Pathology, LR19ES13, Faculty of Medicine, University of Sfax, Tunisia.
¹⁴ Department of Exercise science, Yarmouk University, Irbid, Jordan.
¹⁵ Department of Social Sciences and Humanities, Autonomous University of Occident, Los Mochis, Mexico.
¹⁶ Departamento de Fisioterapia, Instituto Multidisciplinar de Reabilitação e Saúde, Universidade Federal da Bahia, Brazil.
¹⁷ Centro de Educação Física e Desportos, Universidade Federal do Espírito Santo, Vitória, Espírito Santo, Brazil.
¹⁸ Department of Motor Behavior, Faculty of Sport Sciences, University of Tehran, Tehran, Iran.
¹⁹ Department of Creative Industries, Faculty of Communication, Arts and Sciences, Canadian University of Dubai, Dubai, United Arab Emirates.
²⁰ Department of Physical Education and Sport Teaching, Inonu University, Malatya 44000, Turkey.
²¹ University of Aleppo Faculty of Medicine: Aleppo, Aleppo Governorate, Syria.
²² Department of Comparative and Experimental Medicine, Nagoya City University Graduate School of Medical Sciences, Nagoya 467-8601, Japan.
²³ Department of Chemistry, Faculty of Science, Islamic University of Madinah, Madinah, 42351, Saudi Arabia.
²⁴ International Research Center for Food Nutrition and Safety, Jiangsu University, Zhenjiang 212013, China.
²⁵ Department of Physiology, Faculty of Medicine and Nursing, University of the Basque Country, Leioa, Basque Country.
²⁶ Exercise Science Laboratory, School of Kinesiology, Faculty of Medicine, Universidad Finis Terrae, Santiago, Chile.
²⁷ Department of Sport Science and Physical Education, University of Agder, Kristiansand, Norway.
²⁸ Jozef Pilsudski University of Physical Education in Warsaw, Warsaw, Poland.
²⁹ Research Institute for Sport and Exercise, University of Canberra, Canberra, ACT, Australia.
³⁰ Institute of Primary Care, University of Zurich, Zurich, Switzerland.
³¹ Department of Family and Community Medicine, University of Alabama at Birmingham, Birmingham, Alabama, USA.
³² Center for Sports Cardiology, University of Washington, Seattle, Washington, USA.
³³ Center for Elite Sports Research, Department of Neuromedicine and Movement Science, Norwegian University of Science and Technology, Trondheim, Norway.
³⁴ Higher institute of Sport and Physical Education, ISSEP Ksar Saïd, Manouba University, Tunisia.

Abstract

The rise of artificial intelligence (AI) applications in healthcare provides new possibilities for personalized health management. AI-based fitness applications are becoming more common, facilitating the opportunity for individualised exercise prescription. However, the use of AI carries the risk of inadequate expert supervision, and the efficacy and validity of such applications have not been thoroughly investigated, particularly in the context of diverse health conditions. The aim of the study was to critically assess the efficacy of exercise prescriptions generated by OpenAI's Generative Pre-Trained Transformer 4 (GPT-4) model for five example patient profiles with diverse health conditions and fitness goals. Our focus was to assess the model's ability to generate exercise prescriptions based on a singular, initial interaction, akin to a typical user experience. The evaluation was conducted by leading experts in the field of exercise prescription. Five distinct scenarios were formulated, each representing a hypothetical individual with a specific health condition and fitness objective. Upon receiving details of each individual, the GPT-4 model was tasked with generating a 30-day exercise program. These AI-derived exercise programs were subsequently subjected to a thorough evaluation by experts in exercise prescription. The evaluation encompassed adherence to established principles of frequency, intensity, time, and exercise type; integration of perceived exertion levels; consideration for medication intake and the respective medical condition; and the extent of program individualization tailored to each hypothetical profile. The AI model could create general safety-conscious exercise programs for various scenarios. However, the AI-generated exercise prescriptions lacked precision in addressing individual health conditions and goals, often prioritizing excessive safety over the effectiveness of training. The AI-based approach aimed to ensure patient improvement through gradual increases in training load and intensity, but the model's potential to fine-tune its recommendations through ongoing interaction was not fully satisfying. AI technologies, in their current state, can serve as supplemental tools in exercise prescription, particularly in enhancing accessibility for individuals unable to access, often costly, professional advice. However, AI technologies are not yet recommended as a substitute for personalized, progressive, and health condition-specific prescriptions provided by healthcare and fitness professionals. Further research is needed to explore more interactive use of AI models and integration of real-time physiological feedback.

Keywords: AI Challenges; AI Evaluation; ChatGPT; Chatbot; Digital Health; Exercise Optimization; Fitness Algorithms; Machine Learning; Personalized Medicine; Real-time Monitoring.