|Year : 2022 | Volume
| Issue : 2 | Page : 256-261
Artificial intelligence in health professions education
P Ravi Shankar
IMU Centre for Education, International Medical University, Kuala Lumpur, Malaysia
|Date of Submission||09-Oct-2022|
|Date of Acceptance||17-Oct-2022|
|Date of Web Publication||23-Dec-2022|
Dr. P Ravi Shankar
IMU Centre for Education, International Medical University, Kuala Lumpur
Source of Support: None, Conflict of Interest: None
Artificial intelligence (AI) is widely used in medicine. AI may provide low-cost solutions to health problems and is especially important for developing countries. Health-care professionals will play an important role in providing data for educating AI systems and validating these through clinical trials. AI may necessitate changes in the different roles of a physician and possibly other professionals. Intelligent tutoring systems can support student learning by providing individualized feedback and creating personalized learning pathways. Role-plays with an intelligent active agent can enhance students' interaction with computers and activate their sense of responsibility. AI can support personalized learning by intelligent agents, autonomous scoring, and chatbots. AI has an important role to play in supporting simulations, serious games, and the gamification of learning. Learning analytics and educational data mining are two other important applications. Personalized prediction is also an important benefit. AI will supplement the work of educators and can reduce curricular overload by migrating some knowledge to AI algorithms. Routine tasks and responses to routine queries of learners can be provided by AI. AI can support continuing professional education by incorporating longitudinal and innovative formative assessment methods that can help identify knowledge and skill gaps and support learning. The use of AI in curriculum review and assessment has been limited. Data integrity and privacy are important issues to consider. Unconscious bias in the data used to educate AI systems is also possible. Most of the literature is from developed countries and among medical students and residents.
Keywords: Artificial intelligence, health professions education, individualized feedback, personalized learning pathways
|How to cite this article:|
Shankar P R. Artificial intelligence in health professions education. Arch Med Health Sci 2022;10:256-61
| Introduction|| |
Artificial intelligence (AI) implies the use of a computer to model intelligent behavior with minimal human intervention. AI in medicine can be divided into virtual and physical branches. The virtual component is characterized by mathematical algorithms that learn through experience. The physical branch of AI in medicine is characterized by physical objects, medical devices, and robots taking part in the delivery of health care. This article will focus on the virtual branch and explore AI in health professions education (HPE).
| Artificial Intelligence in Medical Care|| |
Two terms often mentioned in relation to AI are machine learning (ML) and deep learning (DL). In ML, a system learns from previously available data and the learning becomes stronger as more data are provided. DL is structured like the human brain with the capability of interpreting data at various levels and working on different datasets simultaneously until the output is produced. The ability of the system to learn by itself is an important characteristic. AI is being increasingly used in medicine and is being applied in dermatology, pathology, ophthalmology, and radiology among other disciplines. AI can improve health systems by improving workflow, has the potential to reduce medical errors, and may help patients process their own data to improve health. AI has the potential to provide low-cost solutions to health problems and will be especially important for low- and middle-income countries. AI is also being applied in cardiology, pulmonary medicine, endocrinology, nephrology, gastroenterology, neurology, and other specialties.
| Health-Care Professionals and Artificial Intelligence|| |
Health-care professionals (HCPs) will play an important role in providing data for educating AI systems and customizing AI for specific locations and situations. HCPs will play an important role in clinical trials to validate new AI systems and in addressing ethical considerations arising from the widespread use of AI in medicine. Education in AI is still not common in health professional schools. An article published in 2021 mentions that there are few concrete plans for incorporating AI into the medical curriculum. AI requires collaboration between several professionals including health care, data scientists, computer engineers, and others and this is still not very common. Medical students should be aware of the fundamental concepts of AI, and how it can reduce expenses, and improve the quality of and access to care. Changes brought about by AI will affect various roles of a physician. AI has been introduced as a triad element in patient–doctor communication. Communication between the doctor and the patient should also factor in the AI system. Regarding the collaborator role physicians should understand that the introduction of AI can empower other HCPs and patients. This may lead to a lessening or transformation of the leadership role doctors play in health-care teams. Furthermore, professions such as data specialists and computer scientists will play an important role in health care. As leaders and health advocates, doctors should work toward transparent and accountable implementation of AI technologies and work with disadvantaged groups to ensure that the adoption of AI is an empowered choice. There will be a change in professional identity and physicians will have to incorporate data science, information science, and engineering tools into their skill set. Learners must be exposed to big data in the context of decision-making, and they must understand the four Vs of big data. These are volume, variety, velocity, and veracity. Big data are constantly being generated in large amounts from a variety of sources at a high speed and the consistency, accuracy, and trustworthiness of data must be established. Students should also understand how data are aggregated, analyzed, and personalized in health-care delivery using AI applications.
| What Will be Addressed in this Article|| |
This article will not focus on how AI should/will be taught to health science students or the areas that should be addressed. We will also not focus further on AI's various uses and potential uses in health care. The focus will be on how AI technologies can be used to support student learning in health professions. The provision of individualized feedback, creating individual learning pathways, and supporting student learning and engagement will be the focus areas. Medical colleges and health science colleges of the future will be transformative rather than function as information providers and AI and ML will strongly influence the learning of students.
| Role of Artificial Intelligence in Supporting Student Learning|| |
AI can reduce the burden on both students and teachers and can offer students effective learning experiences. AI will play an important role in creating individualized learning experiences and in the gamification of learning. Intelligent tutoring systems (ITSs) are an important application of AI in education and HPE. ITS can provide personalized/individualized learning experiences in four ways. These are monitoring input provided by the student, providing appropriate tasks, providing feedback, and applying interfaces for human–computer interactions. In education, there is a growing realization that with the advent of AI the roles of teachers may need to be adjusted. Teachers will have to collaborate and work together with AI systems to advance student learning. Collaborating with AI will also be an important skill for HCPs.
| Providing Feedback to Learners|| |
Knowledge may need to be reorganized and represented according to the learner's reactions and understanding. Feedback is important to develop learners' proximal learning patterns. An artificial neural network is used to provide feedback to the learners so that they can gradually get used to abstract concepts and be able to perform practical exercises. Students can learn from the feedback provided when they are in error. Intelligent algorithms provide automatic feedback to students. Feedback is provided in a timely manner and specific to the student and the student's response. An intelligent virtual laboratory has been developed to provide appropriate feedback to students who face difficulties with completing their practical activities. A learning website, jutge.org was developed to provide feedback to learners and help students learn from their mistakes and progressively solve problems.
| Intelligent Tutoring Systems|| |
Visualization can help make complex topics interesting and entertaining and improve learners' motivation like in game-based learning. Extended/augmented/virtual reality can provide a highly simulated learning context that may be challenging to achieve in real life. The analysis of human emotions and feelings captured by sensors and affective algorithms known as affection computing can enhance human–computer interaction and ITS was modified to incorporate the detection of students' emotional status to also provide them with timely emotional feedback. Students assume various roles during role-plays and ponder on various problems with the activation of the affective domain also. Role-plays with an intelligent artificial agent were used to enhance learners' investment in their interaction with computers. Students' sense of responsibility was also activated, and they may work harder for their intelligent agent than they may do for themselves.
| Personalized Learning|| |
Educational games are a powerful learning method and are successful if the educational design, domain knowledge, gameplay, and affection elements are integrated. AI can help in the integration of the game and the knowledge elements, and help the game adapt in a dynamic manner to the players'/learners' emotional responses and actions. Educational games can take place in the setting of augmented, virtual or extended reality and complex scenarios difficult to recreate in the real world can be presented. In personalized learning, learning objectives, instructional approaches, and content can vary depending on the individual learner's needs. In personalized adaptive learning, the learners' data are continuously fed to the system followed by continuous assessments and real-time feedback. Most systems also incorporate a dashboard to help learners better evaluate their progress and identify potential obstacles. AI can support personalized learning through intelligent agents, autonomous scoring and assessment, and chatbots. We will discuss chatbots in greater detail shortly. AI compiles various information including students' prior learning and academic achievement and can help map out an individualized learning pathway in an economical manner.
| Learning Analytics|| |
Personalizing learning material improves the motivation of the learner and learning outcomes. Adaptive learning systems provide learners with the resources and tools to attain mastery at their own speed. Instructors are provided with information about learners' progress in a continuous manner. Learning analytics (LAs) uses data gathered from learners and analyzed to better understand and optimize the learning process. Data are automatically collected from learners' interactions with learning technologies, and this can be used to predict specific learning needs.
| Chatbots|| |
AI chatbots are being increasingly used in HPE. A chatbot is a machine conversation system that interacts with human users through natural, conversational language. Multiple applications for chatbots in education have been highlighted including quizzing learners about their existing knowledge, creating higher student engagement with a learning task, and mentoring students toward success. Metacognition chatbots help learners better understand their own learning styles and help them reflect on their coursework. Chatbots can provide a direct response to learners at any time of the day and can support health professions students 24 × 7 across geographically distributed locations. Repetitive learner questions can be dealt with effectively by chatbots and faculty may focus on questions requiring greater thought and input. Chatbots have been used to practice night call, interprofessional education consult, lab interpretations, and to learn advanced anatomical concepts. A study skills chatbot provides learners with evidence-based studying strategies to improve their understanding of the content. A faculty development chatbot provides faculty members with just-in-time evidence-based answers to faculty development issues. Most teaching practices can be effectively offered in short bursts demonstrating the skills involved. Chatbots can advance interactions with virtual patients to a true conversation rather than computer-based text interactions and may better develop learner competencies.
| Learning Analytics in Workplace-Based Learning|| |
LAs has been used more in traditional formal education formats. Its use in workplace-based learning has been less. Workplace-based learning is experiential, social, situated, and practice-bound. Data-informed automated feedback can provoke reflection leading to a reframing of oneself as a part of professional growth. Professional learners must chart their learning needs and align these with the needs of the workplace. As the workplace needs change, they must reframe their learning needs. LA must address challenges regarding how these needs can be scaffolded, supported, and enhanced in the workplace. As AI technologies constantly advance LA can be offered in more complex, nontraditional workplace settings. Skills analytics, reflective writing analytics, and dispositional analytics can be helpful. Analytics tools can help the professional reflect on where she/he is on the career journey, and what skills should she/he aspire to acquire. Reflective writing analytics helps with making sense of challenging experiences, changing as a professional, and how to handle professional dilemmas better next time. Dispositional analytics focuses on professional purpose, how to learn better, and how to transfer formal education and training to the job.
| Serious Games and Simulations|| |
Many simulation approaches have been developed to educate students to deliver health-care safely. Serious games and simulations help students learn new skills and experience in real-time the consequences of their decisions and address system anomalies while not putting patients at risk. In addition to inexperienced health-care workers simulations can also help experienced workers learn new skills. Difficult procedures can be rehearsed multiple times and training on patient complications and equipment failure (which may be rare in practice) requiring decisive action and skilled performance can be provided. AI can support serious games and simulations. Gamification is the use of game elements in a nongame context. In HPE, gamification can improve learning behaviors and attitudes toward learning. Game elements such as points, leaderboards, and prizes are used in a nongame context. AI is being increasingly used to support gamification.
| Educational Data|| |
Education data mining (EDM) explores the unique types of data resulting from educational settings to better understand students and their learning environments. EDM focuses on modeling and predicting student progress and creating computer systems that can adapt without needing human intervention in the learning cycle. HPE programs have complex learning situations and span both academic institutions and clinical practice settings. There are several challenges in obtaining data for LA and EDM. Among these is the fragmentation of data across different systems, the data not being continuously updated, being out of sync, and data volume may be low within individual HPE programs. Data fragmentation can be addressed with systems being able to talk to each other, and use of AI systems in assessment may result in assessment results being made available faster. HPE data can be messy. The personalized prediction uses various methods with existing data to predict future outcomes. Health professions schools would like to predict the characteristics of students who are most likely to be successful practitioners; they may want to predict students who are most likely to succeed in licensing examinations and match into postgraduate and fellowship programs. These may use longitudinal datasets over multiple years to improve predictive accuracy.
Data-based methods in HPE are reductionist in nature and this may be a significant limitation, especially considering the importance now provided to noncognitive attributes like empathy, collaboration, and communication skills in HPE. There are also concerns related to data privacy and how the data are obtained as most of this is obtained when students use required learning systems as they may not be in a position to refuse to provide the data.
| Use of Artificial Intelligence in Assessing Learners|| |
AI can be used to automate performance assessment, provide feedback, and predict patient outcomes. ML algorithms can be used to improve the efficiency of assessing surgical skills. Surgeons' experience levels were used to train an algorithm to distinguish between different levels of operative skills. A novel motion tracking system and algorithm were developed to automatically evaluate trainee performance at a pediatric laparoscopic suturing task. ML was used to assess the level of neurosurgical skills among postgraduate learners on a simulator.
| Artificial Intelligence and Health Professions Educators|| |
AI has the potential to carry out routine administrative tasks and repetitive jobs and can free HPE educators to focus on tasks requiring creativity and specialized knowledge. The authors of this paper mention that the basic premise of AI is to supplement and not supplant the work of educators and misunderstanding this concept may be responsible for the hesitancy to adopt AI along with resistance to change. With rapid development and incorporation of new technologies health profession education curricula suffer from information overload. However, the addition of AI can reduce curricular overload by migrating some biomedical and clinical knowledge to AI algorithms. We should carefully examine the amount and type of information that HCPs should memorize in a world where information can be continually accessed without difficulty. Educators using AI should be aware of both the advantages and the risk of bias in big data and algorithms. Data collected by humans can impact the efficacy of AI algorithms and can introduce unconscious bias. The Georgia Institute of Technology in the United States is using an AI application, called Jill Watson to function as a teaching assistant optimizing the use of scarce human resources.
| Artificial Intelligence and Continuing Professional Education|| |
Competency-based approaches are becoming common in HPE. These approaches rely on rich programmatic data bout each learner. Obtaining and managing this vast amount of data and using it to guide future learning has been a challenge. AI enables precision education by identifying individual performance trends and supporting individualized learning pathways. AI can support continuing professional education by incorporating longitudinal and innovative formative assessment methods that can help identify knowledge and skill gaps and support learning. AI can monitor a health-care provider's patient panel and outcomes and recommend appropriate educational resources that can be provided just in time. The Amplifire's platform mines health system metrics and electronic health record data to identify opportunities for improvement and learning that can be used to design system-wide training across the professions.
| Status of Artificial Intelligence in Health Professions Education|| |
The workload of health professions educators can be optimized by automating some processes, thus reducing stress and burnout. Health professions students at risk of suicide can be identified through natural language processing algorithmic models and a smartphone application. A recent review concludes that AI till now, has primarily provided individualized feedback to students. AI has not yet been used in curriculum review and this could be due to the limited digitization of learning management systems which is essential to create a digital curriculum map. The use of AI in assessments has also been limited and this could be due to the lack of availability of the necessary data pool necessary to train AI systems and the sensitive nature of assessments. Data integrity and data privacy are also important issues to be addressed. [Table 1] highlights certain examples of the use of AI in HPE.
|Table 1: Specific instances of using artificial intelligence in health professions education|
Click here to view
The literature shows that AI has been predominantly used in developed nations. There is a scarcity of studies from the developing world. The studies highlighted were predominantly from medicine. HPE has to fulfill strict accreditation and regulatory criteria. Hence, uptake of AI may be slower. Studies from other health professions are required.
| Conclusion|| |
AI is being widely used in health care. HCPs have an important role in educating AI systems, and AI will impact the various roles of a physician. ITS is an important application of AI in education and can personalize the learning experience for students. AI is being used to provide individualized learner feedback. AI can also help with serious games, simulations, and gamification of learning. Chatbots, LAs, and educational data mining are used commonly. AI has also been tried to assess learners. AI can support continuing professional education. Most studies have been from developed nations and among physicians and medical students. Studies among other students and professionals from developing countries are required.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Hamet P, Tremblay J. Artificial intelligence in medicine. Metabolism 2017;69S: S36-40.
Mainali G. Artificial intelligence in medical science: Perspective from a medical student. JNMA J Nepal Med Assoc 2020;58:709-11.
Mintz Y, Brodie R. Introduction to artificial intelligence in medicine. Minim Invasive Ther Allied Technol 2019;28:73-81.
Kulkarni S, Seneviratne N, Baig MS, Khan AH. Artificial intelligence in medicine: Where are we now? Acad Radiol 2020;27:62-70.
Topol EJ. High-performance medicine: The convergence of human and artificial intelligence. Nat Med 2019;25:44-56.
Shankar PR. Artificial intelligence in healthcare in Nepal. Janaki Med Coll J Med Sci 2022;10:1-4.
Briganti G, Le Moine O. Artificial intelligence in medicine: Today and tomorrow. Front Med (Lausanne) 2020;7:27.
Grunhut J, Wyatt AT, Marques O. Educating future physicians in artificial intelligence (AI): An integrative review and proposed changes. J Med Educ Curric Dev 2021;8:23821205211036836.
Khurana M. Keeping pace: The need for digital health education in medical schools. Acad Med 2020;95:1629-30.
Rampton V, Mittelman M, Goldhahn J. Implications of artificial intelligence for medical education. Lancet Digit Health 2020;2:e111-2.
Hamdy H. Medical college of the future: From informative to transformative. Med Teach 2018;40:986-9.
Loeckx J. Blurring boundaries in education: Context and impact of MOOCs. Int Rev Res Open Distrib Learn 2016;17:92-121.
Zhai X, Chu X, Chai CS, Jong MS, Istenic A, Spector M, et al
. A review of artificial intelligence (AI) in education from 2010 to 2020. Complexity 2021;2021:8812542.
Boulay B. Artificial intelligence as an effective classroom assistant. IEEE Intell Syst 2016;31:76-81.
Fenwick T. Pondering purposes, propelling forwards. Stud Contin Educ 2018;40:367-80.
Melo FR, Flores EL, Carvalho SD, Teixeira RA, Loja LF, de Sousa Gomide R. Computational organization of didactic contents for personalized virtual learning environments. Comput Educ 2014;79:126-37.
Petit J, Roura S, Carmona J, Cortadella J, Duch J, Gimnez O. Jutge.org: Characteristics and experiences. IEEE Trans Learn Technol 2018;11:321-33.
Horakova T, Houska M, Domeova L. Classification of the educational texts styles with the methods of artificial intelligence. J Baltic Sci Educ 2017;16:324-36.
Lin HC, Wang CH, Chao CJ, Chien MK. Employing textual and facial emotion recognition to design an affective tutoring system. Turk Online J Educ Technol 2012;11:418-26.
Chase CC, Chin DB, Oppezzo MA, Schwartz DL. Teachable agents and the protégé effect: Increasing the effort towards learning. J Sci Educ Technol 2009;18:334-52.
Thomas JM, Young RM. Annie: Automated generation of adaptive learner guidance for fun serious games. IEEE Trans Learn Technol 2010;3:329-43.
Palanisamy P, Thilarajah S, Chen Z. Providing equitable education through personalised adaptive learning and assessment. In: Back to the Future – ASCILITE. UK: ASCILITE; 2021.
Peng H, Ma S, Spector JM. Personalized adaptive learning: An emerging pedagogical approach enabled by a smart learning environment. Smart Learn Environ 2019;6:1-4.
Khan AI, Al-Badi A. Emerging data sources in decision making and AI. Procedia Comput Sci 2020;177:318-23.
Corral J. Artificially intelligent chatbots for health professions education. In: Digital Innovations in Healthcare Education and Training. Cambridge, Massachusetts: Academic Press; 2021. p. 127-35.
Corral J, Post MD, Bradford A. Just-in-time faculty development for pathology small groups. Med Sci Educ 2018;28:11-2.
Shum SB, Littlejohn A, Kitto K, Crick R. Framing professional learning analytics as reframing oneself. IEEE Trans Learn Technol 2022;15.
Lin AJ, Chen M, Cheng FF. Game for health professional education. Int J Inf Educ Technol 2016;6:972-5.
van Gaalen AEJ, Brouwer J, Schönrock-Adema J, Bouwkamp-Timmer T, Jaarsma AD, Georgiadis JR. Gamification of health professions education: A systematic review. Adv Health Sci Educ Theory Pract 2021;26:683-711.
Corral J, Konstantinidis ST, Bamidis PD. Learning analytics, education data mining, and personalization in health professions education. In: Digital Innovations in Healthcare Education and Training. Cambridge, Massachusetts: Academic Press; 2021. p. 137-50.
Christopher B, Thompson C. Predictive modelling in teaching and learning. In: Handbook of Learning Analytics. Society for Learning Analytics Research; 2017. p. 61-8.
Frank JR, Snell LS, Cate OT, Holmboe ES, Carraccio C, Swing SR, et al.
Competency-based medical education: Theory to practice. Med Teach 2010;32:638-45.
Tolsgaard MG, Boscardin CK, Park YS, Cuddy MM, Sebok-Syer SS. The role of data science and machine learning in Health Professions Education: Practical applications, theoretical contributions, and epistemic beliefs. Adv Health Sci Educ Theory Pract 2020;25:1057-86.
Uemura M, Tomikawa M, Miao T, Souzaki R, Ieiri S, Akahoshi T, et al.
Feasibility of an AI-based measure of the hand motions of expert and novice surgeons. Comput Math Methods Med 2018;2018:9873273.
Oquendo YA, Riddle EW, Hiller D, Blinman TA, Kuchenbecker KJ. Automatically rating trainee skill at a pediatric laparoscopic suturing task. Surg Endosc 2018;32:1840-57.
Winkler-Schwartz A, Yilmaz R, Mirchi N, Bissonnette V, Ledwos N, Siyar S, et al.
Machine learning identification of surgical and operative factors associated with surgical expertise in virtual reality simulation. JAMA Netw Open 2019;2:e198363.
Lomis K, Jeffries P, Palatta A, Sage M, Sheikh J, Sheperis C, et al.
Artificial intelligence for health professions educators. NAM Perspect 2021;2021.
Wartman SA, Combs CD. Reimagining medical education in the age of AI. AMA J Ethics 2019;21:E146-52.
Eicher B, Polepeddi L, Goel A. Jill Watson Doesn't Care if You're Pregnant: Grounding AI Ethics in Empirical Studies. In Proceedings of the 2018 AAAI/ACM Conference on AI, Ethics, and Society; 2018. p. 88-94.
Coentre R, Góis C. Suicidal ideation in medical students: Recent insights. Adv Med Educ Pract 2018;9:873-80.
Chan KS, Zary N. Applications and challenges of implementing artificial intelligence in medical education: Integrative review. JMIR Med Educ 2019;5:e13930.
Hung AJ, Chen J, Che Z, Nilanon T, Jarc A, Titus M, et al.
Utilizing machine learning and automated performance metrics to evaluate robot-assisted radical prostatectomy performance and predict outcomes. J Endourol 2018;32:438-44.
Ariaeinejad A, Samavi R, Chan TM, Doyle TE. A Performance Predictive Model for Emergency Medicine Residents. In Proceedings of the 27th
Annual International Conference on Computer Science and Software Engineering; 2017. p. 28-37.
Chen F, Arora H, Zvara DA, Connolly A, Martinelli SM. Anesthesia myTIPreport: A web-based tool for real-time evaluation of accreditation council for graduate medical education's milestone competencies and clinical feedback to residents. A A Pract 2019;12:412-5.
Hamdy H, Sreedharan J, Rotgans JI, Zary N, Bahous SA, Venkatramana M, et al.
Virtual Clinical Encounter Examination (VICEE): A novel approach for assessing medical students' non-psychomotor clinical competency. Med Teach 2021;43:1203-9.
Khumrina P, Ryanb A, Juddb T, Verspoora K. Diagnostic Machine Learning Models for Acute Abdominal Pain: Towards an e-Learning Tool for Medical Students. In: Proceedings of the 16th
World Congress of Medical and Health Informatics; 2018. p. 447.
McFadden P, Crim A. Comparison of the effectiveness of interactive didactic lecture versus online simulation-based CME programs directed at improving the diagnostic capabilities of primary care practitioners. J Contin Educ Health Prof 2016;36:32-7.
Kolachalama VB, Garg PS. Machine learning and medical education. NPJ Digit Med 2018;1:54.
Chen Y, Wrenn J, Xu H, Spickard A 3rd
, Habermann R, Powers J, et al.
Automated assessment of medical students' clinical exposures according to AAMC geriatric competencies. AMIA Annu Symp Proc 2014;2014:375-84.