Sage Journals: Discover world-class research

Abstract

Artificial intelligence (AI) with its diverse domains such as expert systems and machine learning already has multiple potential applications in medicine.

Based on the latest developments in the multifaceted field of AI, it will play a pivotal role in medicine, with a high transformative potential in multiple areas, including drug development, diagnostics, patient care and monitoring.

In the pharmaceutical industry AI is also rapidly gaining a crucial role. The introduction of innovative medicines requires profound background knowledge and the latest means of communication. This drives us to intensively engage with the topic of medical education, which is becoming more and more demanding due to the dynamic knowledge landscape, among other things, accelerated even more by digitalization and AI.

Therefore, we argue for the incorporation of AI-based tools and methods in medical education, including personalized learning, diagnostic pathways, and data analysis, to prepare healthcare professionals for the evolving landscape of AI in medicine and support the fluency in dealing with AI by regular contact with various AI-based tools (Learning with AI).

Understanding AI's vast potential and its caveats as well as gaining a basic knowledge of how AI works should be an important part of medical education to ensure that physicians can effectively and responsibly leverage AI-based systems in their daily practice and in scientific communication (Learning about AI).

Keywords

Artificial intelligence medical education curriculum

Introduction

Medical education faces increasing challenges due to the rapidly expanding knowledge, accelerating development of novel diagnostics and treatments, and increasingly personalized approaches. These changes are all boosted by digitalization, and recently by the exponentially expanding applications of artificial intelligence (AI).

The term AI was coined by cognitive scientist John McCarthy in a funding application for the first workshop on automatic computers and neural networks.¹ held in 1956 at Dartmouth College, New Hampshire, USA. Half a century later, around 2011, everyday applications of AI appeared, heralded by the introduction of Apple Inc.'s virtual assistant Siri.² The advent of ChatGPT, a chatbot based on a large language model (LLM) developed by OpenAI Inc., further propelled AI into the forefront of public awareness, enabling widely available direct access to a generative AI-based system since late 2022.³

Given the rapidly spreading applications and exponentially growing capabilities of AI in a wide range of fields, its integration and implementation in clinical medicine, patient care and medical research is rapidly gaining momentum. Accordingly, academic medical education is increasingly adopting AI as an versatile technology and as an obligatory subject. In this paper we discuss the background, requirements and opportunities of AI in academic medical education.

Artificial intelligence in medicine

AI has already made inroads into medical practice with highly diverse applications, rapidly expanding capabilities and increasingly widespread use, eg in the form of machine learning algorithms performing image analysis, or LLMs analyzing massive datasets from medical records, clinical trials, or medical literature.^4,5

Some AI-based systems can compete with experts in certain domains and may outperform them in specifically defined tasks. Examples include the assessment of cardiac function,⁶ the detection of pathological lesions in cerebral magnetic resonance images,⁷ or the estimation of gestational age.⁸ Use cases of AI are rapidly expanding in patient care, underscoring the transformative potential of AI in diagnostics, surveillance and decision making.

Beyond patient care, the applications of AI extend to many sectors of the healthcare industry, including pharmaceutical companies, where AI has been highly useful for drug development in recent years, as demonstrated by the rapid introduction of protease inhibitors active against the SARS-CoV-2 virus facilitated by AI-assisted molecular modeling that enabled the virtual screening of millions of compounds^9,10 and guided the decision on which molecule to advance for further development.¹¹

The introduction of such groundbreaking medicines into clinical practice demands a wealth of background knowledge and the latest methods of medical communication. Consequently, medical functions in the pharmaceutical industry are deeply engaged in scientific education and communication. Based on our experiences with the game-changing potential of AI in drug development and the expected impact of AI in the healthcare sector, we perceive a strong need for physicians to actively embrace AI with its capabilities and applications – using AI-based systems and their ramifications in a differentiated and ethically responsible manner. Numerous authors therefore call for educational programs fostering knowledge and skills in AI as part of the core curriculum in medical school.^12-15

The aim of this article is to explore

which areas in medical education of physicians have a high potential to benefit from the integration of AI-based tools and solutions in knowledge acquisition, to create the best individual, practical and effective learning experiences, and to familiarize physicians with the practical application of AI (learning with AI),

which basic knowledge in the field of AI could be particularly relevant and beneficial for physicians to prepare them for the challenges of medical practice in the upcoming age of an increasingly AI-based environment (learning about AI).

Selected domains of AI and their potential use in medical practice

As no consensual classification of AI has been worked out until now, one possible approach is the classification based on competences of the respective AI system (Figure 1).^16,17 The following explanations provide some basic technical background for selected practically important concepts and domains of AI.

Figure 1.

The landscape of artificial intelligence. Although there is no consensual subclassification of the broad field of AI, one option is the definition of AI domains based on competencies of the respective AI systems (figure based on content from ref.¹⁶).

Expert systems

Rule-based expert systems encoding human knowledge were the first step towards genuine AI. However, they perform poorly in problem solving and knowledge acquisition.¹⁸ They remain in use eg in medical alert implementations, identification of drug-drug interactions, or reminder systems.¹⁹

Machine learning

Machine learning (ML) is an umbrella term that refers to a broad range of algorithms that perform intelligent predictions based on a given dataset.²⁰ Three approaches can be distinguished:²¹

In supervised learning, ML systems may acquire knowledge by learning from “experiences” that are compared with pre-known results, and stepwise adjustment of modeling parameters.²² With representative and sufficiently large datasets, the training converges to yield good performance on unseen samples as well. An algorithm type that can be trained using supervised ML is decision-tree based algorithms, eg used for prediction of circulatory failure in intensive care patients.²³

In unsupervised learning, the prospective output is unknown and the algorithm must discover correlations and patterns by itself, eg identifying non-obvious similarities between patients.²⁴ Unsupervised learning was used eg to identify clinical phenotypes of patients with sepsis.²⁵

In reinforcement learning, the algorithm is exposed to a simulated environment where rewards for successful training steps are obtained.²⁶ This approach is explored eg in real-time management of sepsis, to inform measures based on the current patient data and embedded clinician expertise.²⁷

Deep learning

Deep learning uses a machine learning technique that does not require extensive human feature engineering and works with multiple layers of processing units called neurons or perceptrons, combining to artificial neural networks (ANN), with many interconnected computing neurons organized into consecutive layers.^28,29 An ANN passes information through consecutive layers where data are processed layer-by-layer to generate the output.³⁰ Specific deep learning algorithms capture patterns observed in audio, text, or imaging data.^31,32

Deep learning significantly expanded the applications of AI in medicine, because unstructured data (eg patient histories or serial MRI images) can be used to predict relevant patient outcomes.³³ For example, AI-based evaluation of brain imaging data has been shown to be highly useful for quantitative spatial analyses of degenerative cerebral processes in multiple sclerosis.³⁴

The functionality of classic machine learning and deep learning is visualized in Figure 2 using the clinical example of a skin lesion.

Figure 2.

Functionality of deep learning algorithms compared to classic machine learning (figure based on content from refs.^30,80). Classical machine learning techniques involve feature engineering. Expert-directed classification is often required. In contrast, deep learning approaches work with artificial neural networks, usually designed to autonomously extract and classify features from the presented data.

Natural language processing

Large Language Model (LLMs) such as ChatGPT are using specific deep learning algorithms to perform a wide range of natural language processing tasks such as language understanding and text generation.³⁵ LLMs hold promise for numerous fields of medical applications, for example in history taking and documentation, supporting effective and adequate physician-patient communication^36,37 or facilitating diagnostic procedures, eg in psychiatry, by quantitatively detecting communicative indicators of psychoses or mood disorders.³⁸ LLMs also may enhance therapy by measuring predictors of remission in patients at high risk for psychosis.³⁹

There are promising chatbots with possible applications at the physician-patient interface. The conversational chatbot Pi, developed by Inflection AI Inc., is particularly capable to detect and process a broad range of personal characteristics derived from information about everyday life and could be envisaged for example as a companion for patients during hospital stays offering emotional, often downright “sensitive” support.⁴⁰ The LLM Med-PaLM was designed by Google to answer medical questions. Despite current limitations, the potential of utility in medicine is high due to the expected improvement of such LLMs in comprehension, knowledge recall and reasoning in the next years.⁴¹

The principle of deep-learning-based LLMs can also be adopted to generate functional artificial protein sequences, a task which is conceptually related to the creation of grammatically and semantically correct sentences in natural language. ProGen is an LLM trained on using a large, comprehensive protein sequence dataset to design artificial proteins across multiple families and functions.⁴²

In scientific exchange, LLMs serve to extract and summarize large volumes of scientific data almost in real time, whereas evaluating and placing any new information in the context of pre-existing knowledge remains the realm of humans trained in their field. Integrating the respective tasks of humans and LLMs with their synergies in sociotechnical systems creates human-AI “teams” with enhanced capabilities.⁴³

Table 1 lists a range of potential applications of AI in patient care.⁴⁴ These spotlights indicate the rapidly expanding potential of AI in medicine, vastly enhancing, and in some areas exceeding human capabilities.⁴⁵ An imminent rapid expansion in the applications of AI in medicine and human cooperation with AI-based systems must be considered as a given.^44,46 Therefore, the evolving spectrum of concepts, methods, uses and implications of AI based systems should be inherently and comprehensively embedded in medical education – to prepare ongoing physicians to the exposure and the use of such tools in their practice.

Table 1.

Potential applications of AI in the typical domains of patient care. Content modified from ref.⁴⁴

Domain of patient care	Examples
Prevention and prognosis	Embryo selection during in-vitro fertilization Heart rate measurement via smartwatch
Diagnostics	Diagnostic (decision-making) tools (e.g ADA Health)⁶² Evaluation of medical images
Treatment	Identification of adverse drug interactions Supportive treatment of mental disorders by chatbots or apps Precision surgical robotics
Patient monitoring and follow-up	Daily check-ups and documentation of therapeutic progress via app Improvement of patient security
Administration	Automated review of laboratory data and imaging results Reduction of effort of repetitive administrative tasks (eg extraction of patient data from electronic health records)

Looking at current AI learning offerings in medical education in Germany

According to a study published in 2021 by the learning platform KI-Campus in cooperation with Charité Berlin University Medicine, the majority (28 of 39) of medical schools in Germany offered mostly facultative AI-related courses or extracurricular activities, further AI courses were planned by more of the medical schools at this timepoint.¹²

The joint study of KI-Campus and Charité also found that in 2021 AI-related content and skills in Germany were included in the continuing medical education (CME) regulations largely as a component of the additional qualification “Medical Informatics”.¹² However, the range of CME courses offered for continuing education of physicians has increased significantly since.

These insights into the current AI learning offerings in medical education show that physicians in Germany currently largely still fend for themselves to acquire practice-related knowledge and skills in the field of AI. Furthermore, elective courses offered in medical school often allow access only for rather limited numbers of participants per term.

As a lighthouse project of curriculum-independent education on AI, the above-mentioned learning platform KI-Campus⁴⁷ is continually developed in the framework of a research project by a multifaceted consortium of academic and governmental institutions. It offers a broad spectrum of self-study educational formats that are certified, eg, by the State Medical Association of Baden-Württemberg.⁴⁷ Content that specifically addresses medical uses of AI is predominantly provided by Charité – Berlin University Medicine and covers the foundations of AI and its range of uses in medicine.

Dealing with AI-based tools and using them in medical practice will soon be essential in the medical profession, and this most certainly applies to all medical disciplines. Although a certain range of AI courses is already available for medical students and physicians, we plead for the broader incorporation of both AI-based tools and methods (Learning with AI) and of basic AI knowledge (Learning about AI) in medical education. This approach will support the fluency in dealing with AI by regular contact with various AI-based tools and to ensure that physicians can effectively and responsibly leverage AI-based systems in their daily practice and in scientific communication.

In the following section, we first take a deeper look at diverse benefits from integrating AI-based tools in medical education before we discuss AI content which is in our view valuable for physicians to deal with the opportunities and challenges of AI in medical practice and scientific communication.

Learning with AI: benefits of integrating AI-based tools and solutions into medical education

Given the rise of AI and its broad ramifications, the concept of “Learning with AI” should be increasingly actively implemented and promoted in medical education to (i) tailor learning content individually for effective learning experiences and to (ii) familiarize physicians with the various AI applications through everyday involvement, and prepare them for using AI in clinical practice. While doing so, adequate balance of AI-based and teacher-led learning should be maintained,⁴⁸ with an emphasis on intellectual independence and critical thinking.

The number of literature on the topic of AI in medical education has grown tremendously over the last years.⁴⁹ There are several reviews looking at different aspects of current use of AI in medical education. Machine Learning and Deep Learning were identified as most frequently used AI technologies, mostly applied in training labs and surgery domains.⁵⁰ Regarding the primary target groups for AI use, one review published in 2019 including 37 articles has found that the ability of AI to provide immediate feedback makes it to a valuable tool mainly for undergraduate students in medical learning support, whereas the use of AI in continuous medical education (CME) was considerably smaller.⁵¹ However, in another review published in 2023 more than half of the 42 included studies regarding the application of AI in medical education concerned specialized training.⁵² The most common areas of application of AI were found in the teaching process of theoretical and practical knowledge, with AI being used for example in anatomy education, as assistance to medical students in the recognition and diagnosis of medical images or to provide feedback and instructions to surgical interns.⁵²

Studies on effectiveness of AI in medical education are still scarce with usually a small number of participants.⁵² In controlled experiments, there has been a significant improvement in ophthalmology residents’ performance in recognizing pathological myopia when learned via an AI-based training system compared to the classic learning method through traditional lectures.⁵³ Comparing pre-learning and post-learning accuracy, also a significant difference was shown for medical students using an AI-based medical image learning system to highlight hip fracture on a pelvic film in comparison to a conventional learning group.⁵⁴ According to the literature the perception of usefulness and usability of different applications of AI in medical education is mostly positive, whereas user friendliness and flexibility are common critical points when dealing with AI-supported systems.^48,53

When looking at the use of AI in medical education, there are numerous positive effects and just as many challenges. The advantages include flexible and distance learning opportunities, a stress-free learning environment through AI-based clinical setting simulators and an individualized learning experience and guidance through feedback.⁵⁵ On the other hand, the lack of essential infrastructure and technological framework, scarce effectiveness data and ethical issues such as data security are only some of the challenges when aiming to integrate AI broadly in medical education.^55,56

However, considering the emerging number of publications on the topic AI in medical education, indicating an increased interest and increasingly frequent use in the last years, AI will most likely pervade nearly all aspects of medical learning, training, and practical teaching as well as testing knowledge and skills. Table 2 gives an overview over areas in medical education with a supposedly high level of potential benefit from AI-based tools, fleshed out with examples of applications of AI for learning purposes.

Table 2.

Potentially beneficial areas of use of AI in medical education

Areas of use	Potential benefits from using AI-based learning tools
Individual Learning
Personalized learning	AI-based learning platforms adapt to the individual knowledge status, learning pace and learning style. They can provide personalized study schedules, knowledge checks, and resources tailored to a learner's strengths and weaknesses. Ideally, AI adapts the learning pathways, content and presentation modalities to the needs and preferences of the individual, deliver a personalized learning experience and provide real-time feedback and recommendations with live tracking of the progress being made.⁶⁷
Assessment and feedback	AI can automate the grading of quizzes and exams, providing immediate feedback to the learner. It can also track their performance over time, identifying areas where additional study may be required.^67,68
Virtual anatomy and 3D visualization	Learning in anatomy is greatly enhanced by high-quality visualization of structures and functions. AI can assist advanced 3D visualizations of human anatomy, allowing students and physicians to explore and understand complex structures and systems more effectively. Virtual dissection tools may complement or even replace traditional cadaver labs, making learning of anatomy more accessible and less resource-intensive.^69,70
Serious games	AI-driven serious games, which use game design elements within non-game contexts, can be used for detecting diseases and for the evaluation of user performance. The analysis of the data can improve players’ knowledge, skills, and training progress.^71,72
Virtual assistants	AI chatbots and virtual assistants answer physicians’ questions, provide information on medical topics, and assist in the learning process, making educational resources available 24/7.⁷³
Physician-patient communication
Medical simulation	AI-based medical simulations may create realistic virtual patient scenarios, allowing physicians to learn and practice guideline-based diagnostic pathways and explore disease treatment options in a safe and controlled virtual environment. These simulations can mimic a wide range of medical cases and conditions and help students gain quasi-practical experience.^72,74
Medical communication	Chatbots using avatars may support the training of physician-patient communication and simulate case conferences.⁷⁵
Real-Time translation and communication	AI-powered language translation can help bridge communication gaps between physicians and patients who speak various foreign languages, enhancing the ability to take patient histories and provide adequate care.⁷⁶
Diagnostic pathways	AI systems can assist physicians in learning systematic approaches to clinical diagnosis based on patient data, the medical literature, and databases with historical cases.⁷⁷
Scientific communication and education
Data analysis and research	AI tools provide real-time summaries and analyzes of vast amounts of medical data, including patient records, research data, and clinical trials, to identify trends, correlations, and potential paradigm changes. Physicians can use AI to stay up to date with the latest research and advances.⁷⁸
Virtual conferences and collaborations	AI can facilitate virtual medical conferences, allowing physicians to attend lectures, workshops, congresses and facilitate networking with peers and professionals in the field from anywhere in the world.⁷⁹

Learning about AI: perspectives and requirements of AI-related medical education

Perception of AI among practicing physicians and medical students

Especially in the field of radiology, where AI is already used quite widely in medical practice, several studies have been carried out to analyze the perception of AI among medical students and physicians. In a survey of 263 medical students, many were aware of the potential applications and implications of AI in radiology or medicine in general and generally did not expect that the human physician will be replaced.⁵⁷ However another survey has found that students perceive that AI would reduce job prospects for some specialties such as diagnostic radiology, what would have a major impact on the decision to choose diagnostic radiology as a medical specialty.^58,59 In an online survey that included 3018 medical students, AI was predominantly perceived as an assistive technology that can support access of physicians to information and access of patients to healthcare.⁵⁶ In this study most of the students favored structured training on AI applications during medical education (93.8%) as they a perceived critical need of being educated in this field.⁵⁶

The high necessity of becoming educated in AI has been shown also in another study among 390 US medical students. A majority (355/388, 91.5%) of respondents agreed that training in artificial intelligence concepts during medical school would be useful for their future.⁶⁰ The effect of AI courses was demonstrated e.g. in a in post-graduate radiology institutional AI literacy course that successfully improved AI education of participants.⁶¹

AI basic knowledge framework in medical education

Based on our own experiences with the game-changing potential of AI in pharmaceutical industry, and taking into account the existing and imminent application areas of AI in the healthcare sector – we propose an AI basic knowledge framework for medical education, as visualized in Figure 3, inspired by the AI competency framework for medical school published by KI-Campus and Charité – Berlin University Medicine.¹²

Figure 3.

Framework of basic AI knowledge for medical education. Potential content is clustered in five areas, where the area of “general classification of AI in medical practice” – which covers the evaluation and classification of AI output – touches all the other teaching areas.

In this framework explained below in more detail we identified four educations pillars mathematical foundations, essentials of data science, principles of AI functionality and ethical and legal considerations under the roof of the general classification of AI in medical practice.

General classification of AI in medical practice

This area includes the overarching goal of the AI basic knowledge framework: physicians should be able to evaluate whether the output of the AI really it is suitable to relate new insights to previous knowledge.

Furthermore, medical education must convey that there is no competition between physician and AI, but that they can benefit from each other. For instance, the suggestion of the most likely three diagnoses through a diagnostic tool such as ADA Health, an app where patients can get symptom-based diagnoses and recommendations of action in case of illness,⁶² should not be dismissed as competing with the physicians’ diagnostic skills. Rather, an informed evaluation based on the suggestions of these tools may save valuable time – which is freed up to contemplate, discuss the treatment options and making the “right” decisions with the patient. Ideally, AI systems will serve (i) as competent supporters of physicians in diagnostics, treatment decisions and implementation of care, and (ii) as knowledgeable supportive and proactive companions for patients.

The following specific aspects should be covered as part of an introduction to the area of AI:

Definitions of intelligence and AI

Short history of AI

Landscape of AI (see Figure 1)

Strengths and weaknesses of artificial in contrast to human intelligence.

Mathematical foundation: statistics and stochastics

In our point of view, it is not necessary for physicians to understand the complex mathematics behind the algorithms, but a basic understanding of the mathematical principles and statistical background knowledge should be acquired as this is necessary for understanding the core functionality of algorithms. Medical students should develop an general idea of how AI systems generate output from the input data and gain insights into frequently used algorithms and concepts.

Essentials of data science

Data science knowledge to be acquired should include:

Importance of data in the medical context

Data requirements for optimal processing by AI algorithms (eg structured vs unstructured data, standardization)

Data integration as a core component of the broader data management process and interoperability

Introduction of relational databases (eg in the context of electronic medical records)

It should also be mentioned at this point that a physician does not need to become a data scientist to be able to use, evaluate and classify AI output in everyday clinical practice.

Therefore, in our opinion, the AI basic knowledge framework for physicians does not have to cover programming knowledge or database query languages such as SQL.

However, since learning the basics of programming such as using the language Python, enables a deeper understanding of AI and there are numerous application areas in diverse cross-sectional areas of medicine. Offering elective courses to acquire programming knowledge in medical education could be interesting for a subset of students and physicians.

Principles of AI functionality with focus on relevance in the healthcare sector

Future physicians should learn the functions, areas of application and limitations of medically relevant AI areas, as outlined in this article above. These include

Expert systems

Categories of learning: supervised, unsupervised, reinforcement learning

Principles of machine learning, deep learning and neural networks

Natural language processing: principles of function as well as dealing with and limitations of LLMs/chatbots for different fields of application

Basics of image processing

Application areas of robotics in medicine

Ethical and legal considerations

Other relevant content that should be addressed in a medical AI curriculum includes ethics and the medicolegal implications of AI in medicine.

The use of AI in medical education entails a plethora of ethical and legal questions and conundrums. It is essential to ensure that learning with and about AI in medical education consistently aligns with educational goals, maintains high quality standards of medical information and practice, and adheres to current guidelines and procedures consented by human medical experts. Furthermore, the aspects of data security and patient protection should be highlighted.

While current uses of AI in medicine focus on imaging, diagnostics and monitoring, future applications of AI may actively support treatment decisions and help to define and adjust pathways of individual patient care. Therefore, the field of “explainable AI”, which comprises in particular transparency and comprehensibility of the results received from AI and in decision-making systems,⁶³ becomes increasingly important and it should be given appropriate awareness in the knowledge framework.⁶⁴

It is well documented that biased training data sets may lead to biased algorithms that can disadvantage minorities.⁶⁵ It is very important that physicians are aware of the potentially ingrained biases of ML models and are able to judge whether the model training is performed in a way that minimizes the risk of bias.

Conclusions

Medical education faces increasing challenges with the rapid advancement of artificial intelligence that has a growing impact on medical practice. Aspiring physicians need to embrace AI and its expanding relevance for clinical practice. Broadly integrating AI into medical education and providing fundamental AI knowledge will prepare them to responsibly utilize AI in healthcare.

As the writer Franz Kafka observed, “paths are created by walking them.” The path of AI is currently expanding at a breathtaking pace. To avoid being left behind – and to become competent in this new field – aspiring physicians in the age of AI must actively prepare for this journey in alignment with the pledge that unites medical professionals worldwide – that the health and wellbeing of the patients is their first consideration.⁶⁶

Footnotes

Acknowledgment

The authors would like to thank Markus Fischer for the valuable support regarding linguistic adjustment and the submission process.

Author contributions

All authors cooperatively conceptualized the content. Dina Domrös-Zoungrana wrote the original manuscript. Emma Fröling and Christian Lenz contributed significantly to refining the main ideas of the paper. Neda Rajaeean and Sebastian Boie contributed specific information and certain sections on AI functionality. All authors reviewed, edited and finally consented the manuscript.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Disclaimer

The views expressed in this article are those of the authors and do not reflect the official policy of their employer.

Funding

Medical writing assistance provided by Markus Fischer (FischerBioMedical, Homburg/Saar, Germany) and publication fees were funded by Pfizer Pharma GbmH.

ORCID iD

Dina Domrös-Zoungrana

References

Moor

. The Dartmouth College artificial intelligence conference: the next fifty years. AI Mag. 2006;27(4):87-89.

Hoy

. Alexa, Siri, Cortana, and more: an introduction to voice assistants. Med Ref Serv Q. 2018;37(1):81-88.

Fütterer

Fischer

Alekseeva

, et al. ChatGPT in education: global reactions to AI innovations. Sci Rep. 2023;13:15310.

Rajkomar

Dean

Kohane

. Machine learning in medicine. N Engl J Med. 2019;380(14):1347-1358.

Iqbal

Cortés Jaimes

Makineni

, et al. Reimagining healthcare: unleashing the power of artificial intelligence in medicine. Cureus. 2023;15(9):e44658.

Kwan

Cho

, et al. Blinded, randomized trial of sonographer versus AI cardiac function assessment. Nature. 2023;616(7957):520-524.

Barnett

Wang

Beadnall

, et al. A real-world clinical validation for AI-based MRI monitoring in multiple sclerosis. NPJ Digit Med. 2023;6(1):196.

Lee

Bradburn

Craik

, et al. Machine learning for accurate estimation of fetal gestational age based on ultrasound images. NPJ Digit Med. 2023;6(1):36.

Vidhya

Sultana

. Naveen Kumar

MNKM

. Rangareddy

Artificial intelligence's impact on drug discovery and development from bench to bedside. Cureus. 2023;15(10):e47486.

10.

Saramago

Santana

Gomes

, et al. AI-Driven Discovery of SARS-CoV-2 main protease fragment-like inhibitors with antiviral activity in vitro. J Chem Inf Model. 2023;63(9):2866-2880.

11.

Pfizer Inc. Accessed April 30, 2024. www.pfizer.com/sites/default/files/investors/financial_reports/annual_reports/2022/story/data-and-ai-are-helping-to-get-medicines-to-patients-faster/.

12.

Mosch

Back

Balzer

, et al. Lernangebote zu Künstlicher Intelligenz in der Medizin. KI-Campus; 2021, ki-campus.org. Lernangebote_KI_in_der_Medizin_ Studie.pdf.

13.

Katznelson

Gerke

. The need for health AI ethics in medical school education. Adv Health Sci Educ Theory Pract. 2021;26(4):1447-1458.

14.

Spirnak

Antani

. The need for artificial intelligence curriculum in military medical education. Mil Med. 2023;189(5-6):954-958.

15.

Lee

Kim

Lee

, et al. Defining medical AI competencies for medical school graduates: outcomes of a delphi survey and medical student/educator questionnaire of South Korean medical schools. Acad Med. 2024;99(5):524-533.

16.

Humm

Buxmann

Schmidt

. Grundlagen und Anwendungen von KI. In: Künstliche Intelligenz in der Forschung. Ethics of science and technology assessment, Vol 48. Springer; 2022:17.

17.

Kaul

Enslin

Gross

. History of artificial intelligence in medicine. Gastrointest Endosc. 2020;92(4):807-812.

18.

Kendal

Creen

. An introduction to knowledge engineering. Springer; 2007. ISBN 978-1-84628-475-5.

19.

Zhou

Sordo

. Artificial intelligence in medicine. In: Xing

Giger

Min

, eds. Chapter 5. Expert systems in medicine. Academic Press; 2020.

20.

Nichols

Herbert Chan

Baker

MAB

. Machine learning: applications of artificial intelligence to imaging and diagnosis. Biophys Rev. 2019;11(1):111-118.

21.

Gupta

Srivastava

Sahu

Tiwari

Ambasta

Kumar

. Artificial intelligence to deep learning: machine intelligence approach for drug discovery. Mol Divers. 2021;25:1315-1360.

22.

Faouzi

Colliot

. Classic machine learning methods. In: Colliot

, ed. Chapter 2: Machine learning for brain disorders [Internet]. Humana; 2023 Jul 23.

23.

Hyland

Faltys

Hüser

, et al. Early prediction of circulatory failure in the intensive care unit using machine learning. Nat Med. 2020;26(3):364-373.

24.

Kim

Corpuz

Weng

Rohde

. Applying unsupervised machine learning approaches to characterize autologous breast reconstruction patient subgroups: an NSQIP analysis of 14,274 patients. J Plast Reconstr Aesthet Surg. 2024;88:330-339.

25.

Aldewereld

Zhang

Urbano

, et al. Identification of clinical phenotypes in septic patients presenting with hypotension or elevated lactate. Front Med. 2022;9:794423.

26.

Wise

Emery

Radulescu

. Naturalistic reinforcement learning. Trends Cogn Sci. 2023;28(2):144-158.

27.

Cheng

. A value-based deep reinforcement learning model with human expertise in optimal treatment of sepsis. NPJ Digit Med. 2023;6:15.

28.

Tran

Kondrashova

Bradley

Williams

Pearson

Waddell

. Deep learning in cancer diagnosis, prognosis and treatment selection. Genome Med. 2021;13(1):152.

29.

Wang

Casalino

Khullar

. Deep learning in medicine—promise, progress, and challenges. JAMA Intern Med. 2019;179(3):293-294.

30.

Dildar

Akram

Irfan

, et al. Mahnashi MH. Skin cancer detection: a review using deep learning techniques. Int J Environ Res Public Health. 2021;18(10):5479.

31.

Purwins

Virtanen

Schlueter

Chang

Sainath

. Deep learning for aurdo signal processing. IEEE J Sel Top Signal Process. 2019;13(2):206-219.

32.

Shetty

Sissiqa

. Deep learning algorothms ns applications in computer vision. Int J Computer Scirnces Engineer. 2019;7(7):195-201.

33.

Tayefi

Ngo

Chomutare

Dalianis

Salvi

Godtliebsen

. Challenges and opportunities beyond structured data in analysis of electronic health records. WIRES Comp Stat. 2021;13(1):1-19.

34.

Bonacchi

Filippi

Rocca

. Role of artificial intelligence in MS clinical practice. Neuroimage Clin. 2022;35:103065.

35.

Ray

. ChatGPT: a comprehensive review on background, applications, key challenges, bias, ethics, limitations and future scope. Internet of Things and Cyberphysical Systems. 2023;3:121-154.

36.

Mesko

. The ChatGPT (generative artificial intelligence) revolution has made artificial intelligence approachable for medical professionals. J Med Internet Res. 2023;25:e48392.

37.

Sonntagbauer

Haar

Kluge

. Artificial intelligence: how will ChatGPT and other AI applications change our everyday medical practice? Med Klinik Intensivmedizin Notfallmedizin. 2023;118(5):366-371.

38.

Rezaii

Walker

Wolff

A machine learning approach to predicting psychosis using semantic density and latent content analysis. npj Schizophr. 2019;5:9.

39.

Choe

Choi

, et al. Beyond verbal fluency in the verbal fluency task: semantic clustering as a predictor of remission in individuals at clinical high risk for psychosis. J Psychiatry Neurosci. 2023;48(6):E414-E420.

40.

https://pi.ai/talk; last accessed November 22, 2023.

41.

Singhal

Azizi

, et al. Large language models encode clinical knowledge. Nature. 2023;620(7972):172-180.

42.

Madani

Krause

Greene

, et al. Large language models generate functional protein sequences across diverse families. Nat Biotechnol. 2023;41(8):1099-1106.

43.

Berretta

Tausch

Ontrup

Gilles

Peifer

Kluge

. Defining human-AI teaming the human-centered way: a scoping review and network analysis. Front Artif Intell. 2023;6:1250725.

44.

Künstliche Intelligenz im ärztlichen Alltag - Einsatzgebiete in der Medizin: Nutzen, Herausforderungen und Forderungen der FMH [Artificial intelligence in everyday medical practice - areas of application in medicine: benefits, challenges and demands of the FMH]; www.fmh.ch. September 2022.

45.

Khera

Butte

Berkwits

, et al. AI In medicine—JAMA’s focus on clinical outcomes, patient-centered care, quality, and equity. JAMA. 2023;330(9):818-820.

46.

Haug

Drazen

. Artificial intelligence and machine learning in clinical medicine, 2023. N Engl J Med. 2023;388(13):1201-1208.

47.

KI-Campus. First certified online courses on AI in medicine. https://ki-campus.org/press/zertifizierte-medizin-kurse?language_content_entity=de.

48.

Zhang

Mingxuan

Lee

, et al. AI in medical education: global situation, effects and challenges. Educ Inf Technol. 2023;29(4):1-23.

49.

Sun

Yin

Zhao

. Artificial intelligence for healthcare and medical education: a systematic review. Am J Transl Res. 2023;15(7):4820-4828.

50.

Nagi

Salih

Alzubaidi

, et al. Applications of artificial intelligence (AI) in medical education: a scoping review. Stud Health Technol Inform. 2023;305:648-651.

51.

Chan

Zary

. Applications and challenges of implementing artificial intelligence in medical education: integrative review. JMIR Medical Education. 2019;5(1):e13930.

52.

Zhang

Cai

Lee

, et al. AI In medical education: global situation, effects and challenges. Educ Inf Technol. 2024;29(4):4611-4633.

53.

Fang

Han

. Artificial intelligence-based pathologic myopia identification system in the ophthalmology residency training program. Front. Cell Dev. Biol. 2022;10:1053079.

54.

Cheng

Chen

, et al. Artificial intelligence-based education assists medical students’ interpretation of hip fracture. Insights Imaging. 2020;11(1):119.

55.

Zarei

Ghasemi

. The application of artificial intelligence and drug repositioning for the identification of fibroblast growth factor receptor inhibitors: a review. Adv Biomed Res. 2024 Jan 30;13:9. PMID: 38525398; PMCID: PMC10958741.

56.

Civaner

Uncu

Bulut

, et al. Artificial intelligence in medical education: a cross-sectional needs assessment. BMC Med Educ. 2022;22(1):772.

57.

Pinto Dos Santos

Giese

Brodehl

, et al. Medical students’ attitude towards artificial intelligence: a multicentre survey. Eur Radiol. 2019;29(4):1640-1646.

58.

Atalay

Baird

Stib

George

Oueidat

Cronan

. The impact of emerging technologies on residency selection by medical students in 2017 and 2021, with a focus on diagnostic radiology. Acad Radiol. 2023;30(6):1181-1188.

59.

Hathaway

Hogg

Lakhani

. Need for medical student education in emerging technologies and artificial intelligence: fostering enthusiasm, rather than flight, from specialties most affected by emerging technologies. Acad Radiol. 2023;30(8):1770-1771.

60.

Liu

Sawyer

Luna

, et al. Perceptions of US medical students on artificial intelligence in medicine: mixed methods survey study. JMIR Med Educ. 2022;8(4):e38325.

61.

Perchik

Smith

Elkassem

, et al. Artificial intelligence literacy: developing a multi-institutional infrastructure for AI education. Acad Radiol. 2023;30(7):1472-1480.

62.

Ada Health GmbH. Accessed January 17, 2024. https://ada.com/de.

63.

Bouchefry

de Souza

. Learning in big data: knowledge discovery in big data from astronomy and earth observation. In: P. Škoda P, Adam F, eds. Knowledge discovery in big data from astronomy and earth observation. Elsevier;2020:225-249.

64.

Band

Atefeh Yarahmadi

Hsu

C-C

, et al. Application of explainable artificial intelligence in medical health: a systematic review of interpretability methods. Informatics in Medicine Unlocked. 2023;40:101286.

65.

Tejani

Rayan

. Understanding and mitigating bias in imaging artificial intelligence. Radiographics. 2024;44(5):e230067.

66.

Rheinsberg

Parsa-Parsi

Kloiber

Wiesing

. Medical oath: use and relevance of the declaration of Geneva. A survey of member organizations of the world medical association (WMA). Med Health Care Philos. 2018;21(2):189-196.

67.

Varma

Fernando

Ting

Aamir

Sivaprakasam

. The global use of artificial intelligence in the undergraduate medical curriculum: a systematic review. Cureus. 2023;15(5):e39701.

68.

Turner

Hashimoto

Vasisht

Schaye

. Demystifying AI: Current state and future role in medical education assessment. Acad Med. 2023;99(4S Suppl 1):42-47.

69.

Parikh

. Virtual dissection and learning anatomy: a medical student's perspective. Med Sci Educ. 2023;33(6):1427-1428.

70.

Yang

. Technology-Enhanced preclinical medical education (anatomy, histology and occasionally, biochemistry): a practical guide. Adv Exp Med Biol. 2023;1431:65-93.

71.

Abd-Alrazaq

Abuelezz

Hassan

, et al. Artificial intelligence-driven serious games in health care: scoping review. JMIR Serious Games. 2022;10(4):e39840.

72.

Narayanan

Ramakrishnan

Durairaj

Das

. Artificial intelligence revolutionizing the field of medical education. Cureus. 2023;15(11):e49604.

73.

Gutiérrez-Cirlos

Carrillo-Pérez

Bermúdez-González

Hidrogo-Montemayor

Carrillo-Esper

Sánchez-Mendiola

. ChatGPT: opportunities and risks in the fields of medical care, teaching, and research. Gac Med Mex. 2023;159(5):372-379.

74.

Scherr

Halaseh

Spina

Andalib

Rivera

. ChatGPT interactive medical simulations for early clinical education: case study. JMIR Med Educ. 2023;9:e49877.

75.

Liaw

Tan

Bin Rusli

, et al. Artificial intelligence versus human-controlled doctor in virtual reality simulation for sepsis team training: randomized controlled study. J Med Internet Res. 2023;25:e47748.

76.

Herrmann-Werner

Loda

Zipfel

Holderried

Erschens

. Evaluation of a language translation app in an undergraduate medical communication course: proof-of-concept and usability study. JMIR Mhealth Uhealth. 2021;9(12):e31559.

77.

Borgbjerg

Thompson

Salte

Frøkjær

. Towards AI-augmented radiology education: a web-based application for perception training in chest x-ray nodule detection. Br J Radiol. 2023;96(1152):20230299.

78.

Siegel

Rossi

Lubowitz

. Artificial intelligence and machine learning may resolve health care information overload. Arthroscopy. 2024;40(6):1721-1723.

79.

Sandrone

. Medical education in the metaverse. Nat Med. 2022;28(12):2456-2457.

80.

Choi

Coyner

Kalpathy-Cramer

Chiang

Campbell

. Introduction to machine learning, neural networks, and deep learning. Transl Vis Sci Technol. 2020;9(2):14.

Medical Education: Considerations for a Successful Integration of Learning with and Learning about AI

Abstract

Keywords

Introduction

Artificial intelligence in medicine

Selected domains of AI and their potential use in medical practice

Expert systems

Machine learning

Deep learning

Natural language processing

Looking at current AI learning offerings in medical education in Germany

Learning with AI: benefits of integrating AI-based tools and solutions into medical education

Learning about AI: perspectives and requirements of AI-related medical education

Perception of AI among practicing physicians and medical students

AI basic knowledge framework in medical education

General classification of AI in medical practice

Mathematical foundation: statistics and stochastics

Essentials of data science

Principles of AI functionality with focus on relevance in the healthcare sector

Ethical and legal considerations

Conclusions

Footnotes

Acknowledgment

Author contributions

Declaration of conflicting interests

Disclaimer

Funding

ORCID iD

References