Reviewing the current state of virtual reality integration in medical education - a scoping review

Background In medical education, new technologies like Virtual Reality (VR) are increasingly integrated to enhance digital learning. Originally used to train surgical procedures, now use cases also cover emergency scenarios and non-technical skills like clinical decision-making. This scoping review aims to provide an overview of VR in medical education, including requirements, advantages, disadvantages, as well as evaluation methods and respective study results to establish a foundation for future VR integration into medical curricula. Methods This review follows the updated JBI methodology for scoping reviews and adheres to the respective PRISMA extension. We included reviews in English or German language from 2012 to March 2022 that examine the use of VR in education for medical and nursing students, registered nurses, and qualified physicians. Data extraction focused on medical specialties, subjects, curricula, technical/didactic requirements, evaluation methods and study outcomes as well as advantages and disadvantages of VR. Results A total of 763 records were identified. After eligibility assessment, 69 studies were included. Nearly half of them were published between 2021 and 2022, predominantly from high-income countries. Most reviews focused on surgical training in laparoscopic and minimally invasive procedures (43.5%) and included studies with qualified physicians as participants (43.5%). Technical, didactic and organisational requirements were highlighted and evaluations covering performance time and quality, skills acquisition and validity, often showed positive outcomes. Accessibility, repeatability, cost-effectiveness, and improved skill development were reported as advantages, while financial challenges, technical limitations, lack of scientific evidence, and potential user discomfort were cited as disadvantages. Discussion Despite a high potential of VR in medical education, there are mandatory requirements for its integration into medical curricula addressing challenges related to finances, technical limitations, and didactic aspects. The reported lack of standardised and validated guidelines for evaluating VR training must be overcome to enable high-quality evidence for VR usage in medical education. Interdisciplinary teams of software developers, AI experts, designers, medical didactics experts and end users are required to design useful VR courses. Technical issues and compromised realism can be mitigated by further technological advancements.


Introduction
The focus of medical education is gradually shifting towards digital learning methods, with an increasing emphasis on incorporating new technologies facilitated by the rapid progress in computer science, particularly in artificial intelligence (AI).This has led to the adoption and integration of Virtual Reality (VR) as an educational tool for prospective and registered health care professionals, covering an ever-expanding range of use-cases.

Developments of VR technology
Based on the literature, the first ten years of the 21st century can be referred to as the "VR winter" [1], characterised by a lack of public interest in this emerging technology.However, there was continuous but limited research taking place in corporate, academic, and military research facilities worldwide.The widespread adoption of VR was hindered by the high costs involved, with hardware expenses exceeding $35,000 for a so-called head-mounted display (HMD) and over $30,000 for tracking equipment.Moreover, the fragile infrastructure posed additional challenges for its widespread usage.Around 2011, there was a renewed interest in consumer-grade VR, primarily driven by its potential for entertainment.Prominent companies like Valve, NVIDIA, and the start-up Oculus played a significant role in advancing HMD-based VR technology, transitioning it from being exclusive to technical elites in specialised labs to becoming a mainstream medium for content consumption available to the public [1].

Immersive versus screen-based VR
By now, VR can broadly be divided into screen-based applications (as widely used in surgical simulators) and immersive solutions (wearing HMDs).While literature comparing effectiveness between these two approaches is sparse, according to Gutiérrez et al. [2], study participants benefited significantly more from immersive than screen-based training in terms of knowledge gain, which is recommended to be explored in more detail [3].

Development of VR use cases in medical education
Having experienced its primary use case in the surgical field, VR simulations have been shown to be beneficial for robotic surgery training [4].In addition, according to Izard et al. [5], VR tools have proven effective in providing in-depth knowledge of surgical interventions.This technology allows users to repeat all procedural steps as often as needed to accommodate individual learning progress, an approach which would be impractical in realworld settings [5].
As learning can be particularly challenging in critical situations where repeated training is not feasible due to the risk to patients' lives, VR training deserves special attention for acquiring routine in emergency scenarios, such as cardiopulmonary resuscitation [6].
Furthermore, effective and enforced communication, as one example of non-technical skills (NTS), is gaining increasing awareness for healthcare providers.The impact of NTS in medical education and respective learning goals are growing likewise to the students' demands to practise these skills in VR [7].While training applications to develop NTS in VR are still underrepresented, vast developments in AI encourage growing numbers of studies with respective immersive simulations [8].Additionally, evidence for VR simulations being beneficial for communication skills, emotional management, critical thinking, and clinical decision-making is expanding [9][10][11].
Between 2020-2022, the SARS-CoV-2 pandemic fuelled integration of digital solutions into medical curricula [12] which at the same time increased use cases for VR technology as demonstrated by Birrenbach et al. [13]: e.g., to train tasks such as hand disinfection, nasopharyngeal swab-taking as well as the proper wearing and removal of personal protective equipment.

Students' perception of VR
Overall, students' perception of integrating VR into medical education is very positive as shown by De Ponti et al. [14] as well as a survey conducted at our medical faculty in 2022 [7].To further enhance acceptance of VR, its deep integration into medical curricula is required [15].Simultaneously, cost-effectiveness can be realised in the long run compared to traditional training methods that rely on cadavers, manikins or actors [16,17].
This scoping review has been conducted in parallel to the project "medical tr.AI.ning" [18], which aims to develop an AI-based immersive VR learning platform that enables medical students to practise clinical decision-making with interactive virtual patients in realistic environments.

Currently available reviews and gaps
During the initial search on MEDLINE (PubMed), we identified one scoping review which addresses VR in medical education as well but with a limited study cohort of undergraduate or pre-registration medical students [19].Jiang et al. [19] focused on the technology deployment of VR tools, their features and respective study characteristics.Our scoping review goes beyond this publication by exploring additional aspects of integrating VR in medical education, not only regarding undergraduate/pre-registered medical and nursing students but also physicians and registered nurses.As an extension to Jiang et al., we also include reported evaluation methods and results as well as advantages and disadvantages of applying VR in medical education.
Another scoping review covering a similar topic was recently published by Lie et al. [20].Nevertheless, they only included 7 papers that solely focused on immersive VR and searched other databases, such as Academic Search Elite, Education Source, or Google Scholar.Still, they presented valuable recommendations regarding the integration of VR into health professions education while referring to Carl May's general theory of implementation [21].Therefore, they identified 7 categories which must be considered: collaboration, availability, expenses, guidelines, technology/usability, careful design & evaluation and training [20].
Furthermore, we identified a scoping review dealing with the effects of VR in medical education published in February 2023 [22], wherein 11 out of 28 included studies focused on education.However, this study did not include reviews and registered nurses as study cohorts.Apart from reporting clinical outcomes, the main goal of their scoping review was the assessment of specific endpoints in medical training, such as knowledge, skills or confidence.While providing important insights into the positive outcomes of VR use in medical education, our study covers additional aspects, such as requirements, advantages or disadvantages of VR used for educational purposes in the medical field.

Aim of this review
This scoping review aims to provide an overview of the current use and status of VR in medical education and to excerpt the requirements, advantages, and disadvantages associated with integrating this technology for training health care professionals.Furthermore, we assessed whether and how the use of VR technology has been evaluated and which results were derived to establish a foundation on which further VR projects can be developed, evaluated, and integrated into medical curricula.

Review questions
This review aims to address the following questions:

Methods
The presented scoping review was conducted in accordance with the updated JBI methodology for scoping reviews [23] as well as in line with the PRISMA extension for scoping reviews (PRISMA-ScR) [24].The corresponding protocol was developed in alignment with PRISMA-P [25] and has been made publicly available [26].

Inclusion criteria Participants
This scoping review did not include patient recruitment or public involvement itself, but we examined reviews that included studies on the use of VR in the education of medical and nursing students, registered nurses and qualified physicians.

Concept
In alignment with the defined research questions, reviews which cover the application of VR as a tool for health care professional education were considered for this work.As foundation, only reviews being published from 2012 until March 2022 were considered as eligible publications since they most comprehensively cover knowledge gain and developments in this research area.
Open access publications and publications accessible via our institution in English as global scientific language or German as native language of the authors of this review were taken into account.Reviews without focus on medical education, e.g., the application of VR in patient treatment, as well as publications which centred Augmented / Extended / or Mixed Reality were excluded.Although this review also analyses requirements, evaluation concepts and results, advantages and disadvantages of using VR in medical education, the overarching concept is to examine how and in which medical curricula VR has been already used to date.

Context
Findings of this review can contribute as a foundation to development directions for VR applications, their integration into medical education and respective study designs.Considering the still high initial acquisition costs of VR hardware, differences between high-and low-income countries are likely to reveal.Regarding the joint German project "medical tr.AI.ning" [18], it is of particular interest to explore how and to what extent VR has been used in medical education yet in Germany, Europe and in comparison to the rest of the world as well as to identify respective gaps.

Type of sources
Resulting from a preliminary search in MEDLINE (Pub-Med), the literature search was narrowed down to focus only on reviews that topic corresponding studies on VR since they most appropriately summarise knowledge acquisition and developmental directions within this research field while providing scientific evidence.

Search strategy
A three-step search strategy was established to target eligible publications according to the applied methodology [23].After a preliminary search of MEDLINE (PubMed) which was independently conducted by two authors, search terms were compared and a consensus for search terminology and strategy was established for MEDLINE (PubMed), ScienceDirect (Elsevier), Web of Science Core Collection (Clarivate), Cochrane Library (Wiley), and JBI Evidence Synthesis (Wolter Kluwer) (see Table 1 ) and reviewed by the third author.

Study/Source of evidence selection
Following the database search, all identified records were uploaded into Mendeley Desktop (version 1.19.8, 2020) and duplicates were removed.After a pilot test, titles and abstracts were screened by two independent reviewers following the established inclusion criteria.Full texts of potentially eligible studies were retrieved, imported to the JBI System for the Unified Management, Assessment and Review of Information (JBI SUMARI) [27] and independently screened by the two reviewers following the inclusion criteria while documenting reasons for exclusion.Any arising disagreements during the whole screening process were resolved through discussion and consensus or consulting the third reviewer.Results of the evidence selection are presented as PRISMA2020-flow diagram (see Fig. 1) [28].

Data extraction
After full text screening and a data extraction pilot test, information relevant for the research questions were independently retrieved from included reviews by two reviewers using a spreadsheet with a focus on details about the medical specialty, subject and curriculum in which VR was used, details on technical/didactic requirements, evaluation methodology and outcomes, as well as advantages and disadvantages of VR in medical education.The revised draft extraction form is provided in Table 2. Any disagreements arising between the reviewers were resolved through discussions and consensus or consulting the third reviewer.

Data analysis and presentation
Findings are presented following the PRISMA-ScR checklist [24].Evidence is primarily presented in tabular form.A narrative summary accompanies the tabulated and/or charted results and outlines how the results are linked to the research questions.

Included reviews
As an outcome of the applied search strategy from March following criteria: no focus on VR-technology (n = 28), literature research of the respective review conducted before 2012 (n = 5), review published before 2012 (n = 3), medical education not covered (n = 6), or not being a review (n = 4).

Characteristics of included reviews (RQ1)
Table 3 contains an overview of the characteristics of the included reviews regarding country of origin according to the first author as well as language, year of publication, type of review, number of included studies and which types of VR were investigated:

Country of origin
Out of the 69 included reviews, 24 (34.8%) were from Europe, 21 (30.4%) from North America, 14 (20.3%) from Asia, 5 (7.2%) each from South America and Australia, respectively.No reviews originated from Africa.Among European countries, the United Kingdom was most prevalent (n = 12) followed by France (n = 3).Germany was represented with 2 out of 69 included publications.Among North America, the USA was predominant (n = 15), followed by Canada (n = 6).5 publications from Brazil were the only representatives of South America.Among Asia, 4 reviews were included from China as well as 3 from Singapore.

Language
Considering the language of included publications, English was almost exclusively predominant with 67 reviews (97.1%).Only 2 (2.9%) German-language publications were included in this review.

Year of publication
Regarding the number of included reviews and their publication date, there is a clear increasing trend over time: while only 5 out of 69 (7.2%) included articles were published between 2012 and 2014, 12 reviews (17.4%) were published between 2015 and 2017, 24 reviews (34.8%) between 2018 and 2020 and nearly half, i.e., 28 out of 69 included articles (40.6%), published between 2021 and the time point of our literature research in March 2022 (Fig. 2).

Type of review
With 30 out of 69 included articles, nearly half (43.5%) were associated by their authors with a specific type of review or review methodology.27 reviews (39.1%) were classified as systematic review, followed by 4 meta-analysis papers (5.8%) and 3 narrative reviews (4.3%).

Number of included studies
Similar to the non-specification of the review type, 22

Study participants (RQ2)
We found different study subject populations among the studies reported by the reviews (see Table 4): 30 out of 69 (43.5%) reviews focused on studies with qualified physicians only, while 9 reviews (13.0%) addressed medical students only.Exclusively registered nurses were part of studies of 4 reviews (5.8%), while 5 (7.2%) did not explicitly specify which type of participants were included in the considered studies.The remaining reviews focused on studies addressing combinations of participants, such as medical students and qualified physicians (n = 8, 11.6%), medical and nursing students (n = 2, 2.9%) or all previously mentioned groups of participants (medical students, nursing students, qualified physicians and registered nurses) in 11 out of 69 (15.9%) reviews.

Subjects/Curricula (RQ3)
The included reviews covered a broad range of subjects, specialties and skills of medical curricula regarding VR applications (see Fig. 3; Table 5).46 out of 69 (66.7%) reviews addressed more than one subject or specialty while 21 reviews (30.4%) exclusively focused on one of them.In 2 reviews (2.9%), the outlined characteristics of the included studies did not provide clear information on the considered medical subjects.

Surgical specialties
Predominantly, studies within the scope of VR for training in surgery and surgical specialties were discussed by the included reviews with minimally invasive types of surgery (e.g., laparoscopic, endoscopic) being the subject covered by most publications (n = 30, 43.5%).This type of surgery is often associated with general surgery and internal medicine interventions (n = 20, 29.0%), orthopaedics (n = 16, 23.2%) and neurosurgery (n = 11, 15.9%).Additionally, the last group includes VR applications for preoperative planning as well.Overall, 14 specialties related to surgery were represented by the included reviews.Depending on the specialty and procedure, applications are not covering the whole spectrum of possibilities, e.g., simulators for hip surgery or replacement were reported to lag behind other orthopaedic procedures such as knee and shoulder arthroscopy [29,30].

Non-surgical specialties
Regarding non-surgical clinical specialties, emergency medicine as well as associated scenarios were most prevalent (n = 9, 13.0%) followed by paediatrics and urology, each represented by 5 reviews (7.2%).In total, 15 clinical specialties with no focus on surgery were covered by the included reviews besides 3 preclinical subjects, i.e., anatomy (n = 10, 14.5%), physiology and psychology (each n = 1, 1.4%).Furthermore, interdisciplinary and overarching topics, such as basic clinical or NTS (cognitive, interprofessional, social, communicative) were addressed in 14 reviews (20.3%).

Requirements (RQ4)
Based on our findings considering specifications of requirements mentioned in the included reviews, they were classified according to technical, didactical and organisational requirements for the integration of VR in medical education and ordered by frequency of mention (see Fig. 4; Table 6 ).Overall, 42 out of 69 reviews (39.1%) provided information on requirements.Technical requirements consider the acquisition of respective hardware to be able to use VR applications, which concerns all 69 included reviews, followed by sensory/haptic feedback (n = 5, 7.2%) and accurate interaction regarding tissues (n = 3, 4.3%).
Among the didactical requirements, a strong need for well-structured curricula with sound learning theory framework and clearly defined learning objectives was  highlighted most by reviews (n = 15, 21.7%).The concepts of mastery learning, and deliberate practice were particularly mentioned in this context (n = 4, 5.8% and n = 3, 4.3%).

Synopsis of study evaluations (RQ5) Summary of study methodologies
Overall, 65 (94.2%)reviews mentioned evaluation aspects at least for some of their included studies (see summary in Fig. 5).
In line with the above illustrated focus on surgical procedures, evaluation results reported in most reviews (n = 40, 57.9%) addressed performance time and quality (e.g., completion time, procedure or examinations scores, complication rates, error rates during procedures).Results on skills and knowledge acquisition were described in 27 reviews (39.1%).
Evaluations on different levels of validity [31] were presented by 25 out of 69 reviews (36.2%).Construct validity, which questions the ability to discriminate between novice and experts, was mentioned in 14 out of 69 studies (20.3%), followed by transfer validity (n = 13, 18.8%), which describes the correlation between simulation-and actual real performance.Face validity, defined as the appearance of a test being appropriate, is covered in 3 reviews (4.3%).Concurrent validity, which evaluates the relationship to outcomes of another instrument purporting to measure the same construct, and content validity, covering the appropriateness of the contents of test items, are mentioned in one review each (1.4%).
To evaluate the use of VR in medical education, mainly comparisons to traditional methods, such as mannequins, cadaver, desktop learning, written or video instructions, or self-study were reported in 11 reviews (15.9%).Moreover, knowledge assessment (n = 11, 15.9%)  and quasi-experimental studies (pretest vs. posttest; n = 7, 10.1%) were considered.For the assessment of subjective outcomes, surveys or questionnaires were used (n = 6, 8.7%) besides analysing automatic quantitative metrics of simulators (n = 5, 7.2%).

Summary of study results
Among the included reviews, most prevalent were reviews which included at least some information derived from the included studies' evaluations (n = 65, 94.2%).Only 4 reviews (5.8%) did not present explicit details of study results.In the following, the term "mixed results" is defined by the fact that at least one of the included studies in a review did not argue in favour of VR based on comparisons regarding the analysed aspect (see summary in Fig. 6; Table 7 ).
Considering performance time and quality, reduced procedure or performance time after VR simulation was outlined by 20 out of 40 reviews (50.0%), 6 reviews (15.0%) described ambiguous findings considering this aspect.10 reviews (25.0%) detailed increased procedure or examination scores, whereas 5 (12.5%) reported mixed results.For example, Guedes et al. [32] reported that increased examination scores were observed considering basic minimally invasive surgery tasks comparing VR and box trainer groups while no difference could be detected considering other outcomes such as task completion time.The same pattern of unambiguous versus mixed results emerges regarding increased performance (12 vs. 3 reviews) and reduced complication rates (8 reviews vs. 1 review).Increased performance also implied team performance in the case of 3 reviews.Reduced error rates during procedures were described without exceptions in 11 reviews (27.5%) and a significant correlation between performance in surgical VR simulation and mean OSACSS (= Objective Structured Assessment of Cataract Surgical Skill) scores [33] was reported by 2 reviews (5.0%).
While skill acquisition after VR simulation was reported by 11 out of 27 reviews (40.7%) which contained results on skills and knowledge acquisition, 6 (22.2%) presented ambiguous results.The same pattern emerged for knowledge acquisition (10 vs. 5 reviews).Improved spatial understanding and visualisation was reported by 6 out of 27 reviews (22.2%) and one review (3.7%) described increased empathy.Additionally, less guidance by instructors was required to successfully complete a task after simulation (n = 2, 7.4%).
Of the 25 reviews presenting results on different levels of validity [31], 6 (24.0%) reported on more than one level of validity.Beginning with the lowest level and ordered by increasing level of validity, face validity was reported to be demonstrated in 3 reviews (12.0%), content validity in 1 (4.0%), construct validity in 14 (56.0%) and transfer validity in 11 reviews (44.0%).Considering construct validity, 2 reviews pointed out that the coarse distinction between groups of novices and experts regarding performance works well, whereas finer distinction still needs improvement.
Contradictory results considering transfer validity were reported in 2 out of the 25 reviews (8.0%), while 1 review (4.0%) reported in general poor evidence of complete validation.Except for those 3 reviews, none of the  25 reviews reported that the tested validity could not be confirmed by the included studies.
Out of the 18 publications covering results on subjective aspects, i.e., study participants' level of confidence, anxiety, satisfaction, attitude and/or self-efficacy, 13 (72.2%)delineated that satisfaction as well as positive motivation and attitude towards VR applications were demonstrated.In contrast, 4 out of 18 (22.2%)reported mixed results.Regarding the participants' level of confidence and anxiety, 2 reviews (11.1%) concluded that confidence improved, and anxiety or stress reduced, whereas 4 publications (22.2%) reported mixed results.Increased self-efficacy was reported in studies covered by 2 reviews (11.1%).
Considering results on learning curves, a higher improvement could be observed among novices in comparison to experts according to 4 out of 9 reviews (44.4%).Longer learning curves were reported in comparison to other already established learning methods (n = 2, 22.2%).Long-term retention of knowledge was reported with positive and mixed results with equal frequency (each n = 2, 22.2%).
Improved patient outcome and care quality after training with VR applications were reported by all 6 reviews that included results on this aspect.
Among the 3 publications covering cost-effectiveness of VR applications, 2 reviews (66.7%) reported positive results while 1 review (33.3%) claimed a general lack of evidence regarding this topic.

Advantages of VR (RQ6)
VR has become an increasingly popular tool in medical education.The advantages of VR in medical education that we found described in the included reviews can be organised into seven categories: (1) practical aspects, (2) content, (3) skill development, (4) clinical transfer, (5) user experience (6) assessment and feedback, and (7) didactic aspects (Fig. 7; Table 8).mistakes without time constraints (n = 17, 24.6%).The variability of VR scenarios provides multiple use cases consisting of single basic tasks or complex procedures (n = 13, 18.8.%).This also facilitates to cover unexpected or low frequency scenarios as well as scenarios with high performance pressure (n = 7, 10.1%) in realistic immersive environments, leading to more authentic training (n = 7, 10.1%).Additionally, VR can help learners to develop their spatial understanding and visualisation (n = 4, 5.8%) and most recent applications also provide versatile multi-user scenarios that require actual teamwork in VR (n = 1, 1.4%).

Disadvantages of VR (RQ7)
While VR has the potential to revolutionise medical education, it comes with several drawbacks that need to be considered.They can be categorised into (1) financial aspects, (2) technical limitations, (3) didactic and scientific evidence and (4) user experience (Fig. 8; Table 9).
1) One of the biggest challenges relates to financial aspects, the most frequently cited in our literature research (n = 30, 43.5%).The development and design of VR applications for medical education require a multidisciplinary team, which is time-intensive and costly.Additionally, appropriate hardware, personnel and technical support for development and maintenance need to be acquired, users and teachers need to be trained and all these aspects are accompanied by an administrative outlay.2) Despite high pace improvements, technical limitations are still an obstacle.Representing real-life scenarios accurately in VR can be difficult due to various factors, such as using controllers instead of actual instruments, a lack of realistic feedback (e.g., bleedings), missing variations in anatomy, tissue replication, and deformation, or depth perception in case of 2D simulations.Besides that, it is challenging to display authentic complexity of human behaviour, reac-tions and traits across diverse populations (n = 24, 34.7%).Particularly missing or unreliable haptic feedback (e.g., lack of force input, ergonomic limitations) was an often-mentioned disadvantage of VR training (n = 18, 26.1%).3) When it comes to didactics and scientific evidence, inconsistent or missing evidence was reported for the broad range of validity dimensions of acquired knowledge and skills, cost-effectiveness, or cost-benefit-ratio due to the heterogeneity of objective study outcome measures, VR hardware and a missing theoretical learning framework as backbone of studies (n = 24, 34.7%).Furthermore, VR scenarios are often simplified, repetitive, and highly specific, with isolated tasks and not complemented with additional learning material, and accompanied by limited variations (n = 9, 13.0%).Currently, there is still a lack of evidence for long-term retention of acquired knowledge and skills.Apart from surgical simulators, most VR simulations focus on teaching cognitive skills, with few procedural or affective applications.They lack decision-making scenarios, consent processes, effective communication, as well as training of e.g., leadership, which especially features NTS (n = 4, 5.8%).4) Further challenges are related to the user experience.

REPORTED ADVANTAGES Number of Reviews
Review References

Mitigation strategies
Considering the mentioned disadvantages of VR, some reviews reported possible means of mitigation: costs of hard-and software tend to decrease over time [34,35], and cost-effectiveness could be achieved in the long-term [16,36,37].Technological advancements could make up for missing haptic feedback [19,38,39], e.g., by using devices such as gloves [19].Future research of interdisciplinary teams can help to provide standardised systematic guidelines [40] for comparisons at different levels of validity and structured curriculum implementation [41] enables comparisons at larger scale for a higher level of evidence [3,42], circumventing heterogeneity of studies and their outcomes [43].Further investigations on long-term retention of gained knowledge and skills in VR training scenarios are recommended since the evidence base is still lacking [39,41,44,45].This also includes exploring the necessity of refreshing training sessions [46,47] and to critically rethink the use of multiplechoice questionnaires or pure test scores for the assessment of knowledge gain and depth of understanding [38,44].Advancements in Natural-Language-Processing and AI are necessary [48] to create more interactive and realistic training scenarios.Cybersickness can be tackled by reduced latency and increased frames per second [34,49], as well as the use of fixed backgrounds [50] or movement sensors [49].

Discussion
This scoping review summarises findings of a reproducible systematic literature research in March 2022 with a total of 69 included reviews covering a broad range of VR application fields in medical education.
Research questions focused on technical and didactical requirements for the implementation of VR courses, evaluation methods and results, as well as reported advantages and disadvantages of using VR as a medical education tool.To ensure highest reliability with a broad and as complete, current, and precise as possible synopsis, we solely focused on reviews including studies between 2012 and 2022 and added current publications in the discussion.

Characteristics of included reviews
In the last decade, an enhancing relevance of VR in medical education is shown by an increasing number of reviews and studies since 2012.Hotspots are still highincome countries, while low-income countries are rather underrepresented.For example, the fact that we found no review originating from Africa indicates that VR technology by now has not yet become routine, particularly not in low-income countries, which is in agreement with previous findings [3,19,51].Concerning specialties that use VR for educational purposes, we can approve findings of other reviews which identified surgical use cases to be predominant Fig. 8 Most mentioned disadvantages of VR in medical education as reported in the included reviews [19,52,53] -especially minimally invasive types (laparoscopic, endoscopic) -with a clear focus on procedural skills, followed by emergency medicine scenarios.The background reason for implementing VR training in these subjects is most likely linked to one of the biggest advantages reported for this technology: the repeatability of procedures without causing any harm or danger to patients and/or practitioners.By acquiring routine in a safe environment with realistic conditions, using VR courses enhances preparation for real life situations where clinical management is based on algorithms and requires fast decision-making.Even home-based training Table 9 Disadvantages of VR simulations in medical education as reported by included reviews.Linked to RQ7.

REPORTED DISADVANTAGES Number of Reviews
Review References

DIDACTIC ASPECTS AND SCIENTIFIC EVIDENCE
Inconsistent or missing evidence for the broad range of validity dimensions of acquired knowledge and skills, cost-effectiveness or cost-benefit-ratio due to the heterogeneity of objective study outcome measures and VR hardware.Missing theoretical learning framework as backbone of studies 24 [3,11,32,36,43,44,47,60,61,63,70,75,77,84,101,103,104,107,111,113,117] Scenarios are often simplified, repetitive, highly specific with isolated tasks and not complemented with additional learning material, variation is limited Lacking evidence for long-term retention of acquired knowledge and skills 5 [42,[44][45][46]55] Lack of multiplayer-scenarios including face-to-face communications and interconnection with team 5 [19,34,39,47,105] Except for surgical simulators, most VR simulations focus on teaching cognitive skills, rarely on procedural or affective applications, lack of decision-making scenarios, consent processes, effective communication, leadership, i.e., non-technical skills 4 [39,44,48,84] Bias introduction in case of unfamiliarity with immersive VR and potential technical errors 4 [44,64,73,75] Unequal representation of medical specialties in VR applications 2 [3,113] Individual needs of lecturers not necessarily integrated in application design 1 [44]

USER EXPERIENCE
Cybersickness in case of immersive VR using HMDs (e.g., motion sickness, nausea/vomiting, dizziness, cold sweats, asthenopia, fatigue, headache, neck discomfort, blurred vision, post-VR changes in static balance) 10 [34,49,50,61,66,73,85,92,105,107] VR simulations may induce overconfidence 1 [109] Serious content of scenarios can cause stressful experiences, anxieties about performance or interpersonal dynamics 1 [61] Potential of misuse, excessive use or game addiction 1 [61] sessions might be feasible [19,54].Furthermore, the predominance of both surgical use cases and physicians as study participants reported by the included reviews indicates that VR has mainly been investigated and used regarding advanced medical and clinical training instead of basic education of undergraduate medical [51] or nursing students [55].

Technical requirements
According to the high rate of surgical VR applications, realistic haptic feedback and accurate interaction regarding tissues were mentioned as technical requirements.Naturally, manual skills required for surgical disciplines are easier to be embedded in simulations compared to other competencies such as NTS which demand multiplayer modes or highly intelligent virtual agents.
Although the technical advancements already enable extending the range of training scenarios accordingly [19], they are still underrepresented as reflected in our review.Technical issues and barriers must be resolved before the VR application is regularly used to guarantee that they are not negatively affecting the learning process and user experience [20].

Type of VR
Only 11 out of 69 included studies focused on immersive forms of VR, i.e., for which e.g., HMDs are required in contrast to non or less immersive forms such as screenbased applications.This broad conceptual perception and definition of VR is nowadays increasingly shifting towards immersive VR with HMDs [56] due to the technological advancements: increasingly realistic VR training scenarios with higher feeling of situational presence.However -since widespread usage of the technology is still limited -available publications have addressed this form of VR only rarely.The required degree of realism which is serving the learning goals deserves further investigation as well [57].

Didactical requirements
Remarkably, a majority of reviews indicate the need for integration of profound learning theory concepts into respective applications [44] taking into account different levels of learners' experience [46].The development of standardised guidelines for evaluation and integration into medical curricula has been largely lacking up to now [15,51,54,56,57] and might be impeded by deviating degrees of pretraining familiarity of participants with VR.Emblematic of this, the included reviews hardly contained any information on these specific requirements and almost 40% did not cover requirements at all.Concurrently, innovations in medical curricula implementing VR must address how to avoid that users receive incorrect automatically generated evaluation feedback through the application and how to identify responsibilities in case of incorrect feedback [58].Furthermore, it must be discussed how to prevent VR from being maliciously used or causing any decrease of valuable and necessary personal interactions between teachers and learners, also regarding evaluation feedback within or outside of VR applications [58].For example, lower performance of study participants after the simulation was observed when no external feedback from experts was provided in addition to the automatically generated feedback through the application [59].Thus, a hybrid form of evaluation feedback within and outside of the application might be preferable [48] and should be subject of further research [57].
VR can only be useful and effective if it is used in a reasonable way supported by a profound theoretical backbone to prevent from negative effects on the established curricula [44,60].Thus, VR scenarios in the form of serious games can take advantage of intrinsic and extrinsic motivation by gamification [61,62].
Therefore, it is strongly recommended to not consider VR as a completely independent stand-alone solution replacing already established learning formats but to consider it as a potential additionally supporting tool to existing learning programmes [17,35,51,60,[63][64][65][66][67] closing gaps of traditional methods.

Study methodologies and outcomes
The considered reviews clearly indicated the need to check all levels of validity [68] for VR applications, especially transfer validity, and to demonstrate their efficacy which has not been investigated to a sufficient level to date.Further research and controlled trials are required to validate tools considering these aspects.For example, complication rates should comprise postoperative complications besides intraoperative complications [69].
The mixed results reported in some studies can be linked to the already reported missing guidelines for both effective study design, i.e. randomised-controlled trials with large sample sizes [49,55,57,67], and lacking standardised study evaluation measures, e.g. by a combination of subjective (e.g. by experts) and objective, discriminant and reproducible validation assessments [36,63,[70][71][72], quantitative and qualitative data [46] and unified definition of effectiveness criteria [73] for large scale comparisons [56].This includes aspects such as user experience, immersion, cybersickness, cognitive load and ascending levels of validity and is linked to several sets of variables: individual learner, learning environment, context, technology, and pedagogy [73,74].Though, considering comparisons, participant blindness can be hardly achieved due to the nature of VR applications [75].
In addition, long-term retention of acquired knowledge and skills have not been covered sufficiently in studies to date [44,46].Patient benefit and quality of care are strongly recommended as ultimate objectives of testing procedures, but often remain untested [56,76,77].Furthermore, cost-effectiveness should be kept in mind throughout the whole developmental process and finally needs evaluation [3,46,47,70,73].It is recommended to draft respective guidelines based upon and/or adapted from existing or new frameworks, e.g., considering VR treatments in patient care [78], digital health interventions [79], e-learning [74] and shared experiences [80].Other sources could be Kirkpatrick's evaluation framework [52] or Blooms' taxonomy of educational objectives comprising cognitive, procedural and affective learning domains [81] amongst others [82].

Advantages of VR
Our findings indicate that VR technology is increasingly accessible and convenient, even for lower-income countries using low-cost models [83].It offers advantages such as flexibility in terms of time and space, repeatability of training, and cost-effectiveness, which is also linked to the choice of high-fidelity or low-fidelity simulations [30].
"Transfer Effectiveness Ratio" is the only validated measure of cost-effectiveness and should be used in further research on VR used for training.This ratio quantifies the difference between VR and real life in terms of the time (and cost) required to achieve fully competent performance [30,36].
While providing a safe and controlled environment for learning, VR simulations allow users to make mistakes and training without time constraints [84].The variability of VR scenarios enables to practise both skills related to basic tasks and complex procedures, theoretical knowledge and NTS [73], including unexpected or low frequency scenarios as well as high-pressure situations.This makes it interesting for both low-experienced users and high-experienced users aiming to improve complex skills and situations [85,86].Zhang et al. [84] reported that VR simulator training has been shown to be comparable to clinical training with similar outcomes.VR technology could enable patient-specific simulations [41], or enhance pre-operative planning [29,87,88], be adapted to the individual learning preferences and levels of knowledge of the user [42], or even suggest next training steps based upon advanced algorithms [38].
VR enhances training authenticity, helps develop spatial understanding and visualisation skills [43,44], and can offer multi-user scenarios that promote soft skills such as teamwork [11,89].This goes along with the increasing focus on NTS in medical curricula.Advancing AI technology expands the use cases of VR as a psychomotor training tool.Complex scenarios and real time speech recognition foster decision-making, situational awareness, and communication skills.Flin et al. [90] define NTS as cognitive, social, and personal resource skills complementing technical skills and contributing to safe and efficient task performance.According to them, NTS include individual cognitive skills (e.g., situation awareness, decision-making, coping with stress, and management of fatigue) and interprofessional social skills (e.g., cooperation and teamwork, conflict resolution, leadership, empathy) [90].The importance of these competencies is recognised by currently revised learning goals, e.g., in German medical schools [91] and should be examined in more detail as part of future research and development [30].
According to Tang et al. [92], using immersive technology for medical education not only facilitates illustrating complex ideas while removing geographical constraints, but also eases data collection for later performance assessments and evaluation feedback.

Disadvantages of VR
Despite its potential to eminently shape future medical education, VR still comes with several considerable drawbacks.Although prices for hardware are decreasing, financial aspects remain a significant challenge, as to date VR development requires a multidisciplinary team, highperformance hardware, and concrete course-integrations demand technical support, along with training for users and teachers.The multidisciplinary team and the inclusion of end users, such as students and lecturers in the development process can help to successfully implement VR in medical education and contribute to the motivation for higher use and acceptance of VR solutions [20], also by considering the individual needs of lecturers and targeted learning outcomes [64,93].
Furthermore, technical limitations, such as difficulty in accurately representing real-life scenarios, lack of realistic feedback, and challenges in displaying authentic human behaviour, pose obstacles to VR training.Nevertheless, Trehan et al. [65] stated, that "simulation is not meant to eliminate the need for genuine patient interaction and real operating room (OR) experience, but to serve as an important adjunct for safer transition to independent patient care and continued practice".
Didactically, inconsistent or missing evidence considering long-term retention of knowledge and skills, as well as a lack of theoretical frameworks hamper the assessment of competences acquired through VR [19].A good example of severe necessity for VR training concepts is shown by the fact that since 2018 access to the operating room (OR) has been permitted without prior simulation training in some instances [42].Madan et al. [37] suggest combining the strengths of different simulation methods to compensate for their individual weaknesses while providing added value for skill acquisition.At the same time, they emphasise that simulations can only cover a portion of the whole learning while the actual patient must remain as "the final teacher".
In addition, user experience aspects require to be approached, e.g., including cybersickness [85] and technical problems which could strongly affect the user's satisfaction [64,75,94].

Mitigation strategies
The included reviews have already presented some potential solutions for existing drawbacks.However, we can think of further means for mitigation: some of them addressable by technological advancements and AI, e.g., higher degree of realism and representation of real-life variations, more multiplayer-scenarios and intelligent virtual agents to sufficiently train NTS.Variability and completeness of complex training scenarios, including accompanying learning material and curriculum integration could be achieved by future research with interdisciplinary development teams stemming from a university context and thus having direct influence on medical curricula.Careful planning of tutorial sessions or free navigation periods in VR [46] could make up for any novice-bias caused by less familiarity with VR, especially in comparative studies with traditional learning methods [44] and might shorten learning curves in VR simulations.Tutorial sessions could also compensate for divergent perceptions of VR-naive and VR-experienced users prior to training scenarios [70] and deserves future investigation [94].At the same time, it was observed that learners with less experience could be even more easily motivated by fictional scenarios [53,85].
The still unequal representation of medical specialties and corresponding training scenarios might be addressed by prospective research projects expanding the fields of possible applications.Misleading overconfidence after VR simulations [53] could be prevented by the development of more objective measures and respective evaluation feedback.Excessive use, misuse or game addiction could be counteracted by careful application design and consideration of addiction prevention strategies.One review pointed out that the serious content of VR training scenarios could cause stressful experiences and anxieties [61].However, we argue that the same risks can apply for other forms of training scenarios and are even a desired effect since they reflect context and emotions of real-world settings.
It could be beneficial to further explore gender-specific phenomena, e.g., female participants were reported to be more likely to be affected by cybersickness [73] while male showed a tendency for better performance and higher interest regarding laparoscopic simulations [37].
The increasing amount of VR content might offer the opportunity to develop overarching, universal platforms for sharing and making solutions globally available for health care education [19,51,72].This could include shared case libraries [48,56] as well as providing applications in multiple languages [56] and may reduce barriers such as acceptance and assist with the implementation in health care education [72].Though, such a unification and open-source policy is obviously accompanied by divergent financial interests and lack of compatibility between systems.

Strengths and limitations of this review
We have considered in this review all kinds of health care professionals' education and training without restriction to medical students only.A comprehensive literature search strategy was applied in five scientific literature databases.Screening and data extraction were independently conducted by two researchers in parallel and discussed in case of conflicting decisions to increase soundness while decreasing any subjective bias.Following the established methodology, this scoping review does not include an assessment of quality of evidence.Our search was limited to reviews published in English and German.The description of evaluation study designs in the included reviews is provided in an exemplary way in this review but not to full extent, since we focused on the evaluation target measures and results rather than their methods.Additionally, some studies might have been included by more than one of the inspected reviews.Definitions and perceptions of VR have changed over time [95], making comparisons of results partially difficult if the definition has not been clearly stated in the publication.All considered publications were peerreviewed publications of the article type "review".Thus, some other relevant publications, conference papers, books, etc. might have been overlooked.To ensure actuality, we have conducted additional literature research before the submission of this publication and referenced new appropriate findings.

Conclusions
VR in medical education offers a multitude of benefits that span various categories.The technology's practicality, user experience, and skill development potential make it an effective tool in medical education.Furthermore, the versatility of VR in terms of content, assessment, and evaluation feedback, as well as unique didactic advantages make it a valuable asset in medical education today that ultimately seems to improve patient safety and outcome.Currently observed disadvantages concerning financial aspects, technical limitations, didactic aspects and lacking scientific evidence, as well as user experience issues can be addressed by multidisciplinary collaborations in research and development for which this review can provide crucial indications and recommendations.
This review outlined the evolution and the current state of where and how VR technology is rooted in medical education and pointed out the strong need for standardised and validated guidelines for research, evaluation, didactic frameworks, and development projects to guarantee high quality of evidence.Alongside, multidisciplinary collaborations between physicians, medical didactics experts, 3D-designers, VR software developers as well as AI-experts and the end users are needed to create scientifically sound backbones for the development and integration of useful VR applications in medical education.

Fig. 1
Fig. 1 PRISMA2020-flow diagram of search and study selection process.VR: virtual reality

Fig. 2
Fig. 2 Histogram of the year of publication regarding the 69 included reviews

Fig. 3
Fig. 3 Most prevalent subjects and specialties covered by the 69 included reviews.Percentages do not add up to 100 due to overlaps by reviews covering more than one subject or specialty

Fig. 4
Fig.4 Most mentioned requirements for the development of VR applications for medical education as reported by included reviews."Respective hardware" is not mentioned in this figure since it is a basic requirement that was addressed in every included review.

4 )
Regarding clinical transfer, 16 reviews (23.2%) emphasised patient safety when training is supported by VR simulation.5 reviews (7.2%) referred to improved clinical outcomes, one crucial primary concern when assessing effectiveness of medical education methods.Moreover, VR can also include representation of patient-specific information, allowing for more personalised and patient-centred training and preparation (n = 4, 5.8%).5) Focusing on user experience, VR training can improve satisfaction, self-confidence and the educational experience for medical students (n = 6, 8.7%).The inclusion of real-time haptic feedback and multimodal sensorial stimuli improves tactile realism (n = 6, 8.7%).Additionally, users report high levels of satisfaction with VR (n = 5, 7.2%), which apparently can have an emotional impact according to 2 reviews (2.9%).6) When it comes to assessment and feedback, automatic vision-and sensor-based performance measurement allow for bias-free evaluations (n = 11, 15.9%).Instant real-time feedback (n = 7, 10.1%) as well as the possibility of recording training data (n = 5, 7.2%) and support efficient training in a standardised and reproducible manner (n = 6, 8.7%).7) In terms of didactic aspects, VR training addresses self-directed, self-paced, and student-centred learning and deliberate practice, which is seen to be more effective in acquiring certain skills (n = 8, 11.6%).A special advantage was mentioned considering experiential learning that can be achieved through VR, such as dementia training, which is otherwise difficult to simulate (n = 1, 1.4%).

Fig. 5 Fig. 6
Fig. 5 Summary of evaluation aspects reported by included reviews

Fig. 7
Fig. 7 Most mentioned advantages of VR in medical education as reported in the included reviews

Table 1
Full search strategy in PubMed, conducted in March 2022

Table 2
[26]l data extraction form.*=fieldswhich were added or modified in comparison to the originally published protocol[26] * VR, while 18 (26.1%)provided no specifications on the VR modality examined in the included studies.

Table 3
Geographic and basic characteristics of the included reviews.Linked to RQ1.

Table 4
Summary of groups of participants in the studies included in the reviews."All groups" refers to medical and nursing students as well as physicians and nurses.Linked to RQ2.

Table 5
Distribution of subjects and specialties among included reviews.Percentages do not add up to 100 due to overlaps by reviews covering more than one subject or specialty.Linked to RQ3.

Table 6
Distribution of technical, didactical and organisational requirements reported by reviews.Linked to RQ4.

Table 7
Summary of study results as reported by included reviews.Linked to RQ5.

Table 8
Advantages of VR simulations in medical education as reported by included reviews.Linked to RQ6.