Virtual reality training for gastrointestinal endoscopy: A systematic review of efficacy and outcomes

Abstract

BACKGROUND

Traditional gastrointestinal endoscopy (GIE) apprenticeship training raises concerns regarding patient safety, prompting growing adoption of simulation-based training approaches. Virtual reality (VR) simulators provide a risk-free, learner-centered environment in which trainees can develop and refine endoscopic skills prior to clinical practice. Although VR-based training has been increasingly implemented worldwide, its efficacy relative to conventional training methods has yet to be conclusively established.

AIM

To evaluate the effectiveness and outcomes of VR-based training in GIE through a systematic review of interventional studies.

METHODS

This systematic review followed PRISMA 2020 guidelines. PubMed, ScienceDirect, and Google Scholar were searched through June 2025 using Boolean combinations of terms related to VR, simulation, and GIE. Eligible studies were interventional trials evaluating VR-based training that reported changes in endoscopist competency. Procedures included esophagogastroduodenoscopy, colonoscopy, endoscopic retrograde cholangiopancreatography, and endoscopic ultrasound.

RESULTS

VR-based training consistently improved trainee competency, procedural independence, and patient safety compared with baseline performance, with outcomes comparable to conventional training. In esophagogastroduodenoscopy, VR simulation significantly increased independent procedure completion rates (odds ratio: 65.7; 95% confidence interval: 20.1-214.4) and achieved similar procedure times to traditional training (10.5 minutes vs 12.4 minutes). In endoscopic retrograde cholangiopancreatography, VR effectively differentiated skill levels, with experts outperforming novices in key metrics such as papilla visualization and cannulation time (P < 0.05). Endoscopic ultrasound competency improved when VR was combined with clinical training (64.53 ± 4.91 vs 60.09 ± 5.49; P = 0.028). Colonoscopy studies reported higher pass rates and improved objective skill scores in VR-trained groups (100% vs 88.9% pass rate in proficiency-based VR training).

CONCLUSION

These findings support integrating VR simulation as an adjunct to conventional endoscopy training and underscore the need for standardized training frameworks and further research to optimize implementation and assess long-term clinical impact.

Key Words: Virtual reality training; Gastrointestinal endoscopy; Esophagogastroduodenoscopy; Esophagoduodenoscopy; Colonoscopy; Endoscopic ultrasonography

Core Tip: This article synthesizes evidence from 30 interventional studies involving 901 clinicians to evaluate the efficacy of virtual reality (VR) training in gastrointestinal endoscopy. VR simulation improves trainee competency, procedural independence, and patient safety across key procedures, including esophagogastroduodenoscopy, colonoscopy, endoscopic retrograde cholangiopancreatography, and endoscopic ultrasound. Performance gains with VR were comparable to those achieved with conventional training, supporting its role as an adjunct in endoscopic education. Heterogeneity in study design limited meta-analysis, underscoring the need for standardized VR training protocols and further research to assess long-term clinical outcomes.

INTRODUCTION

Gastrointestinal endoscopy (GIE) is widely used for the diagnosis and management of gastrointestinal (GI) disorders[1]. Given its central role in clinical practice, there is an increasing demand for highly skilled endoscopists. Traditionally, endoscopy training has relied on an apprenticeship model involving supervised procedures performed on patients. However, heightened awareness of patient safety has driven greater adoption of simulation-based training. Simulation allows novice endoscopists to practice and refine technical skills in a learner-centered, risk-free environment[1,2]. This approach reduces the risk of patient harm and enhances safety by ensuring that trainees achieve proficiency before clinical practice[2]. Accordingly, effective implementation of simulation training and successful transfer of acquired skills to real-world clinical settings are critical objectives of modern endoscopic education[2,3].

The first mechanical simulators for endoscopy training were introduced in the 20^th century. Over time, these tools have evolved into sophisticated computer-based systems capable of providing real-time feedback to trainees[4]. These advanced platforms generate a virtual reality (VR) environment in which endoscopists can visualize procedural anatomy and receive immediate feedback on key technical parameters, such as applied force and loop formation during endoscopic procedure[4].

VR has been used in endoscopy training for more than a decade, with its adoption expanding worldwide. However, early evidence regarding its effectiveness has been mixed[2]. While available data support VR as a supplement to conventional training, there is no consensus on its ability to replace traditional apprenticeship-based education, a conclusion echoed by prior systematic reviews and meta-analyses[5]. Accordingly, we conducted an updated systematic review to synthesize emerging evidence on the role of VR in training GI endoscopists.

MATERIALS AND METHODS

Protocol and registration

This article was conducted in accordance with the PRISMA 2020 guidelines[6]. The review protocol was registered in the International Prospective Register of Systematic Reviews under registration number CRD420251071274.

Literature search

A comprehensive literature search was conducted independently by two authors to identify relevant studies published up to June 2025. PubMed, Web of Science, and Google Scholar were searched using database-specific strategies. Boolean operators were used to combine keywords as follows: (“simulation” OR “simulator” OR “simulation based”) AND (“virtual reality” OR “augmented Reality” OR “VR”) AND (“gastrointestinal” OR “GI”) AND (“endoscopy” OR “upper endoscopy” OR “colonoscopy” OR “sigmoidoscopy” OR “endoscopic retrograde pancreatography” OR “endoscopic ultrasound”). Reference lists of included studies were manually screened to identify additional eligible articles.

Eligibility criteria

Retrieved studies were assessed against predefined eligibility criteria prior to inclusion. Studies were included if they: (1) Evaluated VR-based training in GI endoscopic procedures; (2) Were interventional in design; (3) Reported changes in endoscopist competency following VR training; and (4) Were primary studies, including randomized clinical trials and nonrandomized interventional studies.

Studies were excluded if they: (1) Were published in languages other than English; (2) Did not incorporate VR as an endoscopy training intervention; (3) Did not involve GIE as the primary procedure; or (4) Failed to report relevant outcome measures.

Data extraction

Data extraction was independently performed by two reviewers. Any discrepancies were resolved through discussion until consensus was reached. In accordance with PRISMA guidelines, records were screened in multiple stages, beginning with title and abstract review to exclude irrelevant studies, followed by full-text assessment for eligibility. All studies meeting the inclusion criteria were retained for data extraction. Extracted data included author information, study setting, sample size, VR technology used, type of GIE, trainee experience level, intervention characteristics, and reported outcomes.

Quality assessment

Risk of bias in the included trials was assessed using the Risk of Bias 2 tool. This tool evaluates bias across five domains: The randomization process, deviations from intended interventions, blinding, outcome measurement, and selective outcome reporting. An overall risk-of-bias judgment was derived from domain-level assessments. Studies were classified as high risk of bias if one or more domains were rated as high risk, as having some concerns if one or more domains raised concerns, and as low risk of bias only if all domains were rated as low risk.

RESULTS

Literature search outcomes

A total of 2131 articles were identified across the databases. After removal of 1581 duplicates, 550 articles were screened for relevance, of which 475 were excluded. Seventy-five full-text articles were assessed for eligibility, and 30 studies met the inclusion criteria and were included in the review. Excluded studies comprised 12 secondary studies, 7 non-English publications, 9 studies that did not report relevant outcomes, 11 that did not include simulation-based training, and 6 that used mechanical simulators. The study selection process is summarized in the PRISMA flow diagram (Figure 1).

Figure 1 A PRISMA flow diagram summarizing the search strategy.

Characteristics of the included studies

This article included 30 trials conducted across diverse geographic regions, including North America, Europe, South America, and Asia. All studies utilized VR-based simulators, such as GI mentor I and II, ENDO VR, AccuTouch colonoscopy simulator, Kyoto Kagaku colonoscopy simulator, M40 Colonoscope training simulator, Endo TS-1 Olympus, Plastic Phantom, and the compact Erlangen Active Simulator for Interventional Endoscopy. Simulated procedures included colonoscopy, sigmoidoscopy, endoscopic retrograde cholangiopancreatography (ERCP), upper GIE, esophagogastroduodenoscopy (EGD), and endoscopic ultrasound (EUS). Across all studies, 901 participants were enrolled, encompassing medical students, interns, residents, fellows, and attending physicians. Detailed study characteristics are summarized in Tables 1 and 2[4-33].

Table 1 Characteristics of the included studies.

Ref.	Setting	Study design	Sample size	Level of training
Piskorz et al[4], 2020	Argentina	Pre- and post-test trial	126	PGY1 gastroenterology residents
Van Sickle et al[5], 2011	United States	Pre- and post-test trial	41	PGY1-4 general surgery residents
Telem et al[6], 2014	United States	Pre- and post-test trial	9	Novice surgical interns
Shirai et al[7], 2008	Japan	RCT	20	PGY1 and PGY2 residents in the gastroenterology and hepatology department
Gerson and Van Dam[8], 2003	United States	RCT	16	Internal medicine residents
Koch et al[9], 2015	Netherlands	RCT	18	Residents starting their gastroenterology training
Ferlitsch et al[10], 2002	Austria	RCT	24	Medical residents and experienced endoscopists
Silva Mendes et al[11], 2022	Portugal	Pre- and post-test trial	23	First-year gastroenterology residents
Buzink et al[12], 2007	Netherlands	Pre- and post-test trial	30	Medical interns or residents in training
McIntosh et al[13], 2014	United Kingdom	RCT	18	Residents PGY2 to PGY4
Park et al[14], 2007	Canada	RCT	24	Residents in internal medicine and gastroenterology in PGY1 to PGY3
Gomez et al[15], 2015	United States	RCT	27	Surgical PGY1 residents
Cohen et al[16], 2006	United States	RCT	45	Gastroenterology fellows
Grover et al[17], 2015	United States	RCT	34	Residents in internal medicine, gastroenterology, and general surgery
Vajpeyi et al[18], 2025	Canada	Pre-test post-test trial	33	Eight expert endoscopists and 25 novice endoscopists
Gao et al[19], 2019	China	RCT	27	Clinical doctors
Arnold et al[20], 2015	Denmark	Pre-test post-test trial	37	Senior doctors, medical doctors, and medical students
Bittner et al[21], 2010	United States	Pre-test post-test trial	12	Six were novice endoscopists, and 6 were expert endoscopists
Sahakian et al[22], 2016	United States	NRCT	10	Four gastroenterology faculty and six fellows
Ferlitsch et al[23], 2010	Austria	RCT	28	28 residents
Georgiou et al[24], 2023	Bulgaria	NRCT	37	7 expert endoscopists and 31 novices
Sedlack[25], 2007	United States	RCT	29	Gastroenterology fellows with different levels of experience
Sedlack et al[26], 2004	United States	RCT	38	PGY2 internal medicine residents
Sedlack and Kolars[27], 2004	United States	RCT	8	First-year GI fellows
Haycock et al[28], 2010	Netherlands, Italy, and the United Kingdom	RCT	36	Novice colonoscopists
Di Giulio et al[29], 2004	Italy	RCT	22	Gastroenterology fellows
Ahlberg et al[30], 2005	Sweden	RCT	22	12 residents and 10 surgeons
Grover et al[31], 2017	Canada	RCT	37	Novice endoscopists
Ende et al[32], 2012	Germany	RCT	28	Medical and surgical residents

PGY: Postgraduate year; RCT: Randomized controlled trial; NRCT: Non-randomized controlled trial; GI: Gastrointestinal.

Table 2 Procedure characteristics from the selected studies meeting the inclusion criteria.

Ref.	Type of virtual training	Endoscopic procedure	Description of procedure	Number of sessions	Outcome assessment
Van Sickle et al[5], 2011	GI mentor II	Colonoscopy	Basic colonoscopy with no anatomical or pathological variation	The trainees needed to achieve proficiency in two tasks before proceeding to another level of training in a maximum of 10 trials	Performance data, training data, and subjective self-evaluation data
Telem et al[6], 2014	GI mentor	Colonoscopy	12-hour GI mentor independent training	2-hour sessions each week for 6 weeks	Performance based on the GAGES-C criteria
Piskorz et al[4], 2020	GI mentor II	Gastroscopy and colonoscopy	8-hour work day	-	Performance data
Shirai et al[7], 2008	GI mentor II	EGD	1-hour training sessions	Five sessions over two weeks	The performance time taken by each trainee group to reach the various stages of EGD
Gerson and Van Dam[8], 2003	-	Sigmoidoscopy	The virtual reality group had unlimited time for 2 weeks. The traditional bedside group had only ten sessions to practice	-	Performance of the residents, including the number of procedures completed independently, those requiring assistance, flexure recognition, and retroflexion
Koch et al[9], 2015	GI mentor II	Colonoscopy	One session included at least five consecutive colonoscopies	They included 10-20 sessions distributed over 5 weeks to 10 weeks	The insertion depth and the rate of cecal intubation during patient-based assessments
Ferlitsch et al[10], 2002	GI mentor	Gastroscopy and colonoscopy	2-hour sessions	5 sessions a week, for 3 weeks	Performance of the trainees in the endobubble, endobasket, and virtual endoscopy cases
Silva Mendes et al[11], 2022	GI mentor II	EGD and colonoscopy	5-hour sessions	One session each week for about 3-4 weeks	Cognitive evaluation of the resident’s performance, endoscopy skill assessment in both EGD and colonoscopy
Buzink et al[12], 2007	GI mentor II	Colonoscopy	Not described	1 session per day for five consecutive days	Performance time
McIntosh et al[13], 2014	GI mentor II	Colonoscopy	The training was to be carried out for a total of between 10 hours and 20 hours	The sessions were distributed over four weeks	Number of proctor assists required per colonoscopy, the procedure time, and the rate of successful cecal intubation
Park et al[14], 2007	AccuTouch colonoscopy simulator	Colonoscopy	The training period was for a period of 2 hours to 3 hours	Not specified	The performance data includes caecum intubation, lack of flaws, and global ratings as assessed by an independent evaluator
Gomez et al[15], 2015	GI mentor II and Kyoto Kagaku colonoscopy	Colonoscopy	-	The total training was for 3 weeks	Performance time and the evaluator rating using the GAGES scores
Cohen et al[16], 2006	GI mentor	Colonoscopy	2-hour session on the simulators	Five sessions over 8 weeks	Objective and subjective competence of the fellows, and the discomfort scores
Grover et al[17], 2015	EndoVR virtual reality endoscopy simulator	Colonoscopy	8 hours of supervised endoscopy sessions	Not reported	JAG DOPS scale scores and performance data
Vajpeyi et al[18], 2025	M40 colonoscope training simulator	Colonoscopy	-	-	Performance data, including performance time
Gao et al[19], 2019	GI mentor II	EUS	-	All the training was carried out over 8 weeks	Test scores after completion of training
Arnold et al[20], 2015	GI mentor II	ERCP	-	-	Motion movement as demonstrated by the different groups of endoscopists according to the level of expertise
Bittner et al[21], 2010	GI mentor II	ERCP	They performed supervised 30-minute sessions	Each participant completed three cases	Validity of the VR tool as a simulator for ERCP
Sahakian et al[22], 2016	Endo VR	ERCP	The participant had unlimited time to complete cases	Each participant completed 6 cases	Procedure time and patient discomfort
Ferlitsch et al[23], 2010	GI mentor	Upper GI endoscopy	2-hour sessions	The participants completed training sessions for as long as 20 hours	Patient discomfort and pain
Georgiou et al[24], 2023	GI mentor	ERCP	3 procedures	-	Performance time
Sedlack[25], 2007	GI Mentor	EGD	15-minute introduction session, then the participant completed the training for 6 hours	The total simulation training included up to 20 simulated cases	Performance data, including procedure time and endoscopy skills
Sedlack et al[26], 2004	AccuTouch VR	Sigmoidoscopy	The participants had simulation training for a total of 3 hours	The total simulation training included 8-10 simulated cases	Subjective and objective assessment of the resident's sigmoidoscopy competency
Sedlack and Kolars[27], 2004	AccuTouch VR	Colonoscopy	The participants received simulation training for about 6 hours	The simulation involved 20-25 cases spread over two days	Objective assessment of the fellows’ performance, including performance time
Haycock et al[28], 2010	Endo TS-1 Olympus	Colonoscopy	The participants received four half-day sessions, leading to 16 hours of simulation training	The sessions were distributed over four days	Colonoscopy proficiency as assessed using the JAG DOPS and Global score
Di Giulio et al[29], 2004	GI mentor	EGD	The participants completed 10 hours of simulation training	The simulation involved 10 cases to be covered in 3 sessions to 5 sessions	Procedure time, rate of successful intubation, and the reasons for failed or assisted intubations
Ahlberg et al[30], 2005	AccuTouch VR	Colonoscopy	-	-	Procedure time and reason for procedure termination, if applicable
Grover et al[31], 2017	Accutouch VR	Colonoscopy	The participants received 4 hours of interactive didactic sessions and 6 hours of supervised simulator training	-	The performance of endoscopists in the two different topics was assessed by the JAG DOPS assessment tool
Ende et al[32], 2012	GI mentor, plastic phantom, and compactEASIE	EGD	The simulation training consisted of 2-hour sessions for a total of 10 hours	Five training sessions distributed over five weeks	Performance data including procedure time and the number of instances where residents required assistance
Hashimoto et al[33], 2018	GI mentor	Colonoscopy	Each participant had to complete a repetitive task in colonoscopy for the FES program	Ten repetitions for each task in the repetitions group. The proficiency group repeated each task until they met the proficiency benchmark in two consecutive tasks	Pass rate and FES scores of each of the participants

GI: Gastrointestinal; FES: Fundamentals of endoscopic surgery; GAGES-C: Global Assessment of Gastrointestinal Endoscopic Skills-Colonoscopy; EGD: Esophagogastroduodenoscopy; ERCP: Endoscopic retrograde cholangiopancreatography; VR: Virtual reality; JAG DOPS: Joint advisory group direct observation of procedural skills; EUS: Endoscopic ultrasound; VR: Virtual reality; AccuTouch VR: AccuTouch virtual reality simulator; EndoVR: Endo virtual reality simulator; TS-1 Olympus: Endoscopy simulator model name; compactEASIE: Compact compact Erlangen Active Simulator for Interventional Endoscopy.

Quality assessment

Figure 2 summarizes the risk of bias across the included studies. Only four studies were judged to have a low risk of bias. Among the remaining studies, concerns most commonly arose from limitations in the randomization process and lack of blinding. Notably, 10 studies were rated as having a high risk of bias due to absence of randomization.

Figure 2 The Risk of Bias 2 assessment tool.

EGD

Six included studies[8,12-16] evaluated the use of VR simulation in EGD training. Measures of post-training competency varied across studies, and only two reported quantitative trainee performance scores following VR training. Ende et al[32] compared VR simulation, clinical training alone, and clinical training supplemented with VR, finding that all approaches significantly improved manual skills from baseline (P < 0.05), with no significant differences between groups. In contrast, Shirai et al[7] compared VR-based and conventional training and reported that a significantly higher proportion of trainees in the traditional training group required assistance to complete the procedure, as indicated by a 1-point rating, compared with the VR group.

Independent procedure completion

Two studies assessed rates of independent procedure completion. Di Giulio et al[29] reported that simulation-based training significantly increased the likelihood of independent completion (odds ratio = 65.7; 95% confidence interval: 20.1-214.4). Similar findings were reported by Shirai et al[7].

Performance time

All included studies assessed differences in procedure time across intervention groups. Each study reported a significant reduction in procedure time from baseline following training. However, Di Giulio et al[29] found no significant difference between the VR and non-VR simulation groups (10.5 minutes vs 12.4 minutes, respectively). Similarly, Shirai et al[7] reported comparable procedure times between VR and conventional training groups (14.40 minutes vs 14.05 minutes, respectively). Ende et al[32] also observed no significant differences when comparing VR simulation alone, clinical training alone, or combined training approaches (P = 0.405).

ERCP

Four included studies evaluated the use of VR simulation in ERCP training[20-24], primarily examining performance differences between expert and novice endoscopists. Across studies, procedure times and task-specific performance consistently varied by expertise level. Georgiou et al[24] compared experts and novices across three ERCP scenarios. During bile duct stone removal, total procedure time did not differ between groups (273 ± 83 seconds vs 273 ± 38 seconds; P = 0.99), although experts required significantly less time to visualize the papilla (89 ± 29 seconds vs 124 ± 45 seconds; P = 0.02) and achieve deep cannulation (127 ± 27 seconds vs 160 ± 54 seconds; P = 0.028). In cystic duct leak management, overall procedure time was similar (P = 0.415); however, experts were significantly faster in papilla visualization (P = 0.008), deep cannulation (P < 0.001), diagnosis (P = 0.001), sphincterotomy (P = 0.001), and procedure completion (P = 0.01). No significant differences were observed for hilar stenosis across measured parameters[24]. Bittner et al[21] similarly reported superior performance by experts compared with novices across most ERCP metrics, supporting the face validity of VR simulation. Consistent findings were reported by Arnold et al[20] and Sahakian et al[22], demonstrating that VR simulation effectively discriminates between expert and novice ERCP performance.

EUS

Gao et al[19] uniquely evaluated the use of VR simulation in EUS training. The study compared combined VR simulation and clinical training with clinical training alone. EUS competency was assessed with a written examination scored out of 100. Trainees receiving combined training achieved significantly higher scores than those undergoing clinical training alone (64.53 ± 4.91 vs 60.09 ± 5.49; P = 0.028). Analysis of variance further demonstrated that VR simulation significantly improved EUS skills regardless of baseline endoscopy experience (P = 0.035)[19].

Colonoscopy and sigmoidoscopy

Most included studies evaluated the impact of VR simulation on colonoscopy training[6,9,13-18,27,28,30], whereas only two focused exclusively on sigmoidoscopy[12,29]. Outcomes and control conditions varied across studies, and endoscopist competency was assessed using diverse scoring systems. Hashimoto et al[33] compared different VR training strategies and reported higher Fundamentals of Endoscopic Surgery examination pass rates in VR-trained groups than in historical controls without simulation. Notably, proficiency-based VR training resulted in higher pass rates than repetition-based training (100% vs 88.9%). McIntosh et al[13] assessed competency using a 5-point Likert scale (1 to 5) and found that VR-trained trainees received significantly higher objective skill ratings than controls, as evaluated by both endoscopy nurses (2.56 ± 0.26 vs 2.05 ± 0.28; P = 0.001) and proctors (2.28 ± 0.21 vs 1.88 ± 0.45; P = 0.02). Similar results were reported by Park et al[14].

Independent colonoscopy completion rate

Parallel-arm trials also evaluated the proportion of trainees who completed the colonoscopy independently. McIntosh et al[13] assessed the need for assistance during colonoscopy and found that trainees in the simulator group required significantly fewer assists than those in the control group (1.94 vs 3.43; P = 0.004).

Performance time

McIntosh et al[13] compared colonoscope insertion times between VR simulator and control groups and found no statistically significant difference between trainees (14.4 vs 14.6; P = 0.37). No other included studies reported comparative procedure time outcomes between groups.

Sigmoidoscopy

Gerson and Van Dam[8] compared VR simulator training with traditional instruction and found that the traditional teaching group achieved significantly higher competency scores (rated on of 5-point scale) than the simulator cohort (3.8 ± 0.2 vs 2.9 ± 0.2; P < 0.001). A significantly greater proportion of trainees in the traditional group also completed sigmoidoscopy independently (72% vs 29%; P = 0.001)[8]. In contrast, Sedlack[25] reported no significant difference between simulator and control groups in independent procedure completion (8 vs 8; P = 0.893) or procedure time per case (24.10 vs 24 ± 1.1; P > 0.05).

DISCUSSION

VR simulators have been increasingly integrated into GIE training over recent years. Rather than serving as a replacement for traditional apprenticeship-based methods, VR simulation functions as an adjunct that enhances conventional endoscopic training[34]. In this review, VR simulation is defined as computer-based simulation technology that enables interactive, immersive experiences designed to replicate procedural and clinical aspects of endoscopy.

Our review included studies evaluating the use of VR simulation across upper GI endoscopic procedures, including ERCP, EUS, and EGD. In EGD training, the evidence suggests that VR simulation improves trainee independence compared with conventional training and enhances novice competency relative to baseline performance. These findings are supported by a Cochrane review reporting that VR-trained trainees completed significantly more procedures independently than those without training (relative risk: 1.25; 95% confidence interval: 1.13-1.39). However, our analysis also showed that improvements in endoscopic skills were generally comparable between VR-based and traditional training approaches. This indicates that VR simulation is at least as effective as conventional methods for endoscopist education. Importantly, incorporating VR into EGD training allows trainees to develop procedural skills before patient-based practice, thereby offering a meaningful advantage in terms of patient safety.

In ERCP training, the studies included in this review primarily evaluated the validity of VR simulators in distinguishing expert from novice performance. Across studies, expert endoscopists consistently outperformed novices across multiple ERCP metrics, particularly in the management of pancreatic pathologies such as cystic duct leakage[24]. These findings support the construct validity of VR simulation as a training and assessment tool for ERCP. However, additional studies are needed to determine whether VR-based training leads to measurable improvements in procedural skills and competency over time. Such evidence will be critical to guide the optimal integration of VR simulation into ERCP training curricula.

In colonoscopy and sigmoidoscopy training, most studies in this review demonstrated significant improvements in endoscopist competency following VR simulation. Several studies also reported higher rates of independent colonoscopy completion among VR-trained trainees compared with controls. However, findings were not entirely consistent. Qiao et al[35] reported no significant difference in independent colonoscopy completion between simulation-based and traditional training groups (89.3% vs 88.9%; P = 0.41). Similarly, sigmoidoscopy outcomes were heterogeneous; one study found higher independent completion rates with traditional instruction[33], whereas another reported no significant difference between training modalities[25]. These inconsistencies highlight variability in study design, outcome measures, and trainee experience, underscoring the need for standardized assessment frameworks in endoscopic training research.

Limitations of this study

The included studies displayed substantial methodological heterogeneity in trainee experience and specialty, assessed outcomes, and VR intervention designs. This variability precluded quantitative meta-analysis and limited our ability to identify the most effective strategies for implementing VR simulation in endoscopy training.

CONCLUSION

This article demonstrates that VR simulation is an effective training modality for GIE, with performance outcomes generally comparable to those of traditional training methods. These findings support integrating VR simulation into endoscopy training curricula to prepare trainees before patient-based practice and enhance patient safety. However, substantial heterogeneity in VR training approaches across existing studies limits definitive conclusions regarding optimal implementation. Future research should focus on developing and evaluating standardized VR training frameworks to determine best practices for integrating VR simulation into clinical endoscopy education.