pmc Endoscopy Endoscopy 3405 thiemesd 0215166 Endoscopy Endoscopy 0013-726X 1438-8812 pmc-is-collection-domain yes pmc-collection-title Thieme Open Access PMC13552228 PMC13552228.1 13552228 13552228 42235541 10.1055/a-2858-7084 ENDOS-2025-26417 1 Original article Learning and deskilling effects of artificial intelligence in colonoscopy among endoscopists with different levels of experience: a pragmatic, prospective trial http://orcid.org/0009-0009-0290-0567 Pedersen Tom Andre MD http://orcid.org/0000-0003-2262-0334 Mori Yuichi Prof. Botteri Edoardo Dr. Engjom Trond Dr. Seip Birgitte Dr. Dimcevski Georg Gjorgji Dr. Havre Roald Flesland Prof. 1 Department of Medicine 60498 Haukeland University Hospital Bergen Norway 2 Department of Medicine Haraldsplass Deaconess Hospital Bergen Norway 3 Bergen Research Group for Advanced Gastrointestinal Endoscopy (BRAGE), Department of Clinical Medicine (K1) 1658 University of Bergen Bergen Norway 4 Clinical Effectiveness Research Group 6305 University of Oslo Oslo Norway 5 Department of Colorectal Cancer Screening Cancer Registry of Norway, Norwegian Institute of Public Health Oslo Norway 6 Department of Research Cancer Registry of Norway, Norwegian Institute of Public Health Oslo Norway 7 Department of Clinical Medicine (K1) University of Bergen Bergen Norway 8 Department of Medicine 60512 Vestfold Hospital Trust Tønsberg Norway 9 Kanalspesialistene AS Bergen Norway Correspondence Tom Andre Pedersen MD Department of Medicine, Haukeland University Hospital Post Office Box 1400 Bergen 5021 Norway tom.andre.pedersen@helse-bergen.no 03 6 2026 9 2026 58 9 520043 1003 1014 22 12 2025 13 4 2026 01 06 2026 09 09 2026 09 09 2026 The Author(s). This is an open access article published by Thieme under the terms of the Creative Commons Attribution License, permitting unrestricted use, distribution, and reproduction so long as the original work is properly cited. (https://creativecommons.org/licenses/by/4.0/). 2026 The Author(s). https://creativecommons.org/licenses/by/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Background The effect of computer-aided detection (CADe) on performance of endoscopists with different experience levels is not well understood. This study assessed whether CADe use promotes learning or deskilling. Methods We performed a prospective, multicenter, registry-based, pragmatic clinical trial. The primary end point was change in the proportion of colonoscopies with detection of at least one polyp ≥5 mm (PDR-5). Each endoscopist performed colonoscopies in three successive phases: 1) before CADe exposure, 2) during CADe use, and 3) after CADe removal. To assess the impact of CADe on PDR-5, we compared phase 1 with phases 2 and 3, adjusting for patient and endoscopist characteristics and endoscopy performance measures. Generalized linear mixed models were used and presented according to endoscopist experience. Results 13 endoscopists (7 inexperienced, 6 experienced) performed 5013 colonoscopies; median patient age was 59 years (interquartile range 44–71) and 2703 (53.9%) were women. Patient sex and age were similar across the two experience groups, whereas indications differed (e.g. more inflammatory bowel disease in the inexperienced group). CADe temporarily increased PDR-5 among inexperienced endoscopists from phase 1 to 2 (31.9% [216/678] to 39.5% [257/651]; odds ratio [OR] 1.43, 95%CI 1.11–1.84). There was no significant change between phases 1 and 3 (31.9% [216/678] to 36.3% [173/476]; OR 1.03, 95%CI 0.79–1.34). There was no significant CADe effect on PDR-5 among experienced endoscopists between phases. Conclusions PDR-5 of inexperienced endoscopists increased when CADe was used. In non-CADe-assisted colonoscopy, no upskilling or deskilling was observed following a period of CADe exposure. Graphical Abstract Medtronic Europe 10.13039/100020192 ESGE (European Society of Gastrointestinal Endoscopy) pmc-status-qastatus 0 pmc-status-live yes pmc-status-embargo no pmc-status-released yes pmc-prop-open-access yes pmc-prop-olf no pmc-prop-manuscript no pmc-prop-legally-suppressed no pmc-prop-has-pdf yes pmc-prop-has-supplement yes pmc-prop-pdf-only no pmc-prop-suppress-copyright no pmc-prop-is-real-version no pmc-prop-is-scanned-article no pmc-prop-preprint no pmc-prop-in-epmc yes pmc-license-ref CC BY Introduction An increased adenoma detection rate (ADR) reduces the risk of post-colonoscopy colorectal cancer (CRC) and death . High-quality colonoscopies are vital for preventing and treating CRC. In recent years, artificial intelligence (AI) has been introduced into colonoscopy for polyp detection (i.e. computer-aided detection [CADe]), which is expected to increase ADR according to randomized trials . However, in contrast to such promising initial results, more recent real-world studies have questioned the effect of CADe . Furthermore, a newly published article even showed a deskilling effect after the introduction of CADe for polyp detection . Nevertheless, there is still a scarcity of evidence on whether the use of CADe contributes to the improvement or deskilling of endoscopists over time. Given the unresolved questions surrounding human–AI interaction in endoscopy, the European Society of Gastrointestinal Endoscopy (ESGE) recently published a curriculum aimed at ensuring the safe and effective use of AI in this field . To address the important question of learning effects from CADe, we evaluated how CADe use affects the polyp detection rate (PDR)-5, defined as the proportion of colonoscopies with detection of at least one polyp ≥5 mm in size; PDR-5 is a surrogate marker for ADR in Gastronet, the Norwegian national endoscopy quality registry . Gastronet utilizes PDR-5 as a convenient quality indicator for colonoscopy, as it correlates with ADR and does not require histological confirmation. In the present study, we aimed to compare PDR-5 of endoscopists in training across three consecutive phases: 1) before CADe exposure, 2) during CADe use, and 3) after CADe removal. We also investigated the same measures for experienced endoscopists. Methods Design and patients We performed a prospective, multicenter, registry-based, pragmatic clinical trial from November 2021 to June 2025, including patients who underwent colonoscopy at three Norwegian outpatient clinics: a university clinic (Haukeland University Hospital), a local hospital (Haraldsplass Deaconess Hospital), and a high-volume outpatient endoscopy practice (Kanalspesialistene). Each endoscopist aimed to perform at least 100 colonoscopies in each of three successive phases: phase 1 – before CADe exposure; phase 2– during CADe use; and phase 3 – after CADe removal. Each endoscopist completed the three consecutive phases in a fixed order, serving as their own controls. We divided the endoscopists into two groups: inexperienced and experienced. We defined “inexperienced” as an endoscopist who had performed between 100 and 1000 colonoscopies, and “experienced” as someone who had performed more than 1000 colonoscopies in their career before the study began. To perform a technically adequate procedure, the participating endoscopists had to have performed at least 100 colonoscopies in their career before inclusion began. The term “pragmatic” refers to a trial designed to assess clinical effectiveness in real-world settings, rather than measuring efficacy in an ideal or controlled environment. This type of trial aims to address the variability found in clinical practices and among patients . The colonoscopies were performed as part of the regular outpatient clinic with registration in our national endoscopy quality registry, Gastronet. To evaluate how CADe affects real-world clinical colonoscopy, we excluded patients with a positive fecal immunochemical test in the Norwegian CRC screening program. Screening was not performed in all participating centers, and much higher ADRs are achieved in screening colonoscopies than those performed for other indications, which could represent a possible selection bias in our study population . Gastronet collects information from both endoscopists and patients, providing data such as cecal intubation rate, bowel cleansing (as measured by the Boston Bowel Preparation Scale [BBPS]), withdrawal time, PDR-5, indication, use of sedation/analgesia, and patient-reported outcome measures, including pain grading and information provided before and after the procedure. Adequate bowel cleansing in Gastronet is defined as BBPS ≥6 with no segment score (right, transverse, left colon) being <2. Gastronet uses PDR-5 as a surrogate marker for ADR. The endoscopist estimates polyp size using a biopsy forceps or polypectomy snare as a reference. PDR-5 is readily available, not dependent on histology, and encompasses both adenomas and serrated polyps. A meta-analysis by Niv et al. examined the correlation between ADR and PDR (polyps of all sizes) and found a conversion factor of 0.68 to calculate ADR from PDR . In Gastronet, the registered withdrawal time is the time spent inspecting the mucosa during withdrawal of the colonoscope from the cecum to the anal canal, rounded to the nearest whole minute, also including polypectomies and/or biopsies if performed . The cases classified as CRC family/“screening” refer to surveillance colonoscopies in families with CRC or colonoscopies that are not part of the official CRC screening program. We used the CADe system GI Genius (Medtronic, Minneapolis, Minnesota USA). Cosmo Intelligent Medical Devices (Dublin, Ireland) developed this CADe system, and Medtronic distributes it. GI Genius is compatible with most endoscopy systems. Technical details have been described previously . We used CADe software version 2 for this study. Computer-aided diagnosis was not available to the endoscopists. Before the study began, the endoscopists underwent standard training on polyp characterization and the use of CADe. They were advised to activate the AI system when withdrawing the colonoscope from the cecum. The CADe system remained active throughout the entire inspection phase during the withdrawal of the colonoscope. The endoscopists used Olympus high-definition video processors and colonoscopes (Olympus Medical Systems, Tokyo, Japan). Study assumption and end points We hypothesized that real-time use of CADe during phase 2 would increase PDR-5 in both inexperienced and experienced endoscopists. We also hypothesized that the increased PDR-5 would persist even after CADe was removed in phase 3. Our primary end points were changes in PDR-5 between phases 1 and 2, and between phases 1 and 3, for both inexperienced and experienced endoscopists, respectively. Ethical considerations Gastronet uses an anonymous registration number linked to the patient, the endoscopist, and the center. Information in Gastronet is collected in accordance with the Norwegian Directorate of Health’s exemption from the duty of confidentiality. The Norwegian Data Protection Authority licenses the register. The concession does not require patient consent for registration; however, patients have the right to opt out of registration and are informed of the procedure in writing. The participating endoscopists provided informed consent for their registered data in Gastronet to be used. The endoscopists underwent a standard course that included adherence to the study protocol and instruction on data registration. The Regional Committee for Medical and Health Research Ethics (REC) in Western Norway approved the study (REC identification number 275068). Power analysis We anticipated a PDR-5 of 30% before CADe use and an improvement to 35% during CADe use (5% absolute improvement) for both experience groups. This expected 5% absolute improvement was based on publications reporting ADR improvement with CADe available at the time of study planning in 2021 . To detect a 5% absolute increase in PDR-5 with a two-sided test, an alpha of 0.05, and a power of 80%, we required 688 colonoscopies in phase 1 and 688 in phase 2 for both inexperienced and experienced groups. We chose to include the same number of colonoscopies in phase 3. This resulted in 4128 colonoscopies. Because some newer studies failed to demonstrate statistical significance, we increased the sample size to 5000 colonoscopies . With this adjustment, the study power increased to 87%. Statistical analysis Continuous variables were presented as medians with interquartile ranges (IQRs) or as means with SDs. Categorical variables were presented as frequencies and percentages. Pearson’s chi-squared test was used to compare categorical variables between groups, and Student’s t test was used to compare continuous variables between groups. Generalized linear mixed models were used to estimate the association between explanatory variables and the dependent variable, PDR-5. The following explanatory variables were entered as fixed effects: CADe phase, patient sex and age, colonoscopy indication, BBPS, cecal intubation rate, sedation/analgesia, inclusion time from first colonoscopy, and endoscopists’ experience. The endoscopist was included as a random effect (subject variable) to account for clustering of procedures within endoscopists. We also performed sensitivity analyses with center as a fixed effect, but the estimated effects of all covariates, as well as the overall conclusions, were not materially different from those that did not include center. For this reason, we did not include center in the model. The explanatory variables with P values of <0.1 in the univariable analyses were included in the multivariable analyses. Withdrawal time was not included in the generalized linear mixed models because it was only collected for patients without biopsies or polypectomies. Pain was also excluded, as only 46.7% (2341/5013) of patients returned the patient-reported outcome measures after endoscopy. We reported odds ratios (ORs) and 95%CIs. Statistical significance was defined as a P value of <0.05 using two-sided tests. Data were analyzed using IBM SPSS Statistics Version 30 (IBM Corp., Armonk, New York, USA). Results We included 5034 colonoscopies from 13 endoscopists who participated in all three phases ( ). After excluding 14 colonoscopies due to insufficient demographic data and 7 due to two colonoscopies being performed in the same patient, we analyzed 5013 colonoscopies performed by 7 inexperienced and 6 experienced endoscopists. Patient inclusion. The 13 endoscopists comprised 12 physicians and 1 surgeon. At the beginning of the study, the inexperienced endoscopist group had a mean age of 41.1 years, whereas the experienced group had a mean age of 50.7 years. Mean inclusion times from the first colonoscopy among the inexperienced vs. experienced groups were 6.8 vs. 2.2 months in phase 1, 17.7 vs. 7.6 months in phase 2, and 23.8 vs. 11.8 months in phase 3, respectively. Patient background characteristics across the three study phases were generally similar with respect to age and sex, in both endoscopist experience groups. However, the indications for colonoscopy differed slightly, with, for example, a higher proportion of symptomatic patients in phase 3. The inexperienced group had more patients with inflammatory bowel disease (IBD) in all phases, and the experienced group had more female patients in phases 1 and 2 ( ; see also Table 1s in the online-only Supplementary material). shows the number of colonoscopies performed per endoscopist and the PDR-5 in the three phases, respectively. Most endoscopists reached at least 100 colonoscopies in phases 1 and 2, but some inexperienced endoscopists performed only a few colonoscopies in phase 3. Table 1s also shows the quality indicators in the three phases for the different experience levels. Patient and endoscopist characteristics, and quality indicators. Total Phase 1 Phase 2 (CADe) Phase 3 P value Bold P values are significant. CADe, computer-aided detection; CRC, colorectal cancer; IBD, inflammatory bowel disease; IQR, interquartile range; PDR-5, proportion of colonoscopies with ≥1 polyp ≥5 mm in size. 1 “Screening” refers to surveillance colonoscopies in families with CRC or colonoscopies that are not part of the official CRC screening program. Patients, n (%) 5013 (100) Inexperienced 678 (100) 651 (100) 476 (100) 0.33 Experienced 1075 (100) 1088 (100) 1045 (100) 0.46 Women 2703 (53.9) Inexperienced 343 (50.6) 330 (50.7) 260 (54.6) Experienced 609 (56.7) 596 (54.8) 565 (54.1) Men 2310 (46.1) Inexperienced 335 (49.4) 321 (49.3) 216 (45.4) Experienced 466 (43.3) 492 (45.2) 480 (45.9) Age, median (IQR), years 59 (44–71) Inexperienced 56 (40–71) 54 (39–69) 61 (43–73) 0.10 Experienced 59 (45–70) 61 (46.25–71.75) 59 (45–70) 0.08 Indication, n (%) Inexperienced <0.001 Experienced 0.03 Symptoms 3015 (60.1) Inexperienced 353 (52.1) 333 (51.2) 300 (63.0) Experienced 658 (61.2) 660 (60.7) 711 (68.0) Surveillance (polyp, CRC) 580 (11.6) Inexperienced 71 (10.5) 107 (16.4) 69 (14.5) Experienced 123 (11.4) 126 (11.6) 84 (8.0) IBD 664 (13.2) Inexperienced 175 (25.8) 163 (25.0) 72 (15.1) Experienced 89 (8.3) 87 (8.0) 78 (7.5) CRC family/“screening” 1 262 (5.2) Inexperienced 42 (6.2) 8 (1.2) 11 (2.3) Experienced 78 (7.3) 69 (6.3) 54 (5.2) Other 458 (9.1) Inexperienced 33 (4.9) 35 (5.4) 23 (4.8) Experienced 118 (11.0) 138 (12.7) 111 (10.6) Missing 34 (0.7) Inexperienced 4 (0.6) 5 (0.8) 1 (0.2) Experienced 9 (0.8) 8 (0.7) 7 (0.7) Proportion of colonoscopies with at least one polyp ≥5 mm in size (PDR-5) per endoscopist in each phase. PDR-5 Phase 1, n/N (%) Phase 2, n/N (%) Phase 3, n/N (%) Experienced endoscopists (n = 6) 78/216 (36.1) 73/213 (34.3) 74/260 (28.5) 31/83 (37.3) 36/103 (35.0) 41/107 (38.3) 11/95 (11.6) 13/100 (13.0) 11/117 (9.4) 93/325 (28.6) 64/312 (20.5) 64/255 (25.1) 13/91 (14.3) 22/100 (22.0) 13/47 (27.7) 78/265 (29.4) 83/260 (31.9) 81/259 (31.3) Total 304/1075 (28.3) 291/1088 (26.7) 284/1045 (27.2) Inexperienced endoscopists (n = 7) 22/96 (22.9) 28/100 (28.0) 3/17 (17.6) 48/104 (46.2) 53/107 (49.5) 35/94 (37.2) 38/106 (35.8) 48/101 (47.5) 50/112 (44.6) 20/91 (22.0) 16/69 (23.2) 3/25 (12.0) 30/98 (30.6) 30/92 (32.6) 25/82 (30.5) 26/71 (36.6) 47/100 (47.0) 53/132 (40.2) 32/112 (28.6) 35/82 (42.7) 4/14 (28.6) Total 216/678 (31.9) 257/651 (39.5) 173/476 (36.3) All endoscopists 520/1753 (29.7) 548/1739 (31.5) 457/1521 (30.0) Univariable and multivariable analyses of PDR-5 and show the results of the univariable and multivariable analyses assessing factors associated with PDR-5. Separate analyses were performed for phase 1 vs. phase 2 ( , Table 2s ) and phase 1 vs. phase 3 ( , Table 3s ) to evaluate the effect of CADe (phase 2) and whether this effect persisted after its use (phase 3). The effects of CADe across study phases and endoscopist experience levels are illustrated in . Proportion (with 95%CI) of colonoscopies with at least one polyp ≥5 mm in size (PDR-5). Results from multivariable analyses. Proportion of colonoscopies with at least one polyp ≥5 mm in size (PDR-5) in phases 1 and 2. Variable/category PDR-5, n/N (%) Univariable analysis, OR (95%CI) Multivariable analysis, OR (95%CI) The univariable analyses were based on generalized linear mixed models, with the endoscopist as a clustering random effect. CADe, computer-aided detection; CRC, colorectal cancer; IBD, inflammatory bowel disease; OR, odds ratio. All univariable analyses with P < 0.1 were included in the multivariable analyses. 1 P < 0.05. 2 P < 0.10. 3 “Screening” refers to surveillance colonoscopies in families with CRC or colonoscopies that are not part of the official CRC screening program. Phase 1 (before CADe) (ref.) All 520/1753 (29.7) – – Inexperienced 216/678 (31.9) – – Experienced 304/1075 (28.3) – – 2 (CADe) All 548/1739 (31.5) 1.09 (0.94–1.26) – Inexperienced 257/651 (39.5) 1.37 (1.09–1.72) 1 1.43 (1.11–1.84) 1 Experienced 291/1088 (26.7) 0.92 (0.76–1.12) – Sex Women (ref.) All 515/1878 (27.4) – – Inexperienced 207/673 (30.8) – – Experienced 308/1205 (25.6) – – Men All 553/1614 (34.3) 1.39 (1.20–1.61) 1 1.33 (1.12–1.56) 1 Inexperienced 266/656 (40.5) 1.56 (1.24–1.97) 1 1.54 (1.20–1.98) 1 Experienced 287/958 (30.0) 1.28 (1.05–1.55) 1 1.14 (0.92–1.41) Age 5-year increase All 1068/3492 (30.6) 1.20 (1.17–1.23) 1 1.17 (1.14–1.20) 1 Inexperienced 473/1329 (35.6) 1.22 (1.18–1.27) 1 1.19 (1.15–1.24) 1 Experienced 595/2163 (27.5) 1.18 (1.14–1.22) 1 1.15 (1.11–1.19) 1 Indication Symptoms (ref.) All 595/2004 (29.7) – – Inexperienced 240/686 (35.0) – – Experienced 355/1318 (26.9) – – Surveillance (polyp, CRC) All 241/427 (56.4) 3.22 (2.58–4.02) 1 2.49 (1.97–3.14) 1 Inexperienced 117/178 (65.7) 3.73 (2.62–5.30) 1 2.69 (1.86–3.89) 1 Experienced 124/249 (49.8) 2.86 (2.15–3.82) 1 2.32 (1.71–3.13) 1 IBD All 82/514 (16.0) 0.42 (0.32–0.55) 1 0.52 (0.39–0.70) 1 Inexperienced 71/338 (21.0) 0.52 (0.38–0.71) 1 0.68 (0.49–0.95) 1 Experienced 11/176 (6.3) 0.21 (0.11–0.40) 1 0.25 (0.13–0.48) 1 CRC family/“screening” 3 All 45/197 (22.8) 0.84 (0.58–1.19) 0.83 (0.58–1.20) Inexperienced 18/50 (36.0) 1.34 (0.72–2.47) 1.36 (0.72–2.58) Experienced 27/147 (18.4) 0.66 (0.42–1.02) 2 0.70 (0.45–1.10) Other All 98/324 (30.2) 1.21 (0.93–1.58) 1.02 (0.78–1.35) Inexperienced 25/68 (36.8) 1.26 (0.74–2.15) 1.03 (0.59–1.79) Experienced 73/256 (28.5) 1.16 (0.85–1.58) 1.00 (0.73–1.38) Missing All 7/26 (26.9) 0.80 (0.32–1.95) 0.83 (0.33–2.12) Inexperienced 2/9 (22.2) 0.60 (0.12–2.97) 0.58 (0.10–3.28) Experienced 5/17 (29.4) 0.85 (0.29–2.54) 0.88 (0.29–2.71) Proportion of colonoscopies with at least one polyp ≥5 mm in size (PDR-5) in phases 1 and 3. Variable/category PDR-5, n/N (%) Univariable analysis, OR (95%CI) Multivariable analysis, OR (95%CI) The univariable analyses were based on generalized linear mixed models, with the endoscopist as a clustering random effect. CADe, computer-aided detection; CRC, colorectal cancer; IBD, inflammatory bowel disease; OR, odds ratio. All univariable analyses with P < 0.1 were included in the multivariable analyses. 1 P < 0.05. 2 P < 0.10. 3 “Screening” refers to surveillance colonoscopies in families with CRC or colonoscopies that are not part of the official CRC screening program. Phase 1 (before CADe) (ref.) All 520/1753 (29.7) – – Inexperienced 216/678 (31.9) – – Experienced 304/1075 (28.3) – – 3 (after CADe) All 457/1521 (30.0) 0.95 (0.82–1.11) – Inexperienced 173/476 (36.3) 1.03 (0.79–1.34) – Experienced 284/1045 (27.2) 0.93 (0.76–1.12) – Sex Women (ref.) All 494/1777 (27.8) – – Inexperienced 185/603 (30.7) – – Experienced 309/1174 (26.3) – – Men All 483/1497 (32.3) 1.25 (1.08–1.46) 1 1.23 (1.04–1.45) 1 Inexperienced 204/551 (37.0) 1.35 (1.06–1.73) 1 1.35 (1.04–1.76) 1 Experienced 279/946 (29.5) 1.19 (0.98–1.45) 2 1.12 (0.90–1.38) Age 5-year increase All 977/3274 (29.8) 1.20 (1.17–1.23) 1 1.17 (1.14–1.20) 1 Inexperienced 389/1154 (33.7) 1.20 (1.16–1.25) 1 1.18 (1.13–1.22) 1 Experienced 588/2120 (27.7) 1.19 (1.15–1.23) 1 1.17 (1.13–1.21) 1 Indication Symptoms (ref.) All 574/2022 (28.4) – – Inexperienced 217/653 (33.2) – – Experienced 357/1369 (26.1) – – Surveillance (polyp, CRC) All 190/347 (54.8) 3.07 (2.42–3.90) 1 2.37 (1.85–3.03) 1 Inexperienced 88/140 (62.9) 3.44 (2.34–5.04) 1 2.74 (1.84–4.08) 1 Experienced 102/207 (49.3) 2.84 (2.09–3.87) 1 2.18 (1.58–2.99) 1 IBD All 62/414 (15.0) 0.44 (0.33–0.60) 1 0.56 (0.41–0.76) 1 Inexperienced 47/247 (19.0) 0.50 (0.35–0.72) 1 0.66 (0.45–0.98) 1 Experienced 15/167 (9.0) 0.33 (0.18–0.58) 1 0.38 (0.21–0.69) 1 CRC family/“screening” 3 All 45/185 (24.3) 0.92 (0.64–1.32) 0.91 (0.63–1.31) Inexperienced 17/53 (32.1) 1.20 (0.65–2.22) 1.19 (0.63–2.23) Experienced 28/132 (21.2) 0.79 (0.51–1.24) 0.80 (0.51–1.26) Other All 100/285 (35.1) 1.51 (1.15–1.98) 1 1.29 (0.97–1.70) Inexperienced 20/56 (35.7) 1.25 (0.70–2.22) 1.04 (0.57–1.90) Experienced 80/229 (34.9) 1.57 (1.56–2.14) 1 1.37 (1.00–1.88) 2 Missing All 6/21 (28.6) 0.97 (0.37–2.59) 0.93 (0.34–2.55) Inexperienced 0/5 (0.0) – – Experienced 6/16 (37.5) 1.41 (0.49–4.07) 1.26 (0.42–3.73) CADe use was significantly associated with an increased PDR-5 among inexperienced endoscopists ( ) (OR 1.43, 95%CI 1.11–1.84). CADe did not have a significant effect on experienced endoscopists. When comparing phase 1 (before CADe) with phase 3 (after CADe), there were no significant effects on PDR-5 overall or within the two endoscopist groups ( ). Sex, age, and indication also significantly affected PDR-5 across all endoscopist groups. Time from first colonoscopy (divided into 3-month intervals) did not affect the PDR-5 significantly in any group ( Table 2s , Table 3s ). Endoscopist-level analysis When looking at the inexperienced group, three deskilled, two unchanged, and two upskilled between phases 1 and 3. Similarly, among the experienced group, three deskilled and three upskilled ( , ). Change in proportion of colonoscopies with at least one polyp ≥5 mm in size (PDR-5) between phases 1 and 3 at the endoscopist level (n = 13). Discussion We evaluated 5013 colonoscopies performed by seven inexperienced and six experienced endoscopists. An important finding with clinical impact was that CADe temporarily increased PDR-5 in inexperienced endoscopists, from 31.9% to 39.5% (OR 1.43, 95%CI 1.11–1.84) when CADe was used in phase 2. Another important finding was that we could not demonstrate upskilling or deskilling following CADe use. PDR-5 in non-CADe-assisted colonoscopies did not change significantly after CADe exposure, suggesting no obvious upskilling or deskilling (there was only a nonsignificant increase in PDR-5 after CAD exposure in the inexperienced group, from 31.9% to 36.3%). We observed no significant increase in PDR-5 during CAD use among experienced endoscopists. This differs from the study by Xu et al., which demonstrated an increased ADR regardless of experience . In fact, our study showed similar results for experienced endoscopists as the meta-analysis by Patel et al. and the study by Maas et al., which demonstrated no effect of CADe on ADR in a real-world setting among experienced endoscopists (nontrainees) . The experienced endoscopists had a higher mean age than the inexperienced group (50.7 vs. 41.1 years), which may influence the effect of CADe through a conservatism bias in the human–AI interaction . We did not observe any effect of CADe among experienced endoscopists, suggesting that their skill in detecting polyps was not affected by CADe. Experienced endoscopists may underestimate, or at least do not overestimate, polyp size, whereas inexperienced endoscopists may be more prone to overestimation. The experienced endoscopists may also be less enthusiastic about a software tool intended to improve their performance, leading to an aversion to using it effectively. Algorithm aversion is addressed as a potential obstacle for human–AI interaction in the ESGE curriculum for safe and effective use of AI in endoscopy . Some endoscopists found the sound signal especially disturbing, but the box overlay did not provoke this aversion, though false-positive markings could be exhausting . Low false-positive rates are essential for CADe usability. GI Genius has to be actively turned on by the endoscopist. In the current study, two endoscopists reported having CADe active throughout the entire examination (including intubation of the colon), while the remaining endoscopists activated CADe on withdrawal from the cecum. The finding of no significant effects of CADe on learning or deskilling among experienced endoscopists differs from the recent study by Budzyn et al., which found a significant deskilling effect among experienced endoscopists while they were exposed to CADe . A study by Troya et al. showed decreased eye travel distance when using CADe , which could explain why such deskilling happened in the study by Budzyn et al. However, there is a distinct difference in study design between our study and the Budzyn study. We evaluated the effect of exposure to CADe “after” CADe was removed, whereas Budzyn et al. investigated how endoscopists perform non-CADe-assisted colonoscopy “while” they were exposed to CADe in the other colonoscopies they performed during the same time period. Thus, it is considered that the effect of exposure to CADe will be stronger in the Budzyn study than in our study, suggesting that the deskilling effect may be a temporal phenomenon that disappears over time after CADe is removed. Looking at the endoscopist-level analysis among the inexperienced endoscopists, three deskilled, two unchanged, and two upskilled between phases 1 and 3. Among the experienced group, three deskilled and three upskilled. This also implies that there has been neither an upskilling nor deskilling effect over time. However, it is challenging to draw such a conclusion based solely on the change in PDR-5 between phases 1 and 3, given that there is always a time effect in which endoscopists may naturally upskill or deskill over time. Mean inclusion time from the first colonoscopy among the inexperienced group was 6.8 months in phase 1, 17.7 months in phase 2, and 23.8 months in phase 3. Such long-term exposure to a medical procedure may play a role in skill development, which needs to be considered in analysis of upskilling and deskilling. This “time effect” problem applies to the recently published paper by Okumura et al. , which showed that, after CADe implementation, detection rates without CADe were maintained and did not decline over time. Among non-CADe colonoscopies, high detectors showed accelerated learning curves, indicating they maintained a higher ADR, whereas low detectors had no significant change in their learning curves. However, the inclusion period was quite long (3 years), compromising the study interpretation due to the possible “time effect” on learning. In all phases, the inexperienced endoscopists had a higher PDR-5 than the experienced endoscopists. This may be related to inexperienced physicians overestimating polyp size or to their more meticulous mucosal evaluation than that of experienced physicians. The median withdrawal time across all phases was 10 minutes (IQR 8–16) for the inexperienced endoscopists and 9 minutes (IQR 7–11) for the experienced endoscopists. The withdrawal time must be evaluated with caution, however, as many examinations were excluded because of biopsies/therapy. Both experience groups had longer withdrawal times during the CADe phase. Other studies have shown similar results, with a slight increase in the withdrawal time using CADe . A study by Schult et al. showed that trainees detected more adenomas than consultants . Following an initial training period of 300 colonoscopies, ADR among trainees remained unchanged . This finding aligns with the study by Gianotti et al. , which demonstrated an increase in ADR after trainees performed more than 140 colonoscopies under supervision, and subsequently outperformed consultants. Thresholds of 325 colonoscopies in male patients and 539 colonoscopies in female patients were determined to be ideal for achieving adequate ADR. In the current study, we defined an experienced endoscopist as having performed ≥1000 colonoscopies during their career at the start of study inclusion. This may be an adequate cutoff when comparing our findings to those in these two previous studies . Our findings that inexperienced endoscopists outperform experienced endoscopists also align with these studies. A strength of our study is the large number of colonoscopies performed by 13 endoscopists, both inexperienced and experienced. The more procedures included per endoscopist, the greater the reliability of the evaluation of changes in PDR-5. We increased the number of included endoscopies to approximately 5000 after the meta-analysis by Patel et al. showed no significant effect of CADe among experienced (nontrainee) endoscopists in a real-world setting . Another strength is the three-phase design, in which all endoscopists included colonoscopies in all phases, serving as their own controls. The colonoscopies were included at three different levels of outpatient care: university hospital, local hospital, and high-volume outpatient endoscopy practice. The design reflects a variety of endoscopy settings and experience levels, making this study broadly relevant for real-world endoscopy. There are several limitations to this study. The most important is the use of PDR-5 as a quality indicator for lesion detection, rather than the gold standard ADR. A meta-analysis by Niv et al. found that a ratio of 0.68 can be used to estimate ADR from PDR for the individual endoscopist or a group of endoscopists before receiving formal results from the pathology department . PDR in colonoscopies is higher than PDR-5, suggesting that PDR-5 may have a ratio closer to 1 compared with PDR. PDR-5, being measured optically, may lead to over- or underestimation of polyp size, and studies have shown significant interphysician variation for estimating polyp size . To collect the large number of colonoscopies (approximately 5000), we decided to use PDR-5. We acknowledge that this quality indicator is less well-validated than ADR. Nevertheless, PDR-5 has been used for many years in Norway, and previous studies have adopted it . Another consideration was our aim to conduct a pragmatic study, in which endoscopists performed examinations according to their usual practice. Although polyps are sent to histology, we use PDR-5 as our quality indicator because it is both readily available and independent of histology. Notably, a strength of using PDR-5 is that it includes both adenomas and sessile serrated lesions, and the latter may account for a substantial proportion of interval cancers . It should be noted that this study may not fully capture the effect of CADe on polyp detection when only polyps ≥5 mm are recorded. However, some criticism of CADe is that it only increases the detection of diminutive polyps with very low malignancy potential. Focusing on polyps ≥5 mm may thus reveal whether CADe affects the more relevant polyps. Future studies are needed to validate PDR-5 as a colonoscopy quality indicator. Some inexperienced endoscopists performed relatively few endoscopies in phase 3, which may have led to underpowering of phase 3. This may have limited the ability to detect modest sustained effects of CADe. Future studies are needed to evaluate the learning effect of CADe among inexperienced endoscopists. As shown in previous studies and in the same region of Norway , patient age, sex, and colonoscopy indication significantly influence ADR/PDR-5. To adjust for this effect, we performed multivariable analyses. A possible limitation of the multivariable analyses is the risk of underpowering. We might have overlooked otherwise significant differences due to limited sample size. In particular, the inexperienced endoscopists had differences in the case-mix between phases 1 and 3, with higher age, more women, more symptoms, and fewer IBD patients in phase 3. This difference in case-mix may have influenced the PDR-5. Many of our patients in routine practice have the indications of symptoms and IBD, respectively. Other studies exclude patients with IBD, but we decided to include this indication as it reflects real-world data. However, it is a limitation that patients with IBD may exhibit different mucosal dysplasia features, which may not be detected by the training set used in GI Genius . Another limitation to the generalizability of this study is that the definition of an experienced endoscopist is not well established in the literature. Finally, we used only one type of AI device, which may limit the generalizability of our results. However, a study comparing CADe devices from different companies showed only minor differences between the producers . Overall, these results support a targeted deployment strategy for CADe, prioritizing its use among trainees and early-career endoscopists, where a clear PDR-5 benefit is observed. The added value for high-volume, experienced endoscopists in real-world practice may be limited. Additionally, the absence of an observable post-exposure reduction in PDR-5 compared with pre-exposure argues against the idea that exposure to CADe erodes skills, at least for nonexperienced endoscopists, although such skill reduction has been observed in another study involving only experienced endoscopists . Conclusion In this study, conducted in a real-world clinical setting, PDR-5 among inexperienced endoscopists increased during CADe use. In non-CADe-assisted colonoscopy, no upskilling or deskilling was observed after CADe exposure, regardless of the endoscopist experience level. Acknowledgement We thank Dr. Gert Huppertz-Hauss and the Gastronet secretariat for providing data. Conflict of Interest T.A. Pedersen has received a research grant from the European Society of Gastrointestinal Endoscopy and Medtronic (AI Research Award 2021). Y. Mori is a Co-Editor-in-Chief for Endoscopy; is a consultant for and has received speaker fees and device on loan from Olympus; and declares loyalty/ownership interest in Cybernet System Corporation. E. Botteri, T. Engjom, B. Seip, G.G. Dimcevski, and R.F. Havre declare that they have no conflicts of interest. Contributorsʼ Statement Tom Andre Pedersen: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Validation, Writing - original draft, Writing - review & editing. Yuichi Mori: Conceptualization, Methodology, Supervision, Writing - original draft, Writing - review & editing. Edoardo Botteri: Formal analysis, Methodology, Validation, Writing - review & editing. Trond Engjom: Conceptualization, Funding acquisition, Investigation, Methodology, Supervision, Validation, Writing - original draft, Writing - review & editing. Birgitte Seip: Conceptualization, Methodology, Supervision, Writing - review & editing. Georg Gjorgji Dimcevski: Conceptualization, Investigation, Methodology, Supervision, Writing - review & editing. Roald Flesland Havre: Conceptualization, Formal analysis, Investigation, Methodology, Project administration, Supervision, Validation, Visualization, Writing - original draft, Writing - review & editing. Supplementary Material Supplementary Material Zusatzmaterial References 1 Kaminski MF Regula J Kraszewska E Quality indicators for colonoscopy and the risk of interval cancer N Engl J Med 2010 362 1795 1803 10.1056/NEJMoa0907667 20463339 2 Corley DA Jensen CD Marks AR Adenoma detection rate and risk of colorectal cancer and death N Engl J Med 2014 370 1298 1306 10.1056/NEJMoa1309086 24693890 PMC4036494 3 Kaminski MF Wieszczy P Rupinski M Increased rate of adenoma detection associates with reduced risk of colorectal cancer and death Gastroenterology 2017 153 98 105 28428142 10.1053/j.gastro.2017.04.006 4 Hassan C Spadaccini M Mori Y Real-time computer-aided detection of colorectal neoplasia during colonoscopy: a systematic review and meta-analysis Ann Intern Med 2023 176 1209 1220 37639719 10.7326/M22-3678 5 Spadaccini M Menini M Massimi D AI and polyp detection during colonoscopy Cancers (Basel) 2025 17 797 40075645 10.3390/cancers17050797 PMC11898786 6 Soleymanjahi S Huebner J Elmansy L Artificial intelligence-assisted colonoscopy for polyp detection: a systematic review and meta-analysis Ann Intern Med 2024 177 1652 1663 10.7326/ANNALS-24-00981 39531400 7 Seager A Sharp L Neilson LJ Polyp detection with colonoscopy assisted by the GI Genius artificial intelligence endoscopy module compared with standard colonoscopy in routine colonoscopy practice (COLO-DETECT): a multicentre, open-label, parallel-arm, pragmatic randomised controlled trial Lancet Gastroenterol Hepatol 2024 9 911 923 39153491 10.1016/S2468-1253(24)00161-4 8 Patel HK Mori Y Hassan C Lack of effectiveness of computer aided detection for colorectal neoplasia: a systematic review and meta-analysis of nonrandomized studies Clin Gastroenterol Hepatol 2024 22 971 INF 38056803 10.1016/j.cgh.2023.11.029 9 Maas MHJ Rath T Spada C A computer-aided detection system in the everyday setting of diagnostic, screening, and surveillance colonoscopy: an international, randomized trial Endoscopy 2024 56 843 850 10.1055/a-2328-2844 38749482 PMC11524745 10 Budzyn K Romanczyk M Kitala D Endoscopist deskilling risk after exposure to artificial intelligence in colonoscopy: a multicentre, observational study Lancet Gastroenterol Hepatol 2025 10 896 903 10.1016/S2468-1253(25)00133-5 40816301 11 Mori Y Kopylov U Sinonquel P Curriculum for safe and effective use of artificial intelligence in endoscopy: European Society of Gastrointestinal Endoscopy (ESGE) Position Statement Endoscopy 2026 58 202 210 10.1055/a-2742-4342 41338282 12 Hoff G Botteri E Hoie O Polyp detection rates as quality indicator in clinical versus screening colonoscopy Endosc Int Open 2019 7 E195 E202 10.1055/a-0796-6477 30705953 PMC6338539 13 Randel KR Schult AL Botteri E Colorectal cancer screening with repeated fecal immunochemical test versus sigmoidoscopy: baseline results from a randomized trial Gastroenterology 2021 160 1085 INF 33227280 10.1053/j.gastro.2020.11.037 14 Niv Y Polyp detection rate may predict adenoma detection rate: a meta-analysis Eur J Gastroenterol Hepatol 2018 30 247 251 10.1097/MEG.0000000000001062 29293111 15 Pedersen TA Engjom T Dimcevski GG Differences in colonoscopy performance among four endoscopy centers in Western Norway: influence of case-mix Endosc Int Open 2025 13 a25469515 10.1055/a-2546-9515 PMC11996018 40230559 16 Cherubini A Dinh NN A review of the technology, training, and assessment methods for the first real-time AI-enhanced medical device for endoscopy Bioengineering (Basel) 2023 10 404 10.3390/bioengineering10040404 37106592 PMC10136070 17 Barua I Vinsard DG Jodal HC Artificial intelligence for polyp detection during colonoscopy: a systematic review and meta-analysis Endoscopy 2021 53 277 284 32557490 10.1055/a-1201-7165 18 Repici A Badalamenti M Maselli R Efficacy of real-time computer-aided detection of colorectal neoplasia in a randomized trial Gastroenterology 2020 159 512 INF 32371116 10.1053/j.gastro.2020.04.062 19 Spadaccini M Iannone A Maselli R Computer-aided detection versus advanced imaging for detection of colorectal neoplasia: a systematic review and network meta-analysis Lancet Gastroenterol Hepatol 2021 6 793 802 10.1016/S2468-1253(21)00215-6 34363763 20 Larsen SLV Mori Y Artificial intelligence in colonoscopy: a review on the current status DEN Open 2022 2 e109 10.1002/deo2.109 35873511 PMC9302306 21 Xu H Tang RSY Lam TYT Artificial intelligence-assisted colonoscopy for colorectal cancer screening: a multicenter randomized controlled trial Clin Gastroenterol Hepatol 2023 21 337 INF 35863686 10.1016/j.cgh.2022.07.006 22 Holzwanger EA Bilal M Glissen Brown JR Benchmarking definitions of false-positive alerts during computer-aided polyp detection in colonoscopy Endoscopy 2021 53 937 940 33137833 10.1055/a-1302-2942 PMC8386281 23 Troya J Fitting D Brand M The influence of computer-aided polyp detection systems on reaction time for polyp detection and eye gaze Endoscopy 2022 54 1009 1014 10.1055/a-1770-7353 35158384 PMC9500006 24 Okumura T Kudo SE Ide Y Long-term impact of computer-aided adenoma detection: a prospective observational study Endoscopy 2026 58 121 129 10.1055/a-2661-2624 40683257 25 Schult AL Hoff G Holme O Colonoscopy quality improvement after initial training: a cross-sectional study of intensive short-term training Endosc Int Open 2023 11 E117 E127 10.1055/a-1994-6084 36712907 PMC9879657 26 Gianotti RJ Oza SS Tapper EB A longitudinal study of adenoma detection rate in gastroenterology fellowship training Dig Dis Sci 2016 61 2831 2837 10.1007/s10620-016-4228-9 27405989 27 Taghiakbari M Djinbachian R Labelle J Endoscopic size measurement of colorectal polyps: a systematic review of techniques Endoscopy 2025 57 460 477 10.1055/a-2502-9733 39793610 28 Djinbachian R Taghiakbari M Alj A Virtual scale endoscope versus snares for accuracy of size measurement of smaller colorectal polyps: a randomized controlled trial Endoscopy 2025 57 443 450 10.1055/a-2475-0244 39557063 29 Hoff G Botteri E Huppertz-Hauss G The effect of train-the-colonoscopy-trainer course on colonoscopy quality indicators Endoscopy 2021 53 1229 1234 10.1055/a-1352-4583 33622001 30 Yeh JH Moi SH Chen CC Predominant serrated molecular signature in postcolonoscopy colorectal cancer: a systematic review and meta-analysis Am J Gastroenterol 2026 121 122 129 10.14309/ajg.0000000000003658 40699236 PMC12746767 31 Abdelrahim M Siggens K Iwadate Y New AI model for neoplasia detection and characterisation in inflammatory bowel disease Gut 2024 73 725 728 10.1136/gutjnl-2023-330718 38395438 32 Troya J Sudarevic B Krenzer A Direct comparison of multiple computer-aided polyp detection systems Endoscopy 2024 56 63 69 10.1055/a-2147-0571 37532115 PMC10736101