Artificial Intelligence Significantly Improves Adenoma Detection Rate but Does Not Affect Polyp Detection Rate in Colonoscopy: A Propensity Score Matching Study

Artificial Intelligence Significantly Improves Adenoma Detection Rate but Does Not Affect Polyp Detection Rate in Colonoscopy: A Propensity Score Matching Study | medRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-P4HH5NV'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search Artificial Intelligence Significantly Improves Adenoma Detection Rate but Does Not Affect Polyp Detection Rate in Colonoscopy: A Propensity Score Matching Study Han Byul Lee , John Mayen Ruben , Byeong Cheol Jeong , Jun Sik Yoon , Seung Jung Yu , Eun Jeong Choi , Dong Woo Kim , Nguyen Quang Thu , David Lee , Soonwhan Kang , Jaeyoung Lee , Eunhye Kang , View ORCID Profile Nguyen Phuoc Long , Hong Sub Lee doi: https://doi.org/10.1101/2025.11.09.25339868 Han Byul Lee 1 Department of Rehabilitation Medicine, Shihwa Medical Center , Gyeonggi-do 15034, Republic of Korea Find this author on Google Scholar Find this author on PubMed Search for this author on this site John Mayen Ruben 2 Department of Internal Medicine, Busan Paik Hospital, Inje University College of Medicine , Busan 47392, Republic of Korea Find this author on Google Scholar Find this author on PubMed Search for this author on this site Byeong Cheol Jeong 3 Onnuri Medical Internal Medicine , Busan 48523, Republic of Korea Find this author on Google Scholar Find this author on PubMed Search for this author on this site Jun Sik Yoon 2 Department of Internal Medicine, Busan Paik Hospital, Inje University College of Medicine , Busan 47392, Republic of Korea Find this author on Google Scholar Find this author on PubMed Search for this author on this site Seung Jung Yu 2 Department of Internal Medicine, Busan Paik Hospital, Inje University College of Medicine , Busan 47392, Republic of Korea Find this author on Google Scholar Find this author on PubMed Search for this author on this site Eun Jeong Choi 2 Department of Internal Medicine, Busan Paik Hospital, Inje University College of Medicine , Busan 47392, Republic of Korea Find this author on Google Scholar Find this author on PubMed Search for this author on this site Dong Woo Kim 2 Department of Internal Medicine, Busan Paik Hospital, Inje University College of Medicine , Busan 47392, Republic of Korea Find this author on Google Scholar Find this author on PubMed Search for this author on this site Nguyen Quang Thu 4 Graduate Institute of Biomedical Sciences, College of Medicine, Chang Gung University , Taoyuan 333, Taiwan Find this author on Google Scholar Find this author on PubMed Search for this author on this site David Lee 5 Ainex Corporation , Seoul 0617, Republic of Korea Find this author on Google Scholar Find this author on PubMed Search for this author on this site Soonwhan Kang 5 Ainex Corporation , Seoul 0617, Republic of Korea Find this author on Google Scholar Find this author on PubMed Search for this author on this site Jaeyoung Lee 5 Ainex Corporation , Seoul 0617, Republic of Korea Find this author on Google Scholar Find this author on PubMed Search for this author on this site Eunhye Kang 5 Ainex Corporation , Seoul 0617, Republic of Korea Find this author on Google Scholar Find this author on PubMed Search for this author on this site Nguyen Phuoc Long 4 Graduate Institute of Biomedical Sciences, College of Medicine, Chang Gung University , Taoyuan 333, Taiwan 6 Molecular Medicine Research Center, College of Medicine, Chang Gung University , Taoyuan 333, Taiwan Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Nguyen Phuoc Long For correspondence: phuoclong{at}mail.cgu.edu.tw hslee{at}paik.ac.kr Hong Sub Lee 2 Department of Internal Medicine, Busan Paik Hospital, Inje University College of Medicine , Busan 47392, Republic of Korea Find this author on Google Scholar Find this author on PubMed Search for this author on this site For correspondence: phuoclong{at}mail.cgu.edu.tw hslee{at}paik.ac.kr Abstract Full Text Info/History Metrics Supplementary material Data/Code Preview PDF ABSTRACT Background Colorectal cancer (CRC) remains a major cause of cancer-related morbidity and mortality worldwide. Endoscopy and adenoma removal are effective in reducing the incidence of CRC. Recent advances in artificial intelligence (AI)-assisted endoscopy have demonstrated the potential to improve detection outcomes. This study aimed to evaluate the effectiveness of AI-assisted endoscopy in improving adenoma detection rate (ADR) and polyp detection rate (PDR) during colonoscopy in a real-world clinical setting. Methods A total of 824 colonoscopies between August 2022 and February 2024 at Inje University Busan Paik Hospital were included in the study. Patients were divided into two groups: AI CAD-assisted colonoscopy (N = 393) and conventional colonoscopy (N = 431). Propensity score matching was then performed using a 1:1 nearest-neighbor algorithm, balancing key covariates, including age, sex, BMI, ASA score, bowel preparation quality, and the ratio of expert endoscopists. Ultimately, 786 patients (393 per group) were included in the final comparative analysis. Logistic regression analyses were used to evaluate the association between AI CAD-assisted colonoscopy and ADR and PDR, adjusting for potential confounders. Results ADR was significantly higher in the AI CAD-assisted group (41.5%) compared to the No-AI CAD group (34.4%) (adjusted OR = 1.380; 95% CI: 1.012–1.885; P = 0.042). PDR was higher in the AI CAD-assisted group (53.2% vs. 46.1%), but the difference was not statistically significant (OR = 1.312; 95% CI: 0.971–1.774; P = 0.077). Older age and higher BMI were positively associated with ADR, while male sex and higher ASA scores were negatively associated. Conclusions AI-assisted CAD colonoscopy was independently associated with improved ADR after adjustment for potential confounders. While the increase in PDR was not statistically significant, the findings support the clinical utility of AI CAD. Larger multicenter prospective studies are warranted to validate these findings and guide the integration of AI tools into routine endoscopic practice. INTRODUCTION Colorectal cancer (CRC) ranks third in incidence and second in mortality among all cancers worldwide [ 1 , 2 ]. In South Korea, colorectal cancer (CRC) is the third most common cancer [ 3 ], and the third leading cause of cancer-related deaths [ 4 ]. Because CRC often presents with symptoms such as occult blood, rectal bleeding, abdominal pain, and changes in bowel habits, mainly at advanced stages (Stage III or IV), early detection through screening methods such as blood tests [ 5 ], microsatellite instability testing [ 6 ], and colonoscopy [ 7 ], is essential. Detecting precancerous lesions such as adenomas and serrated lesions through colonoscopy and removing them with polypectomy is key to reducing mortality [ 8 ]. Among the identified risk factors, the typical precancerous lesions found in histopathological examinations of colorectal polyps are adenomas and serrated lesions. The adenoma-carcinoma pathway accounts for 70–90% of CRC cases, the serrated neoplasia pathway accounts for 10–20%, and microsatellite instability, such as in Lynch syndrome, contributes 2–7% [ 8 ]. Polyps found during screening colonoscopy undergo histopathological examination, and precancerous lesions are removed with polypectomy to prevent progression to colorectal cancer [ 9 , 10 ]. A large-scale, multicenter randomized trial published in 2022 involving adult men and women aged 55 to 64 years from Poland, Norway, Sweden, and the Netherlands showed that the 10-year cumulative risk of colorectal cancer was 0.98 (95% CI, 0.86–1.09) in the group that underwent screening colonoscopy, compared to 1.20 (95% CI, 1.10–1.29) in the group that received only usual care, which corresponds to an 18% relative risk reduction (risk ratio, 0.82; 95% CI, 0.70–0.93) [ 11 ]. Furthermore, studies have demonstrated a strong link between higher detection rates of polyps and adenomas and a lower incidence of colorectal cancer after colonoscopy, highlighting the importance of identifying and removing adenomas during screening [ 12 ]. A growing body of evidence has consistently demonstrated that polypectomy performed during colonoscopy is effective in preventing the development of colorectal cancer. In Korea, screening colonoscopy is recommended every five years for adults aged 50 and older [ 13 ]. The U.S. Preventive Services Task Force has recently updated its guidelines to recommend starting screening at age 45, lowering the age by five years compared to earlier recommendations [ 14 ]. Importantly, the adenoma detection rate (ADR) has long been established as a key quality indicator for colonoscopy [ 15 ], with guidelines recommending a minimum ADR of 35% (40% in men and 30% in women) when performed in asymptomatic individuals aged 45 years and older [ 16 ]. A multicenter study published in 2022 demonstrated a significant association between a median ADR of ≥ 28.3% and a reduced risk of colorectal cancer compared to lower ADRs [ 12 ]. Additionally, to ensure accurate detection and improve ADR, a minimum withdrawal time of six minutes is recommended. The cognitive demands on gastroenterologists during real-time endoscopic procedures are immense, requiring them to process a vast amount of visual data, often exceeding 30 images per second, while making immediate decisions about diagnosis and treatment. This growing cognitive load has led to the development of deep learning-based artificial intelligence (AI) systems, particularly in computer vision, to assist in these complex tasks. AI computer-aided detection (CAD) is a gastrointestinal endoscopy AI software medical device that assists in the easy and accurate detection of lesions, regardless of their size and shape, through deep neural network-based learning. AI CAD displays detected polyps in real-time during endoscopic examinations, enabling healthcare professionals to make immediate medical judgments and take necessary actions. Since 2019, prospective randomized trials have demonstrated that CAD in colonoscopy significantly improves both the polyp detection rate (PDR) and ADR [ 17 ]. These systems typically function by highlighting suspicious polyps on the endoscopic monitor with “alert boxes,” thereby aiding the endoscopist’s visual inspection. The current limitations of these systems, however, often lie in their reduced sensitivity for smaller or more subtle polyps. Future research aims to enhance the detection of flat lesions and improve the real-time differentiation between benign and precancerous polyps. A major challenge in developing AI systems is the initial investment cost. Nonetheless, it will become more efficient compared to the time-consuming and labor-intensive process of traditional colonoscopy [ 18 ]. Since 2019, many international studies have consistently shown that AI-assisted colonoscopy improves detection rates. In a randomized controlled trial, Wang et al. reported significantly higher ADR and PDR with an AI-assisted system called EndoScreener [ 19 ]. These findings were further supported by a subsequent double-blinded RCT by the same group in 2020, which also demonstrated significant improvements in both ADR and PDR [ 20 ]. Likewise, a randomized trial examining the efficacy of real-time CAD for colorectal neoplasia found a significantly higher ADR with the GI-Genius system compared to conventional colonoscopy [ 21 ]. The benefit of AI has also been observed in reducing miss rates. A considerably lower adenoma miss rate in the AI-assisted group compared to the conventional group in the CADeT-CS Trial [ 22 ]. Similarly, a 2022 multicenter study in Japan utilizing the EndoBRAIN-EYE system showed significant increases in both ADR and PDR [ 23 ]. However, the impact of AI CAD on ADR and PDR in real-life clinical settings remains to be elucidated. This study aimed to rigorously assess the impact of a novel AI-assisted polyp detection system, i.e., endoscopy as an AI CAD, on ADR in a real-world setting in a major medical center in South Korea. Using a retrospective review of electronic medical records from our institution, we compared ADR and PDR between colonoscopies performed with and without AI CAD assistance. This investigation provided robust evidence supporting the benefits of integrating AI into endoscopic practice to improve colorectal cancer screening outcomes. MATERIAL AND METHODS Population and Study Design This study was conducted at the Endoscopy Center within the Department of Internal Medicine at Inje University Busan Paik Hospital. The study was approved by the Institutional Review Board of Inje University Busan Paik Hospital (BPIRB2023-11-11). All procedures and data analyses were performed in accordance with the ethical guidelines outlined in the Declaration of Helsinki. Medical records for all colonoscopies performed between August 2022 and February 2024 were retrospectively reviewed. A total of 1,286 patients who underwent colonoscopy were initially enrolled. Patients were excluded if they were under 18 years of age, were pregnant, had poor bowel preparation (Boston Bowel Preparation Scale score < 6), or had a history of familial adenomatous polyposis (FAP), inflammatory bowel disease (IBD), or gastrointestinal cancer or surgery. Patients with incomplete baseline data were also excluded. Clinical outcomes were compared between two patient groups: those who underwent colonoscopy with AI CAD assistance and those who underwent conventional colonoscopy without AI assistance. The primary and secondary endpoints of the study were the ADR and PDR, respectively. These quality indicators were extracted from the medical records and were evaluated for differences between the two groups. Data Collection Procedure Out of 1,286 colonoscopy records, 462 records were excluded, leaving 824 records available for analysis. To minimize selection bias, 1:1 nearest-neighbor propensity score matching (PSM) was performed using the MatchIt package to adjust for potential confounding variables [ 24 ]. The covariates included in the model were age, gender, BMI, ASA score, BBPS, and the ratio of expert endoscopists. After PSM, 786 colonoscopy records remained. AI CAD-assisted colonoscopies (N = 393) and non-AI CAD-assisted colonoscopies (N = 393) were identified following the matching ( Figure 1 ). Of the 462 excluded cases, reasons included being under 18 years old, pregnancy, poor bowel preparation (Boston Bowel Preparation Scale < 6), a family history of familial adenomatous polyposis (FAP), a personal history of inflammatory bowel disease (IBD), and a history of gastrointestinal surgery. Download figure Open in new tab Figure 1. Flow chart for the selection of study participants. Abbreviations: AI CAD, Artificial intelligence computer-aided detection; FAP, Familial adenomatous polyposis; IBD, Inflammatory bowel disease; PSM: Propensity Score Matching. The variables selected for baseline characteristics included age, sex, body mass index (BMI), obesity status, American Society of Anesthesiologists (ASA) physical status classification score, Boston Bowel Preparation Scale (BBPS), examiner proficiency in colonoscopy, and indications for colonoscopy. Obesity status was categorized as non-obese (BMI < 25) or obese (BMI ≥ 25). The ASA score was 1 for a healthy physical status and 2 for mild systemic disease. The BBPS scores were grouped into two categories: scores of 3–5 represented regular bowel preparation, and 6–9 represented good bowel preparation. Colonoscopy proficiency was defined as inexperienced (less than 1 year of colonoscopy practice) or experienced (1 year or more of colonoscopy practice). Indications for colonoscopy were classified into three categories: (1) surveillance after polypectomy, (2) symptomatic examination, and (3) screening without prior disease history. Statistical Analysis Missing values were handled using listwise deletion. Continuous variables are reported as median with interquartile range (IQR), and categorical variables are presented as frequencies and percentages. To assess the normality of the distributions of continuous variables (e.g., age and BMI), the Shapiro-Wilk test was used. Because non-normal distributions were indicated (P < 0.05), the Mann-Whitney U test was used to compare these variables between the two groups. For categorical variables, including sex, obesity status, ASA score, Boston Bowel Preparation Scale (BBPS), indications for colonoscopy, and examiner proficiency, chi-square and Fisher’s exact tests were utilized to evaluate statistical differences in univariate analyses. A multivariable logistic regression analysis was performed with variables that demonstrated statistical significance in univariate analyses. The results were reported as odds ratios (ORs) with 95% confidence intervals (CIs) and corresponding P-values. All statistical analyses were performed using R software (version 4.2.1). All statistical tests were two-tailed, and P-values less than 0.05 were considered statistically significant. RESULTS Baseline Characteristics of the Study Before and After Propensity Score Matching A total of 824 patients who underwent colonoscopy were included in the study. Before PSM, baseline characteristics were compared between the AI CAD-assisted colonoscopy group and the conventional colonoscopy group. There were no significant differences in age, gender, body mass index (BMI), obesity status, or BBPS between the groups. However, a significant imbalance was observed in the American Society of Anesthesiologists (ASA) score and the percentage of procedures performed by expert endoscopists ( Supplementary Table 1 ). Following 1:1 PSM, a final cohort of 786 patients was established, with 393 patients in each group. After matching, all baseline characteristics, including age, gender, BMI, ASA score, BBPS, and expert endoscopist ratio, were well balanced between the groups (P > 0.05 for all comparisons, Table 1 ). View this table: View inline View popup Table 1. Baseline Characteristics After Propensity Score Matching (N = 786). Adenoma Detection Rate Was Significantly Higher in the AI CAD Group We first investigated the ADR of endoscopy with and without the assistance of AI CAD. Before PSM, the ADR was significantly higher in the AI CAD group compared to the conventional colonoscopy (No-AI CAD) group (41.5% vs. 33.6%; N = 824, P=0.024). The multivariable logistic regression analysis indicated that the use of AI CAD was significantly associated with an increased ADR (adjusted OR, 1.436; 95% CI, 1.056–1.956; P = 0.021) ( Supplementary Table 2 ). Other factors positively associated with ADR included BMI and age. On the contrary, being male and having a higher ASA score were negatively associated with ADR. View this table: View inline View popup Download powerpoint Table 2. Logistic Regression Analysis Evaluating the Effect of AI CAD on Adenoma Detection Rate (ADR) in 786 Patients. Abbreviation: AI CAD: Artificial intelligence computer-aided detection. After 1:1 propensity score matching, the ADR in the AI CAD group remained significantly higher (41.5%) than in the No-AI CAD group (34.4%) (N = 786, P = 0.047). The multivariable logistic regression analysis further confirmed that AI CAD use was independently associated with a higher ADR (adjusted OR, 1.380; 95% CI, 1.012–1.885; P = 0.042) ( Table 2 ). Similar to the pre-PSM analysis, BMI and age were positively associated with ADR, and male gender and a higher ASA score were negatively associated. Polyp Detection Rate Was Not Significantly Higher in the AI CAD Group Following the ADR, we further examined if AI CAD significantly improve the PDR in endoscopy before and after PSM. Before PSM, logistic regression analysis demonstrated a significant association between AI CAD use and a higher PDR (OR, 1.354; 95% CI, 1.006– 1.826; P = 0.046) ( Supplementary Table 3 ). Factors positively associated with PDR were age and BMI, whereas being male and a higher ASA score were negatively associated. After PSM, the PDR was higher in the AI CAD group (53.2%) compared with the No-AI CAD group (46.1%), though the difference was not statistically significant (OR, 1.312; 95% CI, 0.971– 1.774; P = 0.077) ( Table 3 ). Consistent with the pre-PSM findings, BMI and age were positively associated with PDR, while male gender and a higher ASA score were negatively associated. View this table: View inline View popup Download powerpoint Table 3. Logistic Regression Analysis Evaluating the Effect of AI CAD on Polyp Detection Rate (PDR) After Propensity Score Matching (N = 786). Abbreviation: AI CAD: Artificial intelligence computer-aided detection. DISCUSSION In this study, we evaluated the clinical utility of AI CAD regarding the ADR and PDR in colonoscopy. After applying propensity score matching to mitigate selection bias, we found a statistically significant increase in ADR with AI-assisted colonoscopy compared with conventional colonoscopy. Although the PDF was also higher in the AI group, this difference was not statistically significant. This improvement in ADR is particularly noteworthy. This finding aligns with prior studies, which have shown a similar increase of approximately 10% to 15% with AI-assisted colonoscopy [ 21 , 25 , 26 ]. The clinical significance of this finding is profound because an elevated ADR is a crucial quality metric directly linked to a reduced risk of interval colorectal cancer. Our results also suggest that AI-based systems can standardize detection quality, potentially reducing operator variability and helping endoscopists avoid missed lesions. AI-based systems, particularly those utilizing deep learning, are being increasingly applied in colonoscopies to provide real-time assistance by detecting and highlighting polyps. By identifying lesions that might otherwise be missed, these systems have been shown in multiple studies to improve both PDR and ADR [ 20 - 23 ]. There may be significant differences in detection rates between experienced and inexperienced colonoscopists. This emphasizes the need to further investigate the influence of colonoscopy experience. In the present study, the expert endoscopist ratio, that is, the proportion of experienced colonoscopists (those with more than one year of experience), was lower in the conventional colonoscopy group compared to experienced colonoscopists in the AI-assisted group (37.4% vs 63.9% before PSM, 41.0% vs 63.9% after PSM). Subsequent large-scale studies should examine whether a higher proportion of experienced colonoscopists might yield a more pronounced increase in ADR and PDR when assisted by AI. It is important to note that one study with a high endoscopy rate (78%) performed by experienced endoscopists found that ADR was not significantly higher in the AI-assisted group [ 27 ]. Additionally, the adenoma miss rate was considerably lower among AI-assisted novice colonoscopists compared to those without AI support, demonstrating a clear advantage of AI-assisted endoscopy. The study also showed that AI-assisted novice colonoscopists were not inferior compared to the experts [ 28 ]. Our work using the PSM study design provides a fairer comparison of the clinical utility of AI CAD, indicating a significantly higher ADR but not PDR in the AI-assisted group. Our findings support the clinical use of AI to improve colonoscopy quality metrics and the potential benefits on the prognosis of the patients. This study has several limitations that must be acknowledged. First, it was a retrospective, single-center study, which may limit the generalizability of our findings to other clinical settings. Second, although PSM was used to adjust for confounding variables, residual confounding from unmeasured factors, such as detailed withdrawal techniques or lesion morphology, could not be completely ruled out. Third, despite the relatively large overall sample size, subgroup analyses were limited by the number of procedures performed by each operator. We could not assess inter-examiner agreement and could only categorize colonoscopists by experience level rather than individual skill. CONCLUSIONS This study found that AI-assisted colonoscopy with AI CAD was associated with a higher detection rate of adenomas, particularly after excluding symptomatic patients and applying propensity score matching. However, the increase in PDR was not statistically significant in the AI CAD-assisted group compared to the no AI-CAD-assisted group. These findings underscore the potential of deep learning-based AI to enhance colonoscopy outcomes, which could improve colorectal cancer prevention. Future research, including multi-center, prospective studies with larger sample sizes and a more comprehensive set of clinical variables, is warranted to validate these results and evaluate the broader applicability of AI CAD in routine clinical practice. Data Availability All data produced in the present study are available upon reasonable request to the authors Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Funding None. Acknowledgement None. REFERENCES 1. ↵ Siegel RL , Wagle NS , Cercek A , Smith RA , Jemal A. Colorectal cancer statistics, 2023 . CA: A Cancer Journal for Clinicians . 2023 ; 73 ( 3 ): 233 – 54 . doi: 10.3322/caac.21772 . OpenUrl CrossRef PubMed 2. ↵ Sung H , Ferlay J , Siegel RL , Laversanne M , Soerjomataram I , Jemal A , et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries . CA: A Cancer Journal for Clinicians . 2021 ; 71 ( 3 ): 209 – 49 . doi: 10.3322/caac.21660 . OpenUrl CrossRef PubMed 3. ↵ Jung K-W , Won Y-J , Kang MJ , Kong H-J , Im J-S , Seo HG . Prediction of Cancer Incidence and Mortality in Korea, 2022 . Cancer Res Treat . 2022 ; 54 ( 2 ): 345 – 51 . doi: 10.4143/crt.2022.179 . OpenUrl CrossRef PubMed 4. ↵ Kang MJ , Won Y-J , Lee JJ , Jung K-W , Kim H-J , Kong H-J , et al. Cancer Statistics in Korea: Incidence, Mortality, Survival, and Prevalence in 2019 . Cancer Res Treat . 2022 ; 54 ( 2 ): 330 – 44 . doi: 10.4143/crt.2022.128 . OpenUrl CrossRef PubMed 5. ↵ Liang PS , Zaman A , Kaminsky A , Cui Y , Castillo G , Tenner CT , et al. Blood Test Increases Colorectal Cancer Screening in Persons Who Declined Colonoscopy and Fecal Immunochemical Test: A Randomized Controlled Trial . Clinical Gastroenterology and Hepatology . 2023 ; 21 ( 11 ): 2951 - 7.e2 . doi: 10.1016/j.cgh.2023.03.036 . OpenUrl CrossRef 6. ↵ Taieb J , Svrcek M , Cohen R , Basile D , Tougeron D , Phelip J-M. Deficient mismatch repair/microsatellite unstable colorectal cancer: Diagnosis, prognosis and treatment . European Journal of Cancer . 2022 ; 175 : 136 – 57 . doi: 10.1016/j.ejca.2022.07.020 . OpenUrl CrossRef PubMed 7. ↵ Komanduri S , Dominitz JA , Rabeneck L , Kahi C , Ladabaum U , Imperiale TF , et al. AGA White Paper: Challenges and Gaps in Innovation for the Performance of Colonoscopy for Screening and Surveillance of Colorectal Cancer . Clinical Gastroenterology and Hepatology . 2022 ; 20 ( 10 ): 2198 - 209.e3 . doi: 10.1016/j.cgh.2022.03.051 . OpenUrl CrossRef 8. ↵ Dekker E , Tanis PJ , Vleugels JLA , Kasi PM , Wallace MB . Colorectal cancer . The Lancet . 2019 ; 394 ( 10207 ): 1467 – 80 . doi: 10.1016/S0140-6736(19)32319-0 . OpenUrl CrossRef PubMed 9. ↵ Rex DK , Schoenfeld PS , Cohen J , Pike IM , Adler DG , Fennerty MB , et al. Quality indicators for colonoscopy . Gastrointestinal Endoscopy . 2015 ; 81 ( 1 ): 31 – 53 . doi: 10.1016/j.gie.2014.07.058 . OpenUrl CrossRef PubMed 10. ↵ Siddique S , Wang R , Yasin F , Gaddy JJ , Zhang L , Gross CP , et al. USPSTF Colorectal Cancer Screening Recommendation and Uptake for Individuals Aged 45 to 49 Years . JAMA Network Open . 2024 ; 7 ( 10 ): e2436358 – e . doi: 10.1001/jamanetworkopen.2024.36358 . OpenUrl CrossRef 11. ↵ Bretthauer M , Løberg M , Wieszczy P , Kalager M , Emilsson L , Garborg K , et al. Effect of Colonoscopy Screening on Risks of Colorectal Cancer and Related Death . New England Journal of Medicine . 2022 ; 387 ( 17 ): 1547 – 56 . doi: 10.1056/NEJMoa2208375 . OpenUrl CrossRef PubMed 12. ↵ Schottinger JE , Jensen CD , Ghai NR , Chubak J , Lee JK , Kamineni A , et al. Association of Physician Adenoma Detection Rates With Postcolonoscopy Colorectal Cancer . JAMA . 2022 ; 327 ( 21 ): 2114 – 22 . doi: 10.1001/jama.2022.6644 . OpenUrl CrossRef PubMed 13. ↵ Lee B-I , Hong SP , Kim S-E , Kim SH , Kim H-S , Hong SN , et al. Korean Guidelines for Colorectal Cancer Screening and Polyp Detection . Clin Endosc . 2012 ; 45 ( 1 ): 25 – 43 . doi: 10.5946/ce.2012.45.1.25 . OpenUrl CrossRef PubMed 14. ↵ Force USPST . Screening for Colorectal Cancer: US Preventive Services Task Force Recommendation Statement . JAMA . 2021 ; 325 ( 19 ): 1965 – 77 . doi: 10.1001/jama.2021.6238 . OpenUrl CrossRef PubMed 15. ↵ Zorzi M , Senore C , Da Re F , Barca A , Bonelli LA , Cannizzaro R , et al. Quality of colonoscopy in an organised colorectal cancer screening programme with immunochemical faecal occult blood test: the EQuIPE study (Evaluating Quality Indicators of the Performance of Endoscopy) . Gut . 2015 ; 64 ( 9 ): 1389 . doi: 10.1136/gutjnl-2014-307954 . OpenUrl Abstract / FREE Full Text 16. ↵ Rex DK , Anderson JC , Butterly LF , Day LW , Dominitz JA , Kaltenbach T , et al. Quality indicators for colonoscopy . Gastrointestinal Endoscopy . 2024 ; 100 ( 3 ): 352 – 81 . doi: 10.1016/j.gie.2024.04.2905 . OpenUrl CrossRef PubMed 17. ↵ Berzin TM , Topol EJ . Adding artificial intelligence to gastrointestinal endoscopy . The Lancet . 2020 ; 395 ( 10223 ): 485 . doi: 10.1016/S0140-6736(20)30294-4 . OpenUrl CrossRef 18. ↵ Shung DL , Byrne MF . How Artificial Intelligence Will Impact Colonoscopy and Colorectal Screening . Gastrointestinal Endoscopy Clinics of North America . 2020 ; 30 ( 3 ): 585 – 95 . doi: 10.1016/j.giec.2020.02.010 . OpenUrl CrossRef PubMed 19. ↵ Wang P , Berzin TM , Glissen Brown JR , Bharadwaj S , Becq A , Xiao X , et al. Real-time automatic detection system increases colonoscopic polyp and adenoma detection rates: a prospective randomised controlled study . Gut . 2019 ; 68 ( 10 ): 1813 . doi: 10.1136/gutjnl-2018-317500 . OpenUrl Abstract / FREE Full Text 20. ↵ Wang P , Liu X , Berzin TM , Glissen Brown JR , Liu P , Zhou C , et al. Effect of a deep-learning computer-aided detection system on adenoma detection during colonoscopy (CADe-DB trial): a double-blind randomised study . The Lancet Gastroenterology & Hepatology . 2020 ; 5 ( 4 ): 343 – 51 . doi: 10.1016/S2468-1253(19)30411-X . OpenUrl CrossRef PubMed 21. ↵ Repici A , Badalamenti M , Maselli R , Correale L , Radaelli F , Rondonotti E , et al. Efficacy of Real-Time Computer-Aided Detection of Colorectal Neoplasia in a Randomized Trial . Gastroenterology . 2020 ; 159 ( 2 ): 512 - 20.e7 . doi: 10.1053/j.gastro.2020.04.062 . OpenUrl CrossRef PubMed 22. ↵ Glissen Brown JR , Mansour NM , Wang P , Chuchuca MA , Minchenberg SB , Chandnani M , et al. Deep Learning Computer-aided Polyp Detection Reduces Adenoma Miss Rate: A United States Multi-center Randomized Tandem Colonoscopy Study (CADeT-CS Trial) . Clinical Gastroenterology and Hepatology . 2022 ; 20 ( 7 ): 1499 - 507.e4 . doi: 10.1016/j.cgh.2021.09.009 . OpenUrl CrossRef PubMed 23. ↵ Ishiyama M , Kudo S-e , Misawa M , Mori Y , Maeda Y , Ichimasa K , et al. Impact of the clinical use of artificial intelligence–assisted neoplasia detection for colonoscopy: a large-scale prospective, propensity score–matched study (with video) . Gastrointestinal Endoscopy . 2022 ; 95 ( 1 ): 155 – 63 . doi: 10.1016/j.gie.2021.07.022 . OpenUrl CrossRef PubMed 24. ↵ Ho D , Imai K , King G , Stuart EA . MatchIt: Nonparametric Preprocessing for Parametric Causal Inference . Journal of Statistical Software . 2011 ; 42 ( 8 ): 1 – 28 . doi: 10.18637/jss.v042.i08 . OpenUrl CrossRef 25. ↵ Lagström RMB , Bräuner KB , Bielik J , Rosen AW , Crone JG , Gögenur I , et al. Improvement in adenoma detection rate by artificial intelligence-assisted colonoscopy: Multicenter quasi-randomized controlled trial . Endosc Int Open . 2025 ; 13 ( continuous publication ). doi: 10.1055/a-2521-5169 . OpenUrl CrossRef 26. ↵ Engelke C , Graf M , Maass C , Tews HC , Kraus M , Ewers T , et al. Prospective study of computer-aided detection of colorectal adenomas in hospitalized patients . Scandinavian Journal of Gastroenterology . 2023 ; 58 ( 10 ): 1194 – 9 . doi: 10.1080/00365521.2023.2212309 . OpenUrl CrossRef PubMed 27. ↵ Schöler J , Alavanja M , de Lange T , Yamamoto S , Hedenström P , Varkey J. Impact of AI-aided colonoscopy in clinical practice: a prospective randomised controlled trial . BMJ Open Gastroenterology . 2024 ; 11 ( 1 ): e001247 . doi: 10.1136/bmjgast-2023-001247 . OpenUrl CrossRef PubMed 28. ↵ Yao L , Li X , Wu Z , Wang J , Luo C , Chen B , et al. Effect of artificial intelligence on novice-performed colonoscopy: a multicenter randomized controlled tandem study . Gastrointestinal Endoscopy . 2024 ; 99 ( 1 ): 91 - 9.e9 . doi: 10.1016/j.gie.2023.07.044 . OpenUrl CrossRef PubMed View the discussion thread. Back to top Previous Next Posted November 11, 2025. Download PDF Supplementary Material Data/Code Email Thank you for your interest in spreading the word about medRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Artificial Intelligence Significantly Improves Adenoma Detection Rate but Does Not Affect Polyp Detection Rate in Colonoscopy: A Propensity Score Matching Study Message Subject (Your Name) has forwarded a page to you from medRxiv Message Body (Your Name) thought you would like to see this page from the medRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Artificial Intelligence Significantly Improves Adenoma Detection Rate but Does Not Affect Polyp Detection Rate in Colonoscopy: A Propensity Score Matching Study Han Byul Lee , John Mayen Ruben , Byeong Cheol Jeong , Jun Sik Yoon , Seung Jung Yu , Eun Jeong Choi , Dong Woo Kim , Nguyen Quang Thu , David Lee , Soonwhan Kang , Jaeyoung Lee , Eunhye Kang , Nguyen Phuoc Long , Hong Sub Lee medRxiv 2025.11.09.25339868; doi: https://doi.org/10.1101/2025.11.09.25339868 Share This Article: Copy Citation Tools Artificial Intelligence Significantly Improves Adenoma Detection Rate but Does Not Affect Polyp Detection Rate in Colonoscopy: A Propensity Score Matching Study Han Byul Lee , John Mayen Ruben , Byeong Cheol Jeong , Jun Sik Yoon , Seung Jung Yu , Eun Jeong Choi , Dong Woo Kim , Nguyen Quang Thu , David Lee , Soonwhan Kang , Jaeyoung Lee , Eunhye Kang , Nguyen Phuoc Long , Hong Sub Lee medRxiv 2025.11.09.25339868; doi: https://doi.org/10.1101/2025.11.09.25339868 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Gastroenterology Subject Areas All Articles Addiction Medicine (568) Allergy and Immunology (863) Anesthesia (300) Cardiovascular Medicine (4435) Dentistry and Oral Medicine (444) Dermatology (382) Emergency Medicine (608) Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1509) Epidemiology (15227) Forensic Medicine (30) Gastroenterology (1124) Genetic and Genomic Medicine (6597) Geriatric Medicine (668) Health Economics (997) Health Informatics (4534) Health Policy (1368) Health Systems and Quality Improvement (1613) Hematology (540) HIV/AIDS (1264) Infectious Diseases (except HIV/AIDS) (15916) Intensive Care and Critical Care Medicine (1103) Medical Education (623) Medical Ethics (146) Nephrology (667) Neurology (6599) Nursing (346) Nutrition (998) Obstetrics and Gynecology (1144) Occupational and Environmental Health (957) Oncology (3332) Ophthalmology (974) Orthopedics (369) Otolaryngology (420) Pain Medicine (436) Palliative Medicine (130) Pathology (663) Pediatrics (1693) Pharmacology and Therapeutics (691) Primary Care Research (711) Psychiatry and Clinical Psychology (5447) Public and Global Health (9230) Radiology and Imaging (2198) Rehabilitation Medicine and Physical Therapy (1370) Respiratory Medicine (1196) Rheumatology (593) Sexual and Reproductive Health (712) Sports Medicine (530) Surgery (712) Toxicology (99) Transplantation (289) Urology (265) (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'a0035cde8b11ad07',t:'MTc3OTUzMTk5MA=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();