Artificial intelligence in (gastrointestinal) healthcare: patients' and physicians' perspectives.

Artificial intelligence (AI) is entering into daily life and has the potential to play a significant role in healthcare. Aim was to investigate the perspectives (knowledge, experience, and opinion) on AI in healthcare among patients with gastrointestinal (GI) disorders, gastroenterologists, and GI-fellows. In this prospective questionnaire study 377 GI-patients, 35 gastroenterologists, and 45 GI-fellows participated. Of GI-patients, 62.5% reported to be familiar with AI and 25.0% of GI-physicians had work-related experience with AI. GI-patients preferred their physicians to use AI (mean 3.9) and GI-physicians were willing to use AI (mean 4.4, on 5-point Likert-scale). More GI-physicians believed in an increase in quality of care (81.3%) than GI-patients (64.9%, χ2(2) = 8.2, p = 0.017). GI-fellows expected AI implementation within 6.0 years, gastroenterologists within 4.2 years (t(76) = - 2.6, p = 0.011), and GI-patients within 6.1 years (t(193) = - 2.0, p = 0.047). GI-patients and GI-physicians agreed on the most important advantages of AI in healthcare: improving quality of care, time saving, and faster diagnostics and shorter waiting times. The most important disadvantage for GI-patients was the potential loss of personal contact, for GI-physicians this was insufficiently developed IT infrastructures. GI-patients and GI-physicians hold positive perspectives towards AI in healthcare. Patients were significantly more reserved compared to GI-fellows and GI-fellows were more reserved compared to gastroenterologists.

Automatic image and text-based description for colorectal polyps using BASIC classification.

Colorectal polyps (CRP) are precursor lesions of colorectal cancer (CRC). Correct identification of CRPs during in-vivo colonoscopy is supported by the endoscopist's expertise and medical classification models. A recent developed classification model is the Blue light imaging Adenoma Serrated International Classification (BASIC) which describes the differences between non-neoplastic and neoplastic lesions acquired with blue light imaging (BLI). Computer-aided detection (CADe) and diagnosis (CADx) systems are efficient at visually assisting with medical decisions but fall short at translating decisions into relevant clinical information. The communication between machine and medical expert is of crucial importance to improve diagnosis of CRP during in-vivo procedures. In this work, the combination of a polyp image classification model and a language model is proposed to develop a CADx system that automatically generates text comparable to the human language employed by endoscopists. The developed system generates equivalent sentences as the human-reference and describes CRP images acquired with white light (WL), blue light imaging (BLI) and linked color imaging (LCI). An image feature encoder and a BERT module are employed to build the AI model and an external test set is used to evaluate the results and compute the linguistic metrics. The experimental results show the construction of complete sentences with an established metric scores of BLEU-1 = 0.67, ROUGE-L = 0.83 and METEOR = 0.50. The developed CADx system for automatic CRP image captioning facilitates future advances towards automatic reporting and may help reduce time-consuming histology assessment.

Effects of a Personalized Smartphone App on Bowel Preparation Quality: Randomized Controlled Trial.

BACKGROUND: Adequate bowel preparation is essential for the visualization of the colonic mucosa during colonoscopy. However, the rate of inadequate bowel preparation is still high, ranging from 18% to 35%; this may lead to a higher risk of missing clinically relevant lesions, procedural difficulties, prolonged procedural time, an increased number of interval colorectal carcinomas, and additional health care costs. OBJECTIVE: The aims of this study are to compare bowel preparation instructions provided via a personalized smartphone app (Prepit, Ferring B V) with regular written instructions for bowel preparation to improve bowel preparation quality and to evaluate patient satisfaction with the bowel preparation procedure. METHODS: Eligible patients scheduled for an outpatient colonoscopy were randomized to a smartphone app group or a control group. Both the groups received identical face-to-face education from a research physician, including instructions about the colonoscopy procedure, diet restrictions, and laxative intake. In addition, the control group received written information, whereas the smartphone app group was instructed to use the smartphone app instead of the written information for the actual steps of the bowel preparation schedule. All patients used bisacodyl and sodium picosulfate with magnesium citrate as laxatives. The quality of bowel preparation was scored using the Boston Bowel Preparation Scale (BBPS) by blinded endoscopists. Patient satisfaction was measured using the Patient Satisfaction Questionnaire-18. RESULTS: A total of 87 patients were included in the smartphone app group and 86 in the control group. The mean total BBPS score was significantly higher in the smartphone app group (mean 8.3, SD 0.9) than in the control group (mean 7.9, SD 1.2; P=.03). The right colon showed a significantly higher bowel preparation score in the smartphone app group (mean 2.7, SD 0.5 vs mean 2.5, SD 0.6; P=.04). No significant differences were observed in segment scores for the mean transverse colon (mean 2.8, SD 0.4 vs mean 2.8, SD 0.4; P=.34) and left colon (mean 2.8, SD 0.4 vs mean 2.6, SD 0.5; P=.07). General patient satisfaction was high for the smartphone app group (mean 4.4, SD 0.7) but showed no significant difference when compared with the control group (mean 4.3, SD 0.8; P=.32). CONCLUSIONS: Our personalized smartphone app significantly improved bowel preparation quality compared with regular written instructions for bowel preparation. In particular, in the right colon, the BBPS score improved, which is of clinical relevance because the right colon is considered more difficult to clean and the polyp detection rate in the right colon improves with improvement of bowel cleansing of the right colon. No further improvement in patient satisfaction was observed compared with patients receiving regular written instructions. TRIAL REGISTRATION: ClinicalTrials.gov NCT03677050; https://clinicaltrials.gov/ct2/show/NCT03677050.

Evaluation of polypectomy quality indicators of large nonpedunculated colorectal polyps in a nonexpert, bowel cancer screening cohort.

BACKGROUND AND AIMS: With the introduction of the national bowel cancer screening program, the detection of sessile and flat colonic lesions ≥20 mm in size, defined as large nonpedunculated colorectal polyps (LNPCPs), has increased. The aim of this study was to examine the quality of endoscopic treatment of LNPCPs in the Dutch screening program. METHODS: This investigation comprised 2 related, but separate, substudies (1 with a cross-sectional design and 1 with a longitudinal design). The first examined prevalence and characteristics of LNPCPs in data from the national Dutch screening cohort from February 2014 until January 2017. The second, with screening data from 5 endoscopy units in the Southern part of the Netherlands from February 2014 until August 2015, examined performance on important quality indicators (technical and clinical successes, recurrence rate, adverse event rate, and surgery referral rate). All patients were part of the national Dutch screening cohort. RESULTS: In the national cohort, an LNPCP was detected in 8% of participants. Technical and clinical success decreased with increasing LNPCP size, from 93% and 96% in 20- to 29-mm lesions to 85% and 86% in 30- to 39-mm lesions and to 74% and 81% in ≥40-mm lesions (P < .001; P = .034). The cumulative recurrence rate at 12 months increased with LNPCP size, from 9% to 22% and 26% in the respective size groups (P = .095). The adverse event rate was 5%. The overall surgical referral rate for noninvasive LNPCPs was 7%. CONCLUSIONS: In this performance of 2 substudies, it was shown that quality parameters for endoscopic resection of large polyps in the Dutch screening cohort are not reached, especially in ≥30-mm polyps. Endoscopic resection of large polyps could benefit from additional training, quality monitoring, and centralization either within or between centers.

Optical diagnosis of colorectal polyp images using a newly developed computer-aided diagnosis system (CADx) compared with intuitive optical diagnosis.

BACKGROUND: Optical diagnosis of colorectal polyps remains challenging. Image-enhancement techniques such as narrow-band imaging and blue-light imaging (BLI) can improve optical diagnosis. We developed and prospectively validated a computer-aided diagnosis system (CADx) using high-definition white-light (HDWL) and BLI images, and compared the system with the optical diagnosis of expert and novice endoscopists. METHODS: CADx characterized colorectal polyps by exploiting artificial neural networks. Six experts and 13 novices optically diagnosed 60 colorectal polyps based on intuition. After 4 weeks, the same set of images was permuted and optically diagnosed using the BLI Adenoma Serrated International Classification (BASIC). RESULTS: CADx had a diagnostic accuracy of 88.3 % using HDWL images and 86.7 % using BLI images. The overall diagnostic accuracy combining HDWL and BLI (multimodal imaging) was 95.0 %, which was significantly higher than that of experts (81.7 %, P = 0.03) and novices (66.7 %, P < 0.001). Sensitivity was also higher for CADx (95.6 % vs. 61.1 % and 55.4 %), whereas specificity was higher for experts compared with CADx and novices (95.6 % vs. 93.3 % and 93.2 %). For endoscopists, diagnostic accuracy did not increase when using BASIC, either for experts (intuition 79.5 % vs. BASIC 81.7 %, P = 0.14) or for novices (intuition 66.7 % vs. BASIC 66.5 %, P = 0.95). CONCLUSION: CADx had a significantly higher diagnostic accuracy than experts and novices for the optical diagnosis of colorectal polyps. Multimodal imaging, incorporating both HDWL and BLI, improved the diagnostic accuracy of CADx. BASIC did not increase the diagnostic accuracy of endoscopists compared with intuitive optical diagnosis.

Optical diagnosis of diminutive polyps in the Dutch Bowel Cancer Screening Program: Are we ready to start?

Background and study aims Implementation of optical diagnosis of diminutive polyps may potentially increase the efficacy and cost-effectiveness of colonoscopies. To adopt such strategy in clinical practice, the Preservation and Incorporation of Valuable endoscopic Innovations (PIVI) thresholds provide the basis to be met: ≥ 90 % negative predictive value (NPV) for diagnosis of adenomatous histology and ≥ 90 % agreement on surveillance intervals. We evaluated this within the Dutch Bowel Cancer Screening Program (BCSP). Patients and methods Endoscopic and histological data were collected from participants of the national bowel cancer screening program with an unfavorable fecal immunochemical test referred for colonoscopy between February 2014 and August 2015 at four endoscopy centers. The "resect and discard" scenario was studied, resecting diminutive polyps without histological evaluation. Agreement between optical diagnosis and histological diagnosis was measured for surveillance intervals according to Dutch, European and American post-polypectomy surveillance guideline. Results Fifteen certified endoscopists participated in this study and included 3028 diminutive polyps. In 2,330 patients both optical and histological diagnosis were available. Optical diagnosis of diminutive polyps showed NPV of 84 % (95 % CI 80-87) for adenomatous histology in the rectosigmoid. Applying the 'resect and discard' strategy resulted in 90.6 %, 91.2 %, 90.9 % agreement on surveillance intervals for the Dutch, European and American guideline respectively. Conclusion Our data representing current clinical practice in the Dutch BCSP practice on optical diagnosis of diminutive polyps showed that accuracy of predicting histology remains challenging, and risk of incorrect optical diagnosis is still significant. Therefore, it is too early to safely implement these strategies.