INTRODUCTION

Inflammatory bowel disease (IBD) is characterized by chronic intestinal inflammation with periods of remission and relapse [1] and categorized as ulcerative colitis, Crohn's disease, or indeterminate colitis based on a combination of clinical, endoscopic, and histologic data [1,2]. Currently, the gold standard for IBD diagnosis is ileocolonoscopy and esophagogastroduodenoscopy (EGD) combined with histology analysis [3]; however, novel machine and deep learning technologies present promising new avenues to significantly enhance our understanding of IBD and optimize remission standards to aim for better patient outcomes. Additional diagnostic modalities, including capsule endoscopy, magnetic resonance enterography, and computed tomography enterography are becoming more common in aiding diagnosis and evaluation [4,5], and the data produced from these techniques can be analyzed by machine and deep learning models to inform management of disease (Fig. 1). Although our understanding of the cause of IBD is still evolving [6], our treatment paradigms have markedly changed over the past decade, as biologic agents have become more prevalent [7], and machine and deep learning technologies are uniquely poised to revolutionize our approach to IBD treatment again. 

Box 1:

no caption available

FIGURE 1:

Machine learning models allow for the identification of disease features. Machine learning models can use data from endoscopy, imaging, and histology to detect lesions, identify strictures, and identify histological features of disease.

Artificial intelligence refers to computing methods capable of completing tasks that traditionally require human intelligence [8]. A summary of artificial intelligence concepts can be found in Table 1. To date, the majority of artificial intelligence technologies in medicine have been assistive, although the Food and Drug Administration (FDA) has approved a fully autonomous system to detect diabetic retinopathy [9]. Past use of computers to aid diagnosis and management of disease has relied on analyzing data using a concrete set of prescribed rules; however, novel machine learning methods allow researchers to ‘train’ or ‘teach’ computer models iteratively by exposing them to larger datasets annotated with clinical findings and disease outcomes [10]. Over time, the machine ‘learn’ patterns in the data and can predict specific outcomes from new data inputs. Assistive artificial intelligence was embraced in studying nongastrointestinal diseases like breast cancer [11–14], and interest in artificial intelligence in gastroenterology has increased over time. The number of publications displayed using a keyword search on PubMed of ‘artificial intelligence’ and ‘gastroenterology’ has increased from just 10 references in 2012 to more than 600 in 2022. Artificial intelligence technologies are particularly well suited to study IBD because diagnosis and management requires the synthesis of large, heterogeneous data such as endoscopic evaluation, radiologic evaluation, and histopathologic evaluation. These techniques can be leveraged to identify the mathematical functions that associate these different types of data, allowing for the integration of multiple diagnostic modalities into a unified model to evaluate disease. Furthermore, use of artificial intelligence may significantly improve clinical workflow by reducing the immense amount of time spent in interpreting data, allowing clinicians to spend more time interacting face-to-face with patients, leading to improved data collection and patient outcomes.

Table 1 - Definitions of artificial intelligence concepts Concept Definition Artificial intelligence Artificial intelligence refers to computational methods through which computers are able to complete activities that traditionally required human intelligence to complete Machine learning Artificial intelligence systems that allow a model or system to learn and improve iteratively as it is exposed to additional data Deep learning Machine learning models that are created using complex algorithms that are inspired by the organization of the human brain with many discrete nodes or ‘neurons’ and can identify important patterns or features in a dataset Autonomous artificial intelligence Autonomous artificial intelligence refers to when a system is making predictions and decisions without any real-time human input, for example, self-driving cars Assistive artificial intelligence Assistive artificial intelligence systems are designed to augment a human operator's ability to perform a certain task Convolutional neural network Convolutional neural networks are a type of deep learning model that uses a series of convolutional layers containing filters to detect patterns of increasing complexity in an input image as the data runs through the model
MACHINE LEARNING IN ENDOSCOPY

The gold-standard for IBD diagnosis is thorough examination of patient-derived biopsies collected during ileocolonoscopy and esophagogastroduodenoscopy. In addition to informing diagnoses, gross examination of the intubated bowel also provides important contextual information about the extent, continuity, and severity of inflammation. Unfortunately, current diagnostic methods cannot definitively classify some patients’ disease following initial examination. Establishing the correct diagnosis as early as possible is crucial because the treatment for ulcerative colitis and Crohn's disease can vary – especially in severe cases, where surgical interventions are considered. In order to address this challenge, Chierici et al.[15] developed a deep learning model to distinguish healthy versus inflamed tissue, ulcerative colitis from healthy tissue, and ulcerative colitis from Crohn's disease. Although the model's performance was best in the first two tasks, it was able to distinguish ulcerative colitis from Crohn's disease fairly accurately with a Matthew's correlation coefficient of 0.688. The results of their proof-of-concept study demonstrate that, with further investigation, a deep learning model capable of distinguishing ulcerative colitis from Crohn's disease could be a powerful tool for clinicians when diagnosing IBD. Although further work in the specific domain of diagnosis is necessary, there has been a surge in interest in leveraging artificial intelligence to assist in characterizing severity of IBD, predicting clinical outcomes, and identifying lesions using endoscopy data.

As target treatment outcomes have evolved over time for IBD, several indices for scoring the severity of inflammation on endoscopy have been developed [16–20]; however, these metrics have not been widely adopted and require clinicians to manually calculate the scores [21]. Several studies have investigated the use of machine learning models to score bowel inflammation severity – particularly in ulcerative colitis – using traditional endoscopy images and video as well as capsule endoscopy footage [22,23,24▪,25]. In one such study, Iacucci et al. developed a model, which predicted endoscopic and histologic remission in ulcerative colitis patients using white light endoscopy and virtual chromoendoscopy video data [23]. Separately, Tekenaka et al. created a deep learning model that can be implemented in real-time during endoscopy to predict histological remission and calculate an ulcerative colitis endoscopic index of severity score (UCEIS) [16,24▪]. Their model's histologic prediction agreed with pathologists’ assessment 97% of the time, demonstrating the model's ability to accurately assess inflammation on endoscopy.

In addition to grading inflammation severity, machine learning is currently being used to predict clinical outcomes based on endoscopy data. Maeda, et al. used an artificial intelligence system called EndoBRAIN-UC and endocytoscopy [26▪▪], a technique involving the use of a microscope attached to the endoscope, to predict the likelihood of active disease or healing in patients undergoing colonoscopy. They found that for patients with a prediction of active disease, the risk of clinical relapse over 12 months was 28.4% versus only 4.9% in patients with predictions of healing, highlighting the power of artificial intelligence tools to predict clinical outcomes and inform clinical management of IBD.

Lastly, there have been significant advances in utilizing machine learning models to identify ulcers and other mucosal features using capsule endoscopy data [27]. Ferreira et al.[28] developed a model that automatically detects ulcers in the small bowel capsule endoscopy data, whereas Afonso et al.[29] further developed the model to automatically quantify ulcers and predict their hemorrhagic potential. Furthermore, the model was able to analyze the footage in 16 min on average, whereas the same video could take a clinician up to 120 min to review. With adoption of this technology, the time spent reviewing capsule endoscopy footage could be spent seeing more patients.

MACHINE LEARNING IN RADIOLOGY

In addition to ileocolonoscopy and biopsy review, additional radiologic modalities are increasingly used to diagnose and characterize IBD, especially when small bowel inflammation is suspected. Researchers have investigated numerous metrics to assess intestinal inflammation and stricture using MRE data [30–32]; however, machine and deep learning are currently being explored as a mechanism to identify features associated with disease and aid diagnosis. The development of such indices for evaluation of IBD is paramount; however, development of these indices is hindered by relying on time-consuming, manual review of MRE data by researchers. machine learning techniques allow for partial or potentially full automation of this review process. Ziselman et al. trained a model with residual connections on annotated T2-weighted MRE scans from the ImageKids study, which accurately segmented MRE images of the terminal ileum and classified images as having normal or increased bowel wall thickness [33–35]. Notably, their model was trained on data that can be derived from MRE without intravenously administered contrast, increasing accessibility to this technique. Similarly, Arkko et al. developed a machine learning method to detect Crohn's disease using quantified motility measurements from MRE data, which performed well in the dataset with high-pixel resolution, low temporal resolution, and high temporal length [36]. However, their model's performance deteriorated by about 30% in the other two datasets tested, highlighting that further investigation is needed to validate this approach.

Computed tomography enterography (CTE) is less commonly used in the evaluation of IBD because of its requisite radiation exposure; however, there have been similar advances using machine learning for this diagnostic method. A deep learning model was developed and demonstrated better performance than human radiologists in assessing bowel fibrosis [37▪]. Additionally, researchers developed a convolutional neural network (CNN) to identify differences in volumetric visceral adipose tissue content visualized on CTE in patients with ulcerative colitis and Crohn's disease, which found that visceral adipose tissue phenotype differed between the two groups [38], demonstrating that CNN-based machine learning models can extract features from data that may inform diagnosis of IBD in the future.

Lastly, Guez et al. developed a multimodal random forest model that integrated MRE data with normalization of biochemical markers including C-reactive protein and fecal calprotectin, and they found that their fusion model could more accurately predict Crohn's disease than a linear regression model based on MaRIA [39,40]. This study exemplifies how machine and deep learning models can integrate multiple types to provide more accurate predictions than traditional statistical methods.

MACHINE LEARNING IN HISTOLOGY

Although mucosal healing is the primary target for the treatment of IBD, histologic remission may become an important and reasonable objective in the near future. The use of artificial intelligence can optimize these efforts using a multipronged approach: automate histopathologic scoring using existing scoring systems, extract meaningful histologic features of disease, and/or develop new metrics of disease activity.

Many histologic scoring indices have been developed for categorizing ulcerative colitis over the years. The Nancy index (NI) [41], was first published in 2017. Using a dataset of 200 images in patients with ulcerative colitis, a novel artificial intelligence system was developed in Nancy, France, to automate the assessment of the Nancy index. The average intra-class coefficient (ICC) among the three human annotators was 0.89, and remarkably, the average ICC between the histopathologists and the artificial intelligence tool was 0.87, illustrating the potential for using artificial intelligence to standardize histological scoring [42▪].

Histologic remission in ulcerative colitis, defined by the European Crohn's and Colitis Organisation (ECCO) as the absence of intraepithelial neutrophils, erosions, and ulcerations [43], has become an important objective of therapy. The Paddington International virtual ChromoendoScopy ScOre (PICaSSO) Histologic Remission Index (PHRI) was developed using biopsies from 307 patients across 11 centers in Europe and North America [18]. The study team wanted to develop a simple histologic index that correlated with published histology scores [41,44,45], endoscopy activity scores [16,46], clinical outcomes, and was conducive to automating scoring using artificial intelligence by focusing on a single-variable neutrophil infiltration. Since developing the PHRI, it has been used to develop multiple artificial intelligence algorithms. Gui et al.[47▪▪] developed a CNN-based model to detect neutrophils in whole slide images and subsequently classify images into either a histologic remission or nonremission class. Villanacci et al.[48] also developed a semi-supervised transfer learning model using images with pixel-level neutrophil annotations from patients with ulcerative colitis to predict the presence of neutrophils. Subsequently, a second model combined the features of all patches in each biopsy to classify the presence or absence of ulcerative colitis with high accuracy. These results demonstrate that designing new metrics that can easily be incorporated into digital pathology artificial intelligence systems to interpret histopathology data and inform clinical management of IBD.

Machine learning can also be used to extract meaningful histologic features of disease. Deep learning has previously been used to analyze intestinal histopathologic images of surgical specimens in patients with Crohn's disease, showing adipocyte shrinkage and mast cell infiltration to predict postoperative recurrence of Crohn's disease within 2 years after surgery [49▪]. Similarly, mucin depletion is a potential histologic indicator of clinical relapse in ulcerative colitis [50]; however, low inter-observer concordance limits its utility as a predictor. Recently, a deep learning model was trained to quantify goblet cell mucus area in whole slide images from patients with ulcerative colitis in endoscopic remission [51▪], and they found that patients with a greater ratio of goblet cell mucus area to the epithelial cell and goblet cell mucus area had a lower relapse rate over 12 months, demonstrating that deep learning models can help to standardize interpretation of image data. Furthermore, researchers developed a deep-learning model to identify eosinophils in tissue from ulcerative colitis patients and quantify eosinophil density [52]. From these calculations, they observed that patients on corticosteroid therapy had significantly lower eosinophil density, demonstrating the capability of deep learning models to identify unique disease features using histopathology images. However, there is still a gap in the literature on using machine learning-based histopathologic analysis methods to search for histologic features of dysplasia resulting from or comorbid with IBD.

CONSIDERATIONS FOR USING ARTIFICIAL INTELLIGENCE IN CLINICAL PRACTICE

Despite artificial intelligence's potential to revolutionize medical practice, several formidable obstacles must be addressed prior to the widespread implementation of artificial intelligence in the clinical setting. Although artificial intelligence is optimally suited to parse the wealth of modern clinical information and find novel insights in IBD (Fig. 2), training and deployment of these models are significantly hindered by the critical lack of large, publicly available, pathologist-annotated datasets, the lack of data sharing, and the relative rarity of IBD. Furthermore, the controls used for comparison are often tissue with no overt signs of inflammation or tissue obtained from participants with other gastrointestinal diseases. Thus, no clear reference dataset exists for ‘what is normal’. The creation of such reference datasets is necessary to allow for extensive testing of artificial intelligence systems so that they can be leveraged to achieve better clinical outcomes. Another limitation of using artificial intelligence comes from having insufficient or inconsistent datasets to use for modeling. When models are trained on a single dataset from one site, the images are often consistent because of standardized tissue staining and slide digitization. Importantly, there may be considerable image differences when using images from other sites limiting the generalizability of models.

FIGURE 2:

Summary of machine learning capabilities in inflammatory bowel disease.

Because of the promise of decreasing human error, artificial intelligence has been lauded as an objective means of assisting in various areas of clinical practice, including disease diagnosis, prevention, monitoring, therapy, and long-term management. Along with immense promise, the use of artificial intelligence in medical practice comes with the necessity to consider the ethical principles of medicine: justice, beneficence, nonmaleficence, and autonomy. As the improper use of artificial intelligence can potentially violate all three of these principles, researchers and doctors must hold each other accountable while creating and using these technologies. The use of artificial intelligence without extensive evidence of improved patient outcomes would violate nonmaleficence, so it is crucial to have a ‘human in the loop’ to oversee the implementation of and decisions of artificial intelligence systems to ensure that they are not causing harm to patients. Additionally, using datasets that are not expansive and diverse can lead to bias in algorithms that could affect how certain groups receive care. Lastly, the implementation of artificial intelligence in clinical practice without informed consent from a patient would violate their autonomy [53,54]. Machine and deep learning methods have the potential to transform the way that physicians diagnose and manage IBD drastically; however, significant care must be taken in developing and implementing these technologies to ensure that they help achieve better clinical outcomes.

CONCLUSION

Machine and deep learning models are becoming increasingly common in medicine. Although further studies are necessary before implementing these technologies in a clinical setting, researchers have demonstrated their utility in aiding the endoscopic, radiologic, and histologic evaluation of IBD. Although there are obstacles to implementing these models in the clinical setting, we will likely see these technologies develop and seek regulatory approval for clinical use in the near future.

Acknowledgements

We would like to thank Adam Greene, PhD, for his editorial insight during the preparation of our manuscript. Figure 1 and 2 were created using Biorender.com.

Financial support and sponsorship

Research reported in this publication was supported by National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) of the National Institutes of Health under award number K23DK117061-01A1 (S.S.) and award number R01DK131491 (S.S.).

Conflicts of interest

There are no conflicts of interest.

REFERENCES AND RECOMMENDED READING

Papers of particular interest, published within the annual period of review, have been highlighted as:

▪ of special interest

▪▪ of outstanding interest

REFERENCES 1. Saez A, Herrero-Fernandez B, Gomez-Bris R, et al. Pathophysiology of inflammatory bowel disease: innate immune system. Int J Mol Sci 2023; 24:1526. 2. Gergely M, Prado E, Deepak P. Management of refractory inflammatory bowel disease. Curr Opin Gastroenterol 2022; 38:347–357. 3. Parigi TL, Mastrorocco E, Da Rio L, et al. Evolution and new horizons of endoscopy in inflammatory bowel diseases. J Clin Med 2022; 11:872. 4. Shaban N, Hoad CL, Naim I, et al. Imaging in inflammatory bowel disease: current and future perspectives. Frontline Gastroenterol 2022; 13:E28–E34. 5. Limpias Kamiya KJL, Hosoe N, Hayashi Y, et al. Video capsule endoscopy in inflammatory bowel disease. DEN Open 2021; 2:E26. 6. Krela-Kaźmierczak I, Zakerska-Banaszak O, Skrzypczak-Zielińska M, et al. Where do we stand in the behavioral pathogenesis of inflammatory bowel disease? The western dietary pattern and microbiota—a narrative review. Nutrients 2022; 14:2520. 7. Turner D, Ricciuto A, Lewis A, et al. International Organization for the Study of IBD. STRIDE-II: An Update on the Selecting Therapeutic Targets in Inflammatory Bowel Disease (STRIDE) Initiative of the International Organization for the Study of IBD (IOIBD): determining therapeutic goals for treat-to-target strategies in IBD. Gastroenterology 2021; 160:1570–1583. 8. Crossnohere NL, Elsaid M, Paskett J, et al. Guidelines for artificial intelligence in medicine: literature review and content analysis of frameworks. J Med Internet Res 2022; 24:e36823. 9. FDA. DE NOVO CLASSIFICATION REQUEST FOR IDX-DR [Internet]. 2018 [cited 4 March 2023]. Available at: https://www.accessdata.fda.gov/cdrh_docs/reviews/DEN180001.pdf. [Accessed 5 January 2023] 10. Chen D, Fulmer C, Gordon IO, et al. Application of artificial intelligence to clinical practice in inflammatory bowel disease - what the clinician needs to know. J Crohn's Colitis 2022; 16:460–471. 11. Chan RCK, To CKC, Cheng KCT, et al. Artificial intelligence in breast cancer histopathology. Histopathology 2023; 82:198–210. 12. Nassif AB, Talib MA, Nasir Q, et al. Breast cancer detection using artificial intelligence techniques: a systematic literature review. Artificial Intell Med 2022; 127:102276. 13. Baxi V, Lee G, Duan C, et al. Association of artificial intelligence-powered and manual quantification of programmed death-ligand 1 (PD-L1) expression with outcomes in patients treated with nivolumab ± ipilimumab. Modern Pathol 2022; 35:1529–1539. 14. Hunter B, Hindocha S, Lee RW. The role of artificial intelligence in early cancer diagnosis. Cancers (Basel) 2022; 14:1524. 15. Chierici M, Puica N, Pozzi M, et al. Automatically detecting Crohn's disease and ulcerative colitis from endoscopic imaging. BMC Med Inform Decis Mak 2022; 22:300. 16. Travis SPL, Schnell D, Krzeski P, et al. Developing an instrument to assess the endoscopic severity of ulcerative colitis: The Ulcerative Colitis Endoscopic Index of Severity (UCEIS). Gut 2012; 61:535–542. 17. Lobatón T, Bessissow T, de Hertogh G, et al. The Modified Mayo Endoscopic Score (MMES): a new index for the assessment of extension and severity of endoscopic activity in ulcerative colitis patients. J Crohns Colitis 2015; 9:846–852. 18. Iacucci M, Daperno M, Lazarev M, et al. Development and reliability of the new endoscopic virtual chromoendoscopy score: the PICaSSO (Paddington International Virtual ChromoendoScopy ScOre) in ulcerative colitis. Gastrointest Endosc 2017; 86:1118.e5–1127.e5. 19. Adler J, Eder SJ, Gebremariam A, et al. Development and testing of a new simplified endoscopic mucosal assessment for Crohn's disease: the SEMA-CD. Inflamm Bowel Dis 2021; 27:1585–1592. 20. Daperno M, Haens GD’, van Assche G, et al. Development and validation of a new, simplified endoscopic activity score for Crohn's disease: the SES-CD. 21. Mohammed Vashist N, Samaan M, Mosli MH, et al. Endoscopic scoring indices for evaluation of disease activity in ulcerative colitis. Cochrane Database Syst Rev 2018; 2018:CD011450. 22. Sutton RT, Zaïane OR, Goebel R, Baumgart DC. Artificial intelligence enabled automated diagnosis and grading of ulcerative colitis endoscopy images. Sci Rep 2022; 12:2748. 23. Iacucci M, Cannatelli R, Parigi TL, et al. PICaSSO group. A virtual chromoendoscopy artificial intelligence system to detect endoscopic and histologic activity/remission and predict clinical outcomes in ulcerative colitis. Endoscopy 2022; 55:332–341. 24▪. Takenaka K, Fujii T, Kawamoto A, et al. Deep neural network for video colonoscopy of ulcerative colitis: a cross-sectional study. Lancet Gastroenterol Hepatol 2022; 7:230–237. 25. Higuchi N, Hiraga H, Sasaki Y, et al. Automated evaluation of colon capsule endoscopic severity of ulcerative colitis using ResNet50. PLoS One 2022; 17:e0269728. 26▪▪. Maeda Y, Kudo SE, Ogata N, et al. Evaluation in real-time use of artificial intelligence during colonoscopy to predict relapse of ulcerative colitis: a prospective study. Gastrointest Endosc 2022; 95:747.e2–756.e2. 27. Xie X, Xiao YF, Zhao XY, et al. Development and validation of an artificial intelligence model for small bowel capsule endoscopy video review. JAMA Network Open 2022; 5:E2221992. 28. Ferreira JPS, de Mascarenhas Saraiva MJCDSDQEC, Afonso JPL, et al. Identification of ulcers and erosions by the novel PillcamTM Crohn's capsule using a convolutional neural network: a multicentre pilot study. J Crohns Colitis 2022; 16:169–172. 29. Afonso J, Saraiva MM, Ferreira JPS, et al. Automated detection of ulcers and erosions in capsule endoscopy images using a convolutional neural network. Med Biol Eng Comput 2022; 60:719–725. 30. Coimbra A, Rimola J, Cuatrecasas M, et al. Magnetic resonance enterography and histology in patients with fibrostenotic Crohn's disease: a multicenter study. Clin Transl Gastroenterol 2022; 13:e00505. 31. Loch FN, Kamphues C, Beyer K, et al. Diagnostic accuracy of magnetic resonance enterography for the evaluation of active and fibrotic inflammation in Crohn's disease. Front Surg 2022; 9:872596. 32. Avila F, Caron B, Hossu G, et al. Magnetic resonance elastography for assessing fibrosis in patients with Crohn's disease: a pilot study. Dig Dis Sci 2022; 67:4518–4524. 33. Ziselman Y, Shinnawi FH, Greer ML, et al. Gadolinium-Free Crohn's Disease Assessment from Magnetic Resonance Enterography Data. In: Proceedings - International Symposium on Biomedical Imaging. IEEE Computer Society. 2022. 34. Weiss B, Turner D, Griffiths A, et al. ImageKids study group. Simple endoscopic score of Crohn disease and magnetic resonance enterography in children: report from ImageKids Study. J Pediatr Gastroenterol Nutr 2019; 69:461–465. 35. Weinstein-Nakar I, Focht G, Church P, et al. ImageKids study group. Associations among mucosal and transmural healing and fecal level of calprotectin in children with Crohn's disease. Clin Gastroenterol Hepatol 2018; 16:1089.e4–1097,e4. 36. Arkko A, Kaseva T, Salli E, et al. Automatic detection of Crohn's disease using quantified motility in magnetic resonance enterography: initial experiences. Clin Radiol 2022; 77:96–103. 37▪. Meng J, Luo Z, Chen Z, et al. Intestinal fibrosis classification in patients with Crohn's disease using CT enterography–based deep learning: comparisons with radiomics and radiologists. Eur Radiol 2022; 32:8692–8705. 38. Zhou Z, Xiong Z, Cheng R, et al. Volumetric visceral fat machine learning phenotype on CT for differential diagnosis of inflammatory bowel disease. Eur Radiol 2023; 33:1862–1872. 39. Guez I, Focht G, Greer MLC, et al. Development of a multimodal machine-learning fusion model to noninvasively assess ileal Crohn's disease endoscopic activity. Comput Methods Programs Biomed 2022; 227:107207. 40. Rimola J, Rodriguez S, García-Bosch O, et al. Magnetic resonance for assessment of disease activity and severity in ileocolonic Crohn's disease. Gut 2009; 58:1113–1120. 41. Marchal-Bressenot A, Salleron J, Boulagnon-Rombi C, et al. Development and validation of the Nancy histological index for UC. Gut 2017; 66:43–49. 42▪. Peyrin-Biroulet L, Rubin D, Weber C, et al. P777 Deployment of an artificial intelligence tool for precision medicine in ulcerative colitis: Preliminary data from 8 globally distributed clinical sites. J Crohns Colitis 2023; 17:i910–i911. 43. Magro F, Doherty G, Peyrin-Biroulet L, et al. ECCO position paper: Harmonization of the approach to ulcerative colitis histopathology. J Crohns Colitis 2020; 14:1503–1511. 44. Mosli MH, Feagan BG, Zou G, et al. Development and validation of a histological index for UC. Gut 2017; 66:50–58. 45. Geboes K, Riddell R, Öst A, et al. A reproducible grading scale for histological assessment of inflammation in ulcerative colitis. Gut 2000; 47:404–409. 46. Schroeder KW, Tremaine WJ, Ilstrup DM. Coated oral 5-aminosalicylic acid therapy for mildly to moderately active ulcerative colitis. New Engl J Med 1987; 317:1625–1629. 47▪▪. Gui X, Bazarova A, del Amor R, et al. PICaSSO Histologic Remission Index (PHRI) in ulcerative colitis: development of a novel simplified histological score for monitoring mucosal healing and predicting clinical outcomes and its applicability in an artificial intelligence system. Gut 2022; 889–898. 48. Villanacci V, Parigi TL, Del Amor R, et al. OP15 A new simplified histology artificial intelligence system for accurate assessment of remission in ulcerative colitis. J Crohn Colitis 2022; 16: (Suppl 1): i015–i17. 49▪. Kiyokawa H, Abe M, Matsui T, et al. Deep learning analysis of histologic images from intestinal specimen reveals adipocyte shrinkage and mast cell infiltration to predict postoperative Crohn disease. Am J Pathol 2022; 192:904–916. 50. Takishima K, Maeda Y, Ogata N, et al. Beyond complete endoscopic healing: Goblet appearance using an endocytoscope to predict future sustained clinical remission in ulcerative colitis. Dig Endosc 2022; 34:1030–1039. 51▪. Ohara J, Nemoto T, Maeda Y, et al. Deep learning-based automated quantification of goblet cell mucus using histological images as a predictor of clinical relapse of ulcerative colitis with endoscopic remission. J Gastroenterol 2022; 57:962–970. 52. Vande Casteele N, Leighton JA, Pasha SF, et al. Utilizing deep learning to analyze whole slide images of colonic biopsies for associations between eosinophil density and clinicopathologic features in active ulcerative colitis. Inflamm Bowel Dis 2022; 28:539–546. 53. Abràmoff MD, Cunningham B, Patel B, et al. Collaborative Community on Ophthalmic Imaging Executive Committee and Foundational Principles of Ophthalmic Imaging and Algorithmic Interpretation Working Group. Foundational considerations for artificial intelligence using ophthalmic images. Ophthalmology 2022; 129:e14–e32. 54. Uche-Anya E, Anyane-Yeboa A, Berzin TM, et al. Artificial intelligence in gastroenterology and hepatology: how to advance clinical practice while ensuring health equity. Gut 2022; 71:1909–1915.