Abstract

1 Maccabi Institute for Research & Innovation, Maccabi Healthcare Services, Tel Aviv, Israel.

2 Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel.

3 Pediatric Gastroenterology Unit, Dana-Dwek Children’s Hospital, Tel-Aviv Sourasky Medical Center, Tel  Aviv, Israel.

4 Predicta Med, LTD.

5 Gastroenterology Department, Sheba Medical Center, Tel Hashomer, Israel.

Although 3,000,000 Americans are afflicted with Inflammatory Bowel Disease (IBD)  encompassing Crohn’s Disease (CD) and Ulcerative Colitis (UC)1, timely diagnosis is a  challenge due to non-specific and overlapping symptoms.2 More than 20% of patients may be  initially misdiagnosed3causing delayed diagnosis and treatment, potentially leading to increased risk of complications2and irreversible mucosal damage.4 Artificial intelligence (AI)  models can alert physicians to patients who would otherwise be misdiagnosed, potentially  improving patient outcomes and reducing costs. We aimed to develop PredictAI, a  proprietary AI Gradient Boosted Decision Tree based machine learning algorithm and test its  accuracy in identifying undiagnosed CD and UC in the primary care setting.

This was a retrospective study of 2,471,267 patients’ electronic medical records (EMR) from  Maccabi Healthcare Services in Israel. Sufficient data was available between the years 2010- 2020, of which 2 consecutive years (2015-2016) were pre-assigned to the test set. Inclusion criteria were: (i) CD or UC ICD code as defined by Maccabi’s Registry,5(ii) no other  autoimmune disease diagnosis, (iii) at least 4 years of data antedating first suspicion by primary care physician (PCP) of IBD were available. First suspicion was defined as any  diagnostic test, procedure, or referral to a specialist, indicating suspicion of IBD. Here we  included adult data only.

Of 2,471,267 patients, 1,214 had a first-time diagnosis of IBD and available antedating data in the years 2015-2016. Of these, PredictAI identified 126, 120, 104, 93 patients 1, 2, 3 and 4  years prior to initial PCP suspicion, respectively. For CD, it predicted 83/229 (36%) of  patients 1 year prior to PCP initial suspicion of disease, 70/211 (33%) patients 2 years prior,

60/179 (33%) patients 3 years prior and 56/148 (38%) patients 4 years prior. Discriminatory  accuracy area under the curve (AUC) was 76%, 75%, 75% and 78%, 1, 2, 3 and 4 years  before initial PCP suspicion, respectively (figure 1). Corresponding early-identification ratio  and discriminatory accuracy for UC was 42/238 (17%, AUC=71%), 50/225 (22%,  AUC=70%), 44/187 (23%, AUC=72%) 37/161 (23%, AUC=74%), for 1,2,3 and 4 years  before initial PCP suspicion of UC, respectively (figure 2). Specificity for CD and UC each  was above 90%.

PredictAI accurately identified CD and UC diagnosis in 17-38% of patients presenting to  primary care up to 4 years prior to PCP’s initial suspicion, potentially reducing time to  diagnosis.

1. Dahlhamer JM, Zammitti EP, Ward BW, Wheaton AG, Croft JB. Prevalence of Inflammatory Bowel Disease  Among Adults Aged ≥18 Years – United States, 2015. MMWR Morb Mortal Wkly Rep. 2016 Oct 28;65(42):1166- 1169. doi: 10.15585/mmwr.mm6542a3. PMID: 27787492.
2. Feuerstein JD, Cheifetz AS. Crohn Disease: Epidemiology, Diagnosis, and Management. Mayo Clin Proc. 2017 Jul;92(7):1088-1103. doi: 10.1016/j.mayocp.2017.04.010. Epub 2017 Jun 7. PMID: 28601423.
3. Wylezinski LS, Gray JD, Polk JB, Harmata AJ, Spurlock CF 3rd. Illuminating an Invisible Epidemic: A systemic  Review of the Clinical and Economic Benefits of Early Diagnosis and Treatment in Inflammatory Disease and  Related Syndromes. J Clin Med. 2019;8(4):493. Published 2019 Apr 11. doi:10.3390/jcm8040493
4. Zane Gallinger, Ryan Ungaro, Jean-Frederic Colombel, Robert S Sandler, Wenli Chen, Delayed Diagnosis of  Crohn’s Disease is Common and Associated with Increased Risk of Disease Complications. Inflammatory Bowel  Diseases, Volume 25, Issue Supplement_1, February 2019, Pages S14–S15.
5. Foundation of a Registry for Inflammatory Bowel Disease at Maccabi Healthcare Services- Connecting Hospitals  with Community Medicine; https://cdn.doctorsonly.co.il/2018/10/13_revital-krib.pdf