Prediction of life-threatening complications of inflammatory bowel disease using neural networks: a practical tool for health care professionals

Abstract

Introduction. Inflammatory Bowel Disease (IBD) including Crohn’s Disease (CD) and Ulcerative Colitis (UC) representsis a challenge for gastroenterologists worldwide, due to its potential to cause life-threatening complications and lead to disability in patients. Aim: to develop a tool that can be used in clinical practice to predict the development of life-threatening complications of IBD through mathematical modeling. Methods. A historical cohort of 291 adult patients with a verified diagnosis of IBD (48% - CD, 52% - UC) who sought medical care from 2020 to 2022 comprised the study base. The outcomes were life-threatening complications including a subgroup of conditions that required urgent surgical intervention. Logistic regression, classification trees and neural network analysis were used to predict the studied outcomes. Results. Life-threatening complications occurred in 22.3% of CD- and in 9.9% of UC patients. The corresponding numbers for urgent surgical complications were 16.5% and 1.3%. Among the constructed mathematical models for both types of outcomes, neural network models demonstrated the highest sensitivity and specificity. Based on the neural network models, two software products named “IBD prognosis: risk of life-threatening complications” and “ IBD prognosis: risk of urgent surgical complications” were developed. For the former, the positive predictive value was 65.0% (95% CI 52.4-75.8) while the negative predictive value was 97.0% (95% CI: 93.9-98.5). For the latter, the corresponding numbers were 77.4% (95% CI: 60.2-87.4) and 99.2% (95% CI: 97.2-99.8). Conclusions. Two tools have been developed for use in clinical practice by gastroenterologists, therapists, and general practitioners to manage IBD patients. Identifying a high-risk IBD patient for developing a life-threatening complication can be used as a foundation for optimizing the therapy used in the treatment of a given patient, potentially saving lives.