Prevalence, incidence, and mortality of inflammatory bowel disease in the Netherlands: development and external validation of machine learning models - Report - MDSpire
Advertisement
Prevalence, incidence, and mortality of inflammatory bowel disease in the Netherlands: development and external validation of machine learning models
Machine Learning Models Reveal IBD Epidemiology Trends in the Netherlands
Overview
Machine learning models accurately identified inflammatory bowel disease (IBD) cases from administrative data, enabling precise estimation of IBD prevalence, incidence, and mortality in the Netherlands from 2013 to 2020. The prevalence was 577.6 per 100,000 in 2020, incidence remained stable at 20.1 per 100,000, and mortality rates among IBD patients increased over time, exceeding those of the general population.
Background
Inflammatory bowel disease (IBD) is a chronic condition with rising prevalence and incidence globally. In the Netherlands, previous data showed increasing incidence and prevalence, with expectations of doubling prevalence by 2030 due to high incidence and low mortality. Large nationwide registries offer opportunities to study IBD epidemiology, but accurate case identification is essential. Machine learning approaches can improve identification of IBD cases in administrative datasets, facilitating large-scale epidemiological studies and healthcare planning.
Data Highlights
Measure
Value (95% CI)
Year
IBD Prevalence per 100,000
577.6 (566.7; 586.2)
2020
IBD Incidence per 100,000
20.1 (18.0; 22.3)
2020
IBD Mortality per 1,000
11.6 (10.5; 11.8)
2020
General Population Mortality per 1,000
9.5
2020
Model AUC for IBD Identification
0.97 (0.96; 0.97)
Validation Cohort
Subtype Classification Accuracy
0.95 (0.94; 0.95)
Validation Cohort
Incidence Year Classification Accuracy
0.88 (0.86; 0.89)
Validation Cohort
Key Findings
Random forest model identified IBD cases with high accuracy (AUC 0.97).
Gradient-boosted trees model classified IBD subtypes with 95% accuracy.
IBD prevalence in the Netherlands was 577.6 per 100,000 in 2020, showing a slower increase than previously expected.
IBD incidence remained stable at approximately 20.1 per 100,000 in 2020.
Mortality rates among IBD patients increased over time, reaching 11.6 per 1,000 in 2020, higher than the general population mortality of 9.5 per 1,000.
Geographical variation in IBD prevalence was observed across the Netherlands.
Clinical Implications
Accurate identification of IBD cases using machine learning on administrative data enables robust epidemiological surveillance and resource planning. The slower-than-expected increase in prevalence suggests a transition towards prevalence equilibrium, impacting healthcare demand projections. Rising mortality rates in IBD patients highlight the need for ongoing monitoring and potentially tailored management strategies to address evolving patient risks.
Conclusion
Machine learning models effectively identify IBD cases and characterize epidemiological trends in the Netherlands, revealing stable incidence, increasing mortality, and a slower rise in prevalence than anticipated. These insights support informed healthcare planning and future research on IBD management.
References
Kaplan & Windsor 2021 -- The Epidemiology of Inflammatory Bowel Disease
IBD South Limburg Cohort Profile 2019 -- Population-Based IBD Research
by Reinier C A van Linschoten, Nikki van Leeuwen, David van Klaveren, Marieke J Pierik, Rob Creemers, Evelien M B Hendrix, Jan A Hazelzet, C Janneke van der Woude, Rachel L West, Desirée van Noord
