Machine learning approach to dissect the clinical heterogeneity of IBD-associated fatigue

Extreme fatigue is a clinical symptom that affects >50% of individuals with Inflammatory Bowel Disease (IBD), with a similar prevalence across many common immune-mediated inflammatory diseases (IMIDs). Despite its ubiquity, human scientific studies have yet to explain the mechanistic basis of this pervasive and complex symptom. One fundamental reason for this, is our inability to account for the clinical heterogeneity of fatigue with its multifactorial nature. We present the conceptual machine learning (ML) framework to dissect the complex nature of fatigue using one of the largest prospectively captured, real-world patient-reported outcome (PROs) on wellbeing from three contemporaneous cohorts (2020-present), totalling 2,970 responses from 2,290 participants across the UK and internationally, including non-IBD controls with 100 lines of clinical metadata. We systematically defined the (1) threshold of fatigue as our primary outcome ([≥]10/14 fatigue days) to build our ML approach, (2) utilised routinely available clinical data that can be used at a population-level analysis, (3) employed seven different ML methods with external validation in 3 different cohorts in UK, Spain and Australia (n=252), (4) employed Shapley Additive Explanations (SHAP) analysis to break down the clinical heterogeneity to allow the examination of clinical predictive factors at an individual level; and finally (5), investigate whether there are distinct clusters of fatigue patients. We found that ML models performed comparably (AUC/C-index [~]0.7) on external validation with SHAP analysis showing interpretable, individualised fatigue drivers and five distinct fatigue phenotypes, including a subgroup of young males with significantly lower fatigue burden. Our data therefore provides the ML roadmap to predict and deconstruct fatigue in IBD and potentially also more widely in IMIDs, enabling patient-level dissection beyond symptom-based classification with the ability to integrate deep molecular data. This is a step towards future clinical-scientific AI models with the immediate clinical application to stratify patients to human experimental studies to better understand the dominant mechanisms that drive fatigue at an individual level.