Biological drifts within normal ranges allow the detection of Crohn's disease patients at high risk of rehospitalization
Homo, A.; Rolland, J.; Bezier, C.; Boutin, R.; Equinet, L.; Maes, N.; Thys, M.; Monin, L.; Louis, E.
Show abstract
Background: Crohn's disease is a chronic relapsing inflammatory bowel disease with an unpredictable clinical course that may lead to recurrent hospitalizations and surgery, making early identification of patients at risk a key challenge in longitudinal monitoring. Objective: To evaluate the prognostic value of blood biomarkers for anticipating hospitalizations in patients with Crohn's disease by moving beyond exclusive reliance on conventional reference intervals toward the analysis of personalized biological drift. The underlying premise is that fluctuations that remain within standard reference ranges (and are therefore invisible to conventional thresholds) may still carry a risk signal when interpreted relative to an individual's optimal baseline. Design: We conducted a retrospective study of 993 patients with Crohn's disease followed at the University Hospital of Liege between 2005 and 2023. Longitudinal laboratory measurements were linked to Crohn's disease-related hospitalizations. Biomarkers were transformed into z-scores relative to optimized and personalized reference populations and classified into drift categories. Time to first hospitalization was analyzed using the Kaplan-Meier method, and recurrent hospitalizations were modeled using Cox models. Results: Hospitalization-free survival differed significantly across drift categories, including for deviations within conventional reference ranges (e.g., albumin, global log-rank p<0.0001). Among 57 biomarkers screened, 32 were significant in the global log-rank analysis, including 5 that were significant for intra-reference drift classes: low lymphocytes (%), low monocytes (%), low albumin, high potassium, and low aspartate aminotransferase. Conclusion: Personalized biomarker drift detects clinically meaningful risk signals that are missed by conventional reference-interval thresholds and may enable earlier risk stratification in Crohn's disease.
Matching journals
The top 10 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Developing And Validating COVID-19 Adverse Outcome Risk Prediction Models From A Bi-National European Cohort Of 5594 Patients 92%
- Identifying organ dysfunction trajectory-based subphenotypes in critically ill patients with COVID-19 92%
- Predicting bloodstream infection outcome using machine learning 91%
Similar papers in this journal
Similar papers in this journal
- Daily variation in blood glucose levels during continuous enteral nutrition in patients on the Intensive Care Unit: a retrospective observational study 92%
- Sedentary lifestyle, Physical Activity, and Gastrointestinal Diseases: Evidence from Mendelian Randomization Analysis 91%
- Precision Diagnostic Approach to Predict 5-year Risk for Microvascular Complications in Type 1 Diabetes 91%
Similar papers in this journal
- Gut microbiota alterations in patients with persistent respiratory dysfunction three months after severe COVID-19 93%
- Children with familial hypercholesterolemia display changes in LDL and HDL function: a cross-sectional study 88%
- BMI is positively associated with accelerated epigenetic aging in twin pairs discordant for BMI 87%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.