Interpretable machine learning applied to high-dimensional salivary proteomics in pediatric inflammatory bowel diseases
Related Experiment Videos
Brittany T Rupp1, Joaquin Reyna2, Ally Giunta3,4
1Department of Oral and Craniofacial Molecular Biology, Virginia Commonwealth University, Richmond, VA, United States.
Background:
Inflammatory bowel diseases (IBDs), including Crohn's disease (CD), ulcerative colitis (UC), and inflammatory bowel disease unclassified (IBD-U), are chronic inflammatory disorders of the gastrointestinal tract. Current methods for classification and longitudinal monitoring are invasive, expensive, and often delayed, limiting timely diagnosis and management. This proof-of-concept study reports the first application of high-dimensional salivary proteomics integrated with interpretable artificial intelligence/machine learning (ML) to define a minimal protein signature for pediatric IBD classification with the ultimate goal of informing therapeutic decision-making.
Methods:
Unstimulated saliva from pediatric CD, UC, and IBD-U patients was analyzed using Alamar Biosciences' NULISAseq 250 Protein Inflammation Panel. Logistic regression with recursive feature elimination identified a candidate minimal discriminative signature. Performance was further assessed using follow-up samples. SHAP (SHapley Additive exPlanations) quantified patient-specific protein contributions and assessed biological similarity of IBD-U to CD and UC.
Results:
Differential abundance analysis between UC and CD revealed 53 significantly different proteins. ML identified a 14-protein signature comprising chemokines/cytokines (CCL1, IFNA1-IFNA13, IL12p70, IL34, TNFSF11/RANKL), receptors/ligands (CD40LG, ICOSLG, IL1R2, IL17RA), structural/tissue-remodeling proteins (CD93, GFAP, SPP1), and growth factors/immune modulators (GDF2, GZMA). The proposed model achieved 97.5% balanced accuracy in first-visit pilot samples and 80.2% with matched follow-up samples. SHAP revealed patient-specific drivers and suggested biological alignment of IBD-U cases toward CD- or UC-like profiles.
Conclusions:
This first-in-field integration of salivary proteomics with interpretable artificial intelligence/ML demonstrates that accurate, noninvasive classification of pediatric IBD is possible using minimal biomarker sets. This approach establishes a scalable framework for future longitudinal monitoring and supports earlier and more precise therapeutic interventions.