Interpretable machine learning applied to high-dimensional salivary proteomics accurately classifies pediatric inflammatory bowel diseases

Product:

Inflammation Panel 250

Topic:

Inflammation

Disease Area:

IBD

Sample Type:

Saliva

Topic/Product:

Inflammation Panel 250

Disease Area:

IBD

Sample Type:

Saliva

Abstract

Background and aims Inflammatory bowel diseases (IBD), including Crohn’s disease (CD), ulcerative colitis (UC), and IBD-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 study reports the first application of high-dimensional salivary proteomics integrated with interpretable artificial intelligence/machine learning (AI/ML) to define a minimal protein signature for pediatric IBD classification with the goal of informing therapeutic decision-making.

Methods Unstimulated saliva from pediatric CD, UC, and IBD-U patients was analyzed using Alamar Biosciences’ NULISAseq Inflammation Panel 250 (250 proteins). Logistic regression with recursive feature elimination identified a minimal discriminative signature. Performance was tested in independent follow-up samples. SHapley Additive exPlanations (SHAP) 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 model achieved 96.2% overall accuracy in first-visit samples and 86.4% overall accuracy in follow-up testing. SHAP revealed patient-specific drivers and suggested biological alignment of IBD-U cases toward CD-like or UC-like profiles.

Conclusions This first-in-field integration of salivary proteomics with interpretable AI/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.

Authors & Affiliations

Brittany T. Rupp 1, Joaquin Reyna 2, Ally Giunta 3,4, Theresa Weaver 1, Kelly Chason 4, Jinze Liu 5,6, Ajay S. Gulati 3,7,8, Kevin M. Byrd 1,6,7

Affiliations:
1. Department of Oral and Craniofacial Molecular Biology, Virginia Commonwealth University,
Richmond VA., USA.
2. Department of Bioinformatics, Virginia Commonwealth University, Richmond VA, USA.
3. Department of Pediatrics, Division of Gastroenterology, University of North Carolina at Chapel
Hill, Chapel Hill, NC, USA.
4. Children’s Research Institute, School of Medicine, University of North Carolina at Chapel Hill,
Chapel Hill, NC, USA.
5. Department of Biostatistics, Virginia Commonwealth University, Richmond, VA, USA.
6. Massey Comprehensive Cancer Center, Virginia Commonwealth University, Richmond, VA,
USA.
7. Center for Gastrointestinal Biology and Disease, University of North Carolina at Chapel Hill,
Chapel Hill, NC, USA.
8. Department of Pathology and Laboratory Medicine, University of North Carolina at Chapel Hill,
Chapel Hill, NC, USA.

Steve Williams, MD, PhD

CSO

Dr. Willams serves as the company’s Chief Scientific Officer. He was previously Chief Medical Officer at Standard Biotools and at SomaLogic where he pioneered the discipline for discovery and validation of predictive, diagnostic and prognostic models using machine-learning applied to large-plex proteomics. 20 such tests were used for drug characterization, safety and efficacy when incorporated in clinical drug trials at Pharma/Biotech and 17 different multivariate tests were validated and translated into regulated healthcare uses. Prior to SomaLogic, Dr. Williams was at Pfizer in the UK and the USA as a clinical triallist in Translational Medicine, and subsequently as VP, Global Clinical Technology. He sat on the National Advisory Council for the National Institute of Biomedical Imaging and Bioengineering, the Executive Committee for the FNIH Biomarkers Consortium, and worked with FDA and PhRMA on developing evidentiary standards for biomarker qualification. Dr. William’s medical training was in London, at Charing Cross and Westminster Medical School, followed by a PhD in medicine/physiology at the same institution and training in Radiology at the University of Newcastle Upon Tyne. Steve is co-inventor on 26 proteomics patents and author/coauthor on multiple foundational proteomics manuscripts.

Justin McAnear

CFO

Mr. McAnear serves as the company’s Chief Financial Officer. He brings over 25 years of operational and financial leadership experience across various sectors and was instrumental in taking 10x Genomics public in 2019, serving as its CFO for over five years. Mr. McAnear served for over 3 years as Tesla’s VP of Worldwide Finance and Operations, supporting landmark initiatives such as the Model X and Model 3 launches and Solar City acquisition. He also held various roles at Apple and J&J earlier in his career and served as a naval officer and aviator for over 9 years.