Skip to content

Author

Gilles Boire

2 papers indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Aug 2026

Complement Pathway Heterogeneity in Rheumatoid Arthritis Uncovered Through Longitudinal Serum Proteomics

Rheumatoid arthritis (RA) is a heterogeneous autoimmune disease. Despite established serological classification and advances in therapeutic strategies, around 40% of patients exhibit inadequate responses to first-line DMARDs. Current clinical tools remain limited in their ability to predict treatment response and risk of joint erosion, posing a significant challenge to precision medicine. There is a critical need for novel biomarkers that can reliably forecast disease outcomes and guide therapeutic decisions. Integrating protein signatures with clinical features may enhance risk stratification in rheumatic diseases.[1] In this study, we employed data-independent acquisition mass spectrometry (DIA-MS) to perform a comprehensive longitudinal serum proteomic analysis in RA patients, aiming to identify biomarkers predictive of disease activity and progression. This study was conducted on serum samples collected from 107 DMARD- and steroid-free RA patients (48 seronegative, 59 seropositive) enrolled in the EUPA cohort ( NCT00512239 ), between 2005 and 2019, at the CIUSSS de l’Estrie-CHUS, Québec, Canada.[2] Serial sera collected at baseline and at the 12-month follow-up visit were analyzed by DIA-MS. MS data were analyzed using DIA-NN software for peptides/proteins identification and quantification. Statistical analyses were performed using R: Differentially abundant proteins (DAPs) across clusters were identified by Mann-Whitney U test with Bonferroni correction. KEGG enrichment analyses were conducted using the PathfindR R package, with BH correction. Generalized estimating equations (GEE) models were adjusted for age, sex, serology, and symptom duration, with FDR correction using Storey’s q-value. Proteomic profiling quantified 869 serum proteins, of which 368 passed quality control and filtering for downstream statistical analyses. 1-Principal component analysis (PCA) followed by hierarchical clustering of baseline (Figure 1A), and 12-month proteomic data suggested patients could be grouped into 2 clusters. 2-KEGG pathway analysis of DAPs between patient clusters showed significant enrichment of complement and coagulation cascades pathway. 3-These clusters exhibited distinct serum levels of complement-related proteins (Figure 1B). 4-C68omparison of patient trajectories between baseline and 12-month serum-based clusters (Figure 1C), revealed that transitions between clusters were accompanied by significant modulation of complement-related protein levels (Figure 1D). 5-GEE models identified 29 proteins (q-value ≤0.05) associated with binomial 12-month disease activity, defined as DAS28-CRP ≤2.6 or ≥3.2, which were enriched for KEGG complement and coagulation cascades pathway. Figure 1. A) Hierarchical Clustering on PCA was performed on RA patients’ baseline serum proteome. B) Baseline serum complement pathway (KEGG hsa04610) protein levels were normalized by z-scores. Average z-scores per patient are shown as a color gradient over PCA plot. C) Sankey plot of patient trajectories between baseline and 12-month proteome-based clusters. D) Average serum complement proteins z-scores of patients grouped by cluster trajectories BH-adjusted p ≤ 0.05 = * and ≤ 0.001 = ***. This study highlights the heterogeneity of circulating complement components in early RA patients, suggesting that complement protein profiles may reflect underlying disease mechanisms beyond general inflammation and could inform future biomarker-driven approaches to RA management. [1.] Carrasco-Zanini J. Nat Med 2024;30:2489-98. [2.] Carrier N. J Rheumatol 2024;52:119-27.

Benoît Marchand, N. Carrier, Elizabeth Beaulieu et al. · 0 citations
Preprint Aug 2026

DoctorAgents: an agentic framework to iteratively refine AutoML pipeline for small clinical temporal data

Clinical machine learning (ML) has the potential to support high-stakes medical decision-making, but reliable deployment is often constrained by scarce, heterogeneous, and temporal complexity. Developing effective ML pipelines for such data remains time-consuming and error-prone, while existing automated machine learning (AutoML) systems only partially address this challenge because they largely rely on brute-force search over predefined spaces and lack explicit reasoning and memory. We therefore reformulate AutoML for small clinical data from exhaustive search to reasoning-driven refinement. We propose DoctorAgents, an agentic AI framework that autonomously constructs and optimizes end-to-end ML pipelines through specialized large language model (LLM) agents for generation, validation, and refinement. DoctorAgents backpropagates natural-language feedback through textual gradient descent to perform targeted updates without exhaustive search. Experiments across diverse clinical tasks show that DoctorAgents consistently outperforms established AutoML baselines while producing more interpretable task-specific representations.

Ruilin Wang, Bozhong Wang, Elizabeth Kourbatski et al. · 1 citation