Abstract
Quantitative MRI (qMRI) biomarkers summarize knee joint health, but their value for forecasting progression, and for simulating interventions over time, is not established. Prior work derived a 110-dimensional principal-component (PC) space of knee qMRI biomarkers and reported cross-sectional associations with osteoarthritis (OA) outcomes. We extend that static space into a longitudinal, counterfactual digital twin. It is a recurrent model that rolls the biomarker state forward, predicts event hazard, and simulates the model-implied consequence of altering individual biomarkers or a modifiable clinical factor (body-mass index, BMI). On the publicly available OAI dataset we study two outcomes. For incident OA in initially healthy knees, imaging carries modest prognostic signal (C-index 0.55) that clinical demographics alone do not (0.51), and model-based perturbation responses are directionally consistent with associations from a prior analysis of an overlapping OAI cohort, including a small, correctly signed BMI response in pooled repeated cross-fitting. For total knee replacement (TKR), discrimination appears far higher (0.77), but a stratified analysis shows it is driven by baseline radiographic severity. Within severity strata the imaging signal vanishes. The twin therefore distinguishes genuine biomarker prognosis from a severity-detection confound. This property is essential for trustworthy counterfactual simulation. All results use hardened, leakage-controlled evaluation.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/RHEUM_AI_011.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: https://papers.miccai.org/miccai-2026-sat/supp/RHEUM_AI_011_supp.pdf
Link to Open Review
Open Review Page: https://openreview.net/forum?id=rOJ3gobt6n
BibTex
@InProceedings{HoyGab_Interpretable_MICCAISAT2026,
author = { Hoyer, Gabrielle AND Majumdar, Sharmila},
title = { { Interpretable Counterfactual Simulation of Knee Osteoarthritis Progression from qMRI Biomarkers } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17271},
month = {pending},
page = {pending}
}