aiqbench analysis
This paper presents the winning entry for the MoCha 2026 Benchmark on Parkinsonian Gait, demonstrating that significant performance gains come from subject-level posterior aggregation and transductive calibration rather than fine-tuning frozen encoders. By averaging per-walk posteriors and calibrating decision operating points without labels, the system achieved a macro-F1 of 0.6945, outperforming both the runner-up and the organizers' baseline on unseen clinical sites. The study highlights that rigorous cross-validation strategies and careful handling of data grouping are critical bottlenecks in medical AI benchmarks, offering a clear methodological roadmap for improving generalization in domain-specific health applications.
Aggregate, Don't Adapt: Subject-Level Posterior Aggregation and Transductive Calibration for Cross-Site Parkinsonian Gait Severity
arXiv:2608.20587v1 Announce Type: cross Abstract: We describe the winning entry to the MoCha 2026 Benchmark and Challenge on Parkinsonian Gait, which predicts MDS-UPDRS gait severity from canonicalized SMPL motion recorded at clinical sites unseen during training. The system reaches 0.6945 macro-F1 on the hidden test and ranked first of 58 entries, ahead of the runner-up at 0.5807 and the organizers' baseline at 0.4289, on a frozen public motion encoder with a single $4\times512$ linear layer. Nearly all of the margin comes from three stages usually treated as bookkeeping: reproducing the reference benchmark's exact head recipe, averaging per-walk posteriors within the subject grouping the organizers ship, and a label-free transductive calibration of the feature mean and the decision operating point. Fine-tuning the encoder lost in four distinct forms, and ten alternative encoders were worse. Every ablation number is a paid read on the hidden test, because our own leave-two-cohort-out cross-validation proved anti-correlated with the deciding score over eleven configurations. We give the negative record in full, and identify our largest gain, subject-level aggregation, as the binding ceiling on this benchmark.
Abstract:We describe the winning entry to the MoCha 2026 Benchmark and Challenge on Parkinsonian Gait, which predicts MDS-UPDRS gait severity from canonicalized SMPL motion recorded at clinical sites unseen during training. The system reaches 0.6945 macro-F1 on the hidden test and ranked first of 58 entries, ahead of the runner-up at 0.5807 and the organizers' baseline at 0.4289, on a frozen public motion encoder with a single $4\times512$ linear layer. Nearly all of the margin comes from three stages usually treated as bookkeeping: reproducing the reference benchmark's exact head recipe, averaging per-walk posteriors within the subject grouping the organizers ship, and a label-free transductive calibration of the feature mean and the decision operating point. Fine-tuning the encoder lost in four distinct forms, and ten alternative encoders were worse. Every ablation number is a paid read on the hidden test, because our own leave-two-cohort-out cross-validation proved anti-correlated with the deciding score over eleven configurations. We give the negative record in full, and identify our largest gain, subject-level aggregation, as the binding ceiling on this benchmark.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2608.20587 [cs.CV]
(or arXiv:2608.20587v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2608.20587 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 274 words
Related coverage
RACER: Role-Aligned Competence Estimation for Human-AI Routing
arXiv · Sep 21, 2026
State of the Field: AI for Epistemics and Coordination
LessWrong · Oct 2, 2026
From Migration to Calibration: Preserving Agent Capabilities across Models, Jurisdictions, and Scale
arXiv · Sep 29, 2026
The J-Space Debate, Agent Swarms, and Pacing Frontier AI - Digital Minds Newsletter #4
LessWrong · Sep 18, 2026