aiqbench analysis
PhenoBench establishes a rigorous, auditable framework for evaluating tabular foundation models and language models on complex clinical data from the Human Phenotype Project. The study reveals that while these emerging models offer aggregate improvements over standard baselines, their gains in specific healthcare tasks are often marginal without cohort-specific fine-tuning. This work provides a critical baseline for assessing the practical utility of large AI models in multimodal medical settings.
PhenoBench: Mapping What a Deeply Phenotyped Human Cohort Can Tell Us
arXiv:2609.06080v1 Announce Type: new Abstract: Deeply phenotyped cohorts combine clinical, imaging, molecular, and wearable observations across timescales from seconds to years. This breadth can reveal which measurements inform which health-related questions, but heterogeneous analyses are not directly comparable. We present PhenoBench, an executable benchmark built around the Human Phenotype Project, in which more than 13,000 participants have completed the initial visit. Each question fixes the target, eligible population, timing, and allowed information; its evaluation contract specifies the split, metric, baseline, and claim boundary. The benchmark defines 90 clinically grounded tasks across 15 domains and 26 input modalities. Measurements showed question- and representation-dependent predictive value, including positive, near-zero, and negative changes in held-out performance relative to matched baselines. We used PhenoBench to evaluate emerging tabular foundation models across 160 matched regression comparisons spanning 52 tasks. These models ranked above standard task-specific models in aggregate but, averaged across the three pretrained models within each cell, improved on ridge by a median of only 0.004 $R^2$ (95% CI, 0.002--0.006). We then used the same cohort data and evaluation contracts to evaluate 14 language models, collectively covering 40 tasks spanning phenotype recovery, classification, follow-up forecasting, and participant ordering. Without cohort-specific fitting, language models made informative predictions on some tasks, but showed task-specific capability gaps, shared failures of scale, and rarely surpassed models fitted on the same fields. PhenoBench turns a multimodal longitudinal cohort into a versioned, auditable evaluation system where new questions, measurements, and models can be added without redefining existing comparisons.
Abstract:Deeply phenotyped cohorts combine clinical, imaging, molecular, and wearable observations across timescales from seconds to years. This breadth can reveal which measurements inform which health-related questions, but heterogeneous analyses are not directly comparable. We present PhenoBench, an executable benchmark built around the Human Phenotype Project, in which more than 13,000 participants have completed the initial visit. Each question fixes the target, eligible population, timing, and allowed information; its evaluation contract specifies the split, metric, baseline, and claim boundary. The benchmark defines 90 clinically grounded tasks across 15 domains and 26 input modalities. Measurements showed question- and representation-dependent predictive value, including positive, near-zero, and negative changes in held-out performance relative to matched baselines. We used PhenoBench to evaluate emerging tabular foundation models across 160 matched regression comparisons spanning 52 tasks. These models ranked above standard task-specific models in aggregate but, averaged across the three pretrained models within each cell, improved on ridge by a median of only 0.004 $R^2$ (95% CI, 0.002--0.006). We then used the same cohort data and evaluation contracts to evaluate 14 language models, collectively covering 40 tasks spanning phenotype recovery, classification, follow-up forecasting, and participant ordering. Without cohort-specific fitting, language models made informative predictions on some tasks, but showed task-specific capability gaps, shared failures of scale, and rarely surpassed models fitted on the same fields. PhenoBench turns a multimodal longitudinal cohort into a versioned, auditable evaluation system where new questions, measurements, and models can be added without redefining existing comparisons.
33 pages; 5 main figures and 5 supplementary figures. Project website: this https URL . Code and benchmark materials: this https URL
Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
arXiv:2609.06080 [cs.LG]
(or arXiv:2609.06080v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.06080 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 · 362 words
Related coverage
Podcast: Securing AI Agents: Identity, Authorization, and the DPACT Framework
InfoQ · Sep 21, 2026
Why the current tech backlash feels different
The Verge · Sep 10, 2026
Dyna-2: General Purpose Physical Agents with Hour-Long Autonomy — Jason Ma, Dyna Robotics
Latent Space · Oct 13, 2026
Skydio CEO Adam Bry on why Silicon Valley shouldn’t draw red lines for drone use
The Verge · Jun 15, 2026