Towards a General Intelligence and Interface for Wearable Health Data
Girish Narayanswamy, Maxwell A. Xu, A. Ali Heydari, Samy Abdel-Ghaffar, Marius Guerard, Kara Vaillancourt, Zhihan Zhang, Jake Garrison
Abstract
While ubiquitous wearable sensors capture a wealth of behavioral and physiological information, effectively transforming these signals into personalized health insights is challenging. Specifically, converting low-level sensor data into representations capable of characterizing higher-level states is difficult due to high phenotypic diversity and variation in individual baseline health, physiology, and lifestyle factors. Moreover, collecting wearable data paired with health outcome annotations is laborious and expensive, and retrospective annotation remains practically unfeasible, contributing to a scarcity of data with high-quality labels. To overcome these limitations, we propose a foundation model for wearable health that is pretrained on more than one trillion minutes of unlabeled sensor signals drawn from a large cohort of five million participants. We demonstrate that the joint scaling of model capacity and pretraining data volume leads to systematic improvements in performance, as evaluated on a diverse set of 35 health prediction tasks, spanning cardiovascular, metabolic, sleep, and mental health, as well as lifestyle choices and demographic factors. We find that this population scale representation unlocks label-efficient few-shot learning and generative capabilities for robust daily metric estimation. To further leverage this learned representation, we deploy a classroom of LLM agents to autonomously search the space of downstream predictive heads built on the model embeddings, showing broad performance improvements that increase with LLM model capacity. Finally, we show how integrating these downstream predictors into a Personal Health Agent can support model responses that are more relevant, contextually aware, and safe, and we validate this via 1,860 ratings from a cohort of clinicians.
AI Impact Assessments
(1 models)Scientific Impact Assessment: "Towards a General Intelligence and Interface for Wearable Health Data"
1. Core Contribution
This paper introduces SensorFM, a foundation model for wearable health data pretrained on over one trillion minutes of multimodal sensor data from 5 million participants. The core contributions are threefold: (1) establishing scaling laws for wearable sensor foundation models showing that joint scaling of model capacity and data volume yields predictable downstream improvements; (2) demonstrating that the learned representations generalize across 35 diverse health prediction tasks spanning cardiovascular, metabolic, mental health, sleep, lifestyle, and demographic domains; and (3) showing that integrating SensorFM predictions into a Personal Health Agent yields clinician-rated improvements in response quality comparable to using ground-truth clinical measurements.
The paper also introduces an LLM-driven "classroom" framework for automated downstream head design, conducting over 30,000 experiments to adapt embeddings to specific tasks, and demonstrates generative capabilities for missing data imputation that improve daily health metric estimation.
2. Methodological Rigor
Strengths in rigor:
Methodological concerns:
3. Potential Impact
Direct impact on digital health: This work could fundamentally shift wearable health analytics from bespoke, task-specific pipelines to a general-purpose embedding interface. The demonstration that simple linear probes on frozen embeddings outperform engineered feature baselines on 34/35 tasks is practically significant, dramatically lowering the barrier to developing new health applications.
Clinical screening and risk stratification: The ability to predict cardiovascular risk scores, metabolic markers, and mental health screening scores from passively collected sensor data could enable population-level screening. The generative infilling capability (99.7% step count accuracy with 60 minutes of missing data) addresses a real-world limitation of intermittent sensor wear.
AI agent integration: The demonstration that SensorFM predictions are statistically non-inferior to ground-truth clinical labels when used by a health agent is particularly impactful for consumer health applications. This validates a pathway from raw sensor data to personalized health coaching.
Broader ML impact: The agentic classroom framework for automated model adaptation, while not the primary contribution, introduces a practical methodology for efficiently adapting foundation model embeddings across many downstream tasks simultaneously.
4. Timeliness & Relevance
This paper arrives at a critical intersection of three trends: the maturation of foundation models, the explosive growth of consumer wearables (~500M+ devices globally), and increasing use of LLMs for health queries. The scarcity of labeled health data paired with wearable signals has been a persistent bottleneck; self-supervised pretraining at this scale directly addresses this limitation. The integration with LLM-based health agents anticipates the near-term convergence of sensing and conversational AI in consumer health products.
5. Strengths & Limitations
Key Strengths:
Notable Limitations:
Overall Assessment: This is a landmark paper in wearable health AI that establishes scaling laws, demonstrates broad generalization, and validates end-to-end clinical utility. While the proprietary data and single-ecosystem evaluation limit reproducibility and generalizability claims, the scale of evidence and breadth of evaluation set a new standard for the field. The work is likely to accelerate both academic research and commercial development in digital health foundation models.
Generated May 22, 2026
Comparison History (57)
Paper 2 presents a foundation model for wearable health pretrained on an unprecedented scale (1 trillion minutes, 5 million participants), demonstrating broad applicability across 35 health prediction tasks with concrete clinical validation. Its real-world impact potential is enormous given the ubiquity of wearable devices and growing demand for personalized health. While Paper 1 makes important theoretical contributions to AI alignment (impossibility theorem for ELK), its impact is more narrowly concentrated in the AI safety theory community. Paper 2's combination of scale, methodological breadth, practical applicability, and clinical validation suggests broader and more immediate scientific impact.
Paper 2 has higher likely scientific impact due to its massive-scale unlabeled pretraining (trillion-minute, 5M-person cohort), broad evaluation across 35 clinically relevant tasks, and clear real-world deployment path (Personal Health Agent with clinician-rated validation). Its contributions span representation learning, health prediction, and human-AI interfaces, making cross-field impact plausible. Paper 1 is novel methodologically (label-free consistency RL for spatial reasoning) but is narrower in application scope and likely to influence a smaller set of downstream domains compared with population-scale wearable health modeling.
Paper 2 presents a foundation model for wearable health pretrained on an unprecedented scale (1 trillion+ minutes from 5 million participants), addressing a fundamental challenge in digital health. Its breadth of impact spans 35 health prediction tasks across multiple medical domains, with clinical validation. The combination of foundation model scaling laws for wearable data, few-shot health prediction, and a clinician-validated Personal Health Agent represents a potentially transformative contribution with broad real-world healthcare applications. Paper 1, while novel in formulating unsupervised AI monitoring, addresses a narrower problem with more limited downstream impact.
Paper 2 presents a massive-scale foundation model trained on unprecedented wearable data (5 million participants), demonstrating broad applicability across 35 diverse health tasks. Its integration with LLMs and clinical validation strongly indicate immense real-world impact in digital health, personalized medicine, and wearable tech, overshadowing Paper 1's more narrowly focused diagnostic framework for LLM safety.
Paper 2 presents a foundation model pretrained on data from 5 million participants (over 1 trillion minutes of sensor data), addressing fundamental challenges in wearable health—a rapidly growing field with enormous real-world clinical applications. Its contributions span model pretraining at unprecedented scale, few-shot learning, agentic downstream optimization, and clinician-validated personal health agents, with broad impact across healthcare, ML, and wearable technology. While Paper 1 is a significant and timely contribution to AI-assisted peer review, its impact is more narrowly scoped to the academic review process itself, whereas Paper 2 has transformative potential for personalized medicine and population health.
Paper 2 presents a foundation model for wearable health pretrained on data from 5 million participants (1 trillion+ minutes), representing a massive scale-up in health AI. It addresses a fundamental challenge in digital health with broad applications across cardiovascular, metabolic, sleep, and mental health domains. The integration with LLM agents for personalized health creates a novel paradigm with significant real-world clinical impact, validated by clinician ratings. Its breadth of impact across healthcare, AI, and wearable technology exceeds Paper 1's more incremental (though solid) contribution to LLM inference optimization.
Paper 2 likely has higher near-term scientific impact due to its demonstrated scale (5M participants, >1T minutes), extensive empirical validation across 35 tasks, and concrete methodological contributions (scaling laws, few-shot transfer, generative estimation, LLM-agent search over predictive heads, clinician-rated Personal Health Agent). It targets an immediate, high-volume real-world data stream with clear translational pathways. Paper 1 is highly novel and potentially transformative but is primarily a conceptual framework with infrastructure/governance discussion and less methodological/experimental evidence, making its impact more speculative and longer-horizon.
Paper 1 likely has higher impact: it introduces a massive-scale wearable-health foundation model (trillion+ minutes; 5M participants) with demonstrated gains across 35 clinically relevant tasks, few-shot/label-efficient learning, generative estimation, and clinician-rated deployment as a Personal Health Agent—strong real-world translational potential and broad healthcare applicability. Paper 2 is a valuable, timely benchmark for interactive spatial reasoning with solid evaluation design, but its primary contribution is measurement infrastructure; impact depends on community adoption. Overall, Paper 1’s scale, clinical relevance, and immediate application potential suggest larger scientific and societal impact.
Paper 2 demonstrates unprecedented scale, utilizing over one trillion minutes of data from five million participants to train a foundation model for wearable health. Its application to 35 diverse health prediction tasks—ranging from cardiovascular to mental health—and its validation by clinicians highlight massive real-world impact potential in personalized medicine and public health. While Paper 1 introduces a clever algorithmic improvement for LLM agents, Paper 2's massive dataset, broad physiological applicability, and direct implications for human well-being give it a significantly higher potential for transformative scientific and societal impact.
Paper 2 presents a foundation model for wearable health pretrained on an unprecedented scale (1 trillion minutes, 5 million participants), addressing a fundamental challenge in digital health. Its breadth of impact spans 35 health prediction tasks across multiple domains, demonstrates few-shot learning capabilities, and integrates with a clinician-validated Personal Health Agent. The combination of massive-scale pretraining, clinical validation, and practical deployment potential gives it broader real-world impact across healthcare, AI, and wearable technology. Paper 1, while innovative in co-evolving RL training harnesses, addresses a narrower ML methodology problem.