我们推出 SensorFM,这是一个面向可穿戴健康领域的基础模型,基于来自五百万人的超过一万亿分钟传感器数据进行预训练。通过协同扩展模型规模与数据量,SensorFM 学习到一种通用的人体生理表征,该表征可迁移至 35 项健康预测任务,支持标签高效适配与数据填充,并能作为个人健康智能体的基础工具。
快速链接
据估算,目前已有数十亿台可穿戴设备在使用中,它们能够精确追踪心率、运动、皮肤温度、血氧水平以及睡眠情况,时间跨度从数天、数周到数月不等。这种持续、纵向的生理与行为数据流,为预防性、个性化健康提供了最有前景的原材料之一。然而,将这些底层信号转化为有意义的洞察仍然困难重重。首先,每个人的基础生理状况、生活方式和健康水平差异巨大,因此对某个人预示风险的信号模式,对另一个人可能毫无意义。其次,训练模型所需的标签——确诊诊断、实验室结果、经过验证的问卷——成本高昂、收集缓慢,且基本无法回溯获取。因此,大多数可穿戴健康模型都是逐个针对单一结果构建的,采用定制化的监督学习流程,目标范围狭窄,难以在人类健康的广泛领域中进行泛化。
在《迈向可穿戴健康数据的通用智能与接口》一文中,我们采取了不同的方法。我们推出了 SensorFM,这是一个大型传感器基础模型,能够直接从大规模人群的无标签可穿戴数据中学习。SensorFM 基于来自五百万名同意参与的受试者、超过一万亿分钟的多模态传感器信号进行预训练,学习到一种单一、可复用的感知人体生理表征——该表征可迁移至心血管、代谢、睡眠、心理健康以及生活方式和人口统计学因素等多个领域。据我们所知,这是迄今为止用于训练模型的最大、最多样化的可穿戴数据集。
从一万亿分钟的传感器数据中学习
为了构建预训练语料库,我们从五百万名已同意将其数据用于健康与福祉研究的人员中采样了去标识化数据,数据采集时间为2024年9月至2025年9月。该数据集覆盖超过100个国家、美国全部50个州,以及20多款Fitbit和Pixel Watch设备型号。我们从每个人身上采集了数周的数据,最终获得了超过20亿小时——即超过一万亿分钟——的分钟级分辨率信号。
SensorFM 接收来自五种传感器模态的34个一分钟聚合特征:光电容积描记法(PPG)、加速度测量、皮肤电活动(EDA)、皮肤温度和高度计。这些特征共同捕捉了完整24小时窗口内的心率与心率变异性、血氧饱和度、睡眠阶段、运动与步数、皮肤电导以及体温。
SensorFM 不依赖标签,而是通过自监督重建进行学习,其基础是 LSM-2 方法及其自适应与继承掩码(AIM)框架。这是一个关键的设计选择,因为可穿戴设备的数据缺失和碎片化(例如,一段时间内数据不可用)是常态,其原因多种多样,包括传感器电源循环、设备脱离手腕、省电运行模式以及传感器开关切换。传统的自监督方法假设输入是完整且不间断的,因此要么被迫对数据缺口进行插补(这可能会引入偏差),要么丢弃不完整的窗口(这会丢弃有价值的数据)。AIM 不走这两条路:它将现实世界中的数据缺失视为一种自然现象,并直接从不完整的记录中学习,将来自真实数据缺口的继承 token 与为了重建目标而人为掩码的 token 结合起来,并将两者视为等价。其结果是一种在构建时就具备缺失感知能力的表征。SensorFM 不仅容忍碎片化数据,还能有效地利用它,如下文生成结果所示。
SensorFM 通过缺失感知的掩码重建,在超过一万亿分钟的多模态传感器数据上进行了预训练。
同时扩展模型和数据规模是值得的。
任何基础模型的核心问题都在于:规模是否能够转化为能力。我们开展了一系列系统性的规模扩展实验,在预训练数据量(从约200万传感器小时到20亿传感器小时)和模型规模(从10万参数到1亿参数)两个维度上均跨越了四个数量级。
结果呈现出清晰且令人鼓舞的信号:随着数据和容量的增长,预训练损失呈现可预测的下降趋势,并且——关键的是——这些收益能够延续到下游健康任务中。在完整的500万人群语料库上训练的最大模型(SensorFM-B),其重建损失相比最小变体降低了31%,在下游分类任务上的性能平均提升9%(AUC),在回归任务上的性能平均提升21%(皮尔逊系数)。
最大的改进来自于同时扩展两个维度。按比例增加数据和容量,在生成式预训练和判别式下游性能上均能产生近乎线性的增益——而且这条曲线尚未显现出饱和迹象。在多个模型变体中,SensorFM-B在35项任务中胜出33项。
将数据和模型容量协同扩展数个数量级,能够在预训练和下游性能两方面都带来增益,且未出现饱和迹象。
一种表征,覆盖多个健康领域
为了测试所学表征的通用性,我们在35项判别式健康任务上评估了SensorFM,这些任务来自三项独立的、经机构审查委员会批准的前瞻性研究,共涉及13,985名参与者。任务涵盖六大类别:心血管健康、代谢风险、心理健康、睡眠、人口统计和生活方式。
为了直接检验嵌入向量的质量,我们冻结了SensorFM编码器,仅在其之上训练一个轻量级线性分类头——然后与基于工程化特征训练的监督基线进行对比。以下是我们的发现:
- 广泛泛化能力:在SensorFM嵌入向量上训练的线性探针,在35项任务中的34项上超越了基于工程化特征的监督基线,且无需任何任务特定的架构。
- 生理特征的隐式学习:加入人口统计学特征(年龄、性别等)会带来小幅提升——但随着模型规模扩大,这种提升效果会减弱,这表明更大的模型在预训练过程中已经隐式地捕捉到了与生理相关的特征。
- 在难以测量的病症上表现突出:扩展预训练对于抑郁症和焦虑症这类病症尤其有价值,这些病症因人而异,且在传感器数据中只留下微弱痕迹。SensorFM 似乎能够学习到通常掩盖这些信号的个体差异,并提取出跨人群共有的模式。
- 标签效率:仅使用一小部分有标签的样本,SensorFM 就能迅速超越仅依赖人口统计学特征和手工特征工程的基线模型——这在高质量标签稀缺的医疗保健领域是一项重要特性。
在冻结的 SensorFM 嵌入向量上使用线性探测法,在各类判别性健康任务中的相对性能提升。
一个用于构建预测头的“智能体课堂”
一个通用嵌入向量的价值,取决于将其适配到具体任务所需的工作量。传统上,要将嵌入向量转化为每个新端点的强预测器,需要手动进行特征工程、架构选择和超参数调优——这些繁琐的工作会随着任务数量的增加而变得更加困难。
为了实现自动化,我们构建了一个智能体“课堂”:一组相互协作与竞争的 LLM 智能体,它们迭代地生成、测试并优化可执行代码,以在 SensorFM 嵌入向量之上构建预测头。在我们的实验中,该系统探索了超过 30,000 个候选解决方案。
在 20 个分类任务中的 16 个以及 15 个回归任务中的 12 个上,由智能体设计的预测头击败了简单的线性探测法。有两个模式尤为突出:解决方案的质量随搜索过程单调提升,并且与底层 LLM 的能力成正比——能力更强的模型(例如更新版本的 Gemini)能产生更好的解决方案,而智能体之间的协作则有助于能力较弱的模型缩小差距。
一个由 LLM 智能体组成的“课堂”迭代地编写、测试和优化代码,以在 SensorFM 嵌入向量上构建预测头,并随着时间的推移不断改进解决方案。
为个人健康智能体奠定基础
最后,我们探究了 SensorFM 是否具备端到端的实用性——即作为一种工具,将 AI 健康教练与用户自身的生理数据相连接。我们将 SensorFM 集成到一个个人健康智能体中,并对比了三种从 31 位真实参与者档案生成健康摘要的条件:
- 人口统计学信息 + 日常可穿戴设备指标 + SensorFM 预测结果
- 人口统计学信息 + 日常可穿戴设备指标 + 真实测量值
- 人口统计学信息 + 日常可穿戴设备指标(仅基线)
一组临床医生在不了解实验条件的情况下,依据五个评分维度——背景、相关性、可论证性、个性化程度以及潜在危害——对生成的摘要进行评分,在超过 40 小时的专家评估中产生了 1860 个评分。研究结果令人瞩目。在所有评分维度上,加入 SensorFM 预测结果后,智能体的响应质量均显著优于基线水平。并且,基于 SensorFM 预测结果与基于真实测量值来指导智能体,两者之间没有统计学上的显著差异——这意味着模型推断出的结果对智能体的帮助程度,几乎与使用真实标签一样好。
将 SensorFM 作为个人健康智能体的工具,能够提升临床医生对响应的评分。
结论
SensorFM 预示着可穿戴健康研究领域的一次转变:从大量定制化的、针对单一结果的模型,转向一种单一的、通用的人类生理表征,这种表征可以灵活、高效且大规模地进行适配。通过从超过一万亿分钟的无标签传感器数据中学习,它能够泛化应用于心血管、代谢、睡眠和心理健康等多个领域;支持高效利用标签的适配和稳健的日常指标估算;能够通过智能体课堂自动实现专业化;并能将个人健康智能体建立在用户自身的信号之上。
致谢
我们感谢来自 Google Research、Google DeepMind 以及学术界的合著者和合作者对本工作的贡献。
- 标签:
- 生成式 AI
- 机器智能
快速链接
其他相关文章
2026 年 6 月 30 日 介绍 TabFM:面向表格数据的零样本基础模型 * 数据管理 · * 机器智能 · * 产品
2026年6月26日 利用冻结多 token 预测在 Pixel 设备上加速 Gemini Nano 模型 * 机器智能 · * 移动系统 · * 自然语言处理
2026年6月24日 思考以回忆:推理如何解锁大语言模型中的参数化知识 * 生成式 AI · * 机器智能 · * 自然语言处理
We present SensorFM, a foundation model for wearable health pre-trained on more than one trillion minutes of sensor data from five million people. By co-scaling model size and data, SensorFM learns a general-purpose representation of human physiology that transfers to 35 health prediction tasks, supports label-efficient adaptation and data infilling, and can serve as a grounding tool for a Personal Health Agent.
Quick links
Estimates suggest that billions of wearable devices are now in use, precisely tracking heart rate, movement, skin temperature, blood-oxygen levels, and sleep across days, weeks, and months. This continuous, longitudinal stream of physiology and behavior provides one of the most promising raw materials for preventive, personalized health. Yet turning those low-level signals into meaningful insights remains hard. First, baseline physiology, lifestyle, and health vary enormously from person to person, so a pattern that signals risk in one individual may not in another. Second, the labels needed to train models — confirmed diagnoses, lab results, validated questionnaires — are expensive, slow to collect, and essentially impossible to gather retrospectively. As a result, most wearable health models have been built one outcome at a time, with bespoke, supervised pipelines that target a narrow endpoint and struggle to generalize across the full breadth of human health.
In “Towards a General Intelligence and Interface for Wearable Health Data”, we take a different approach. We introduce SensorFM, a Large Sensor Foundation Model that learns directly from unlabeled wearable data at population scale. Pre-trained on over one trillion minutes of multimodal sensor signals drawn from five million consented participants, SensorFM learns a single, reusable representation of sensed human physiology — one that transfers across cardiovascular, metabolic, sleep, and mental health, as well as lifestyle and demographic factors. To our knowledge, this is the largest and most diverse wearable dataset used to train a model to date.
Learning from a trillion minutes of sensor data
To build the pre-training corpus, we sampled de-identified data from five million people who had consented to the use of their data for health and wellness research, captured between September 2024 and September 2025. The dataset spans more than 100 countries, all 50 U.S. states, and over 20 Fitbit and Pixel Watch device models. From each person we drew several weeks of data, yielding over two billion hours — more than a trillion minutes — of minute-resolution signals.
SensorFM ingests 34 one-minute aggregate features derived from five sensor modalities: photoplethysmography (PPG), accelerometry, electrodermal activity (EDA), skin temperature, and altimetry. Together these capture heart rate and heart-rate variability, blood-oxygen saturation, sleep stages, motion and steps, skin conductance, and temperature over a full 24-hour window.
Rather than relying on labels, SensorFM learns through self-supervised reconstruction, building on theLSM-2 approach and its Adaptive and Inherited Masking (AIM) framework. This is a crucial design choice, because missing and fragmented data (e.g., stretches of time where data is not available) is the norm with wearable devices, caused by a variety of factors such as sensors’ power-cycle, devices coming off the wrist, power saving modes of operation, and sensors switching on and off. Conventional self-supervised methods assume complete, uninterrupted inputs and so are forced to either impute the gaps (which can introduce bias) or discard incomplete windows (which throws away valuable data). AIM takes neither path: it treats real-world missingness as a natural artifact and learns directly from incomplete recordings, combining the tokens inherited from genuine gaps with those artificially masked for the reconstruction objective and treating the two as equivalent. The result is a representation that is missingness-aware by construction. SensorFM does not just tolerate fragmented data, it uses it productively, as the generative results below show.
SensorFM is pre-trained on over a trillion minutes of multimodal sensor data via missing-aware masked reconstruction.
Scaling model and data together pays off
A central question for any foundation model is whether scale translates into capability. We ran a systematic set of scaling experiments, spanning four orders of magnitude in both pre-training data volume (from roughly 2 million to 2 billion sensor-hours) and model size (from 100K to 100M parameters).
The result is a clean, encouraging signal: pre-training loss falls predictably as data and capacity grow, and — crucially — those gains carry over to downstream health tasks. The largest model (SensorFM-B) trained on the full five-million-person corpus reduces reconstruction loss by 31% over the smallest variant, and improves downstream performance by an average 9% performance gain (AUC) on classification tasks and 21% of performance gain (Pearson Coefficient) on regression tasks.
The biggest improvements come from scaling both dimensions at once. Increasing data and capacity in proportion produces near-linear gains in both generative pre-training and discriminative downstream performance — and the curve shows no sign of saturating. Across model variants, SensorFM-B wins on 33 of 35 tasks.
Co-scaling data and model capacity by orders of magnitude yields gains in both pre-training and downstream performance, with no sign of saturation.
One representation, many health domains
To test how general the learned representation really is, we evaluated SensorFM across 35 discriminative health tasks drawn from three independent, Institutional Review Board Approved prospective studies with a total of 13,985 participants. The tasks span six categories: cardiovascular health, metabolic risk, mental health, sleep, demographics, and lifestyle.
To probe the quality of the embeddings directly, we kept the SensorFM encoder frozen and trained only a lightweight linear head on top of it — then compared against supervised baselines trained on engineered features. Here’s what we found:
- Broad generalization: Linear probes on SensorFM embeddings outperform the feature-engineered supervised baseline on 34 of 35 tasks, without any task-specific architecture.
- Physiology learned implicitly: Adding demographic features (age, sex, etc.) gives a modest boost — but that boost shrinks as the model scales, suggesting larger models implicitly capture physiologically relevant traits during pre-training.
- Strength in hard-to-measure conditions: Scaled pre-training is especially valuable for conditions like depression and anxiety, which vary from person to person and leave only faint traces in sensor data. SensorFM appears to learn the individual differences that usually bury such signals and pick out the patterns that carry across people.
- Label efficiency: With only a small fraction of labeled examples, SensorFM quickly surpasses both demographic-only and feature-engineered baselines — an important property in healthcare, where high-quality labels are scarce.
Relative performance improvement of linear probing on frozen SensorFM embeddings across discriminative health tasks.
A “classroom” of agents to build prediction heads
A general embedding is only as useful as the effort required to adapt it. Traditionally, turning embeddings into a strong predictor for each new endpoint demands manual feature engineering, architecture selection, and hyperparameter tuning — tedious work that gets harder as the number of tasks grows.
To automate this, we built an agentic “classroom”: a set of collaborating and competing LLM agents that iteratively generate, test, and refine executable code to build prediction heads on the SensorFM embeddings. Across our experiments the system explored more than 30,000 candidate solutions.
The agent-designed heads beat a simple linear probe on 16 of 20 classification tasks and 12 of 15 regression tasks. Two patterns stood out: solution quality improves monotonically over the search, and it scales with the capability of the underlying LLM — more capable models (such as more recent versions of Gemini) produce better solutions, while collaboration between agents helps less capable models close the gap.
A “classroom” of LLM agents iteratively writes, tests, and refines code to build prediction heads on SensorFM embeddings, improving solutions over time.
Grounding a Personal Health Agent
Finally, we asked whether SensorFM is useful end-to-end — as a tool that grounds an AI health coach in a person’s own physiology. We integrated SensorFM into a Personal Health Agent and compared three conditions for generating health summaries from 31 real participant profiles:
- Demographics + daily wearable metrics + SensorFM predictions
- Demographics + daily wearable metrics + ground-truth measurements
- Demographics + daily wearable metrics only (baseline)
A panel of clinicians, blinded to condition, rated the resulting summaries across five rubric dimensions — context, relevance, justifiability, personalization, and potential for harm — producing 1,860 ratings over more than 40 hours of expert evaluation. The findings are striking. Adding SensorFM predictions significantly improved responses over the baseline on every rubric dimension. And there was no statistically significant difference between grounding the agent in SensorFM predictions versus actual ground-truth measurements — meaning the model’s inferences served the agent about as well as the real labels would have.
Using SensorFM as a tool for a Personal Health Agent improves clinician-rated responses.
Conclusion
SensorFM points toward a shift in wearable health research: away from many bespoke, single-outcome models and toward a single, generalist representation of human physiology that can be adapted flexibly, efficiently, and at scale. By learning from over a trillion minutes of unlabeled sensor data, it generalizes across cardiovascular, metabolic, sleep, and mental health; supports label-efficient adaptation and robust daily-metric estimation; can be specialized automatically by an agentic classroom; and can ground a Personal Health Agent in a person’s own signals.
Acknowledgements
We would like to thank our co-authors and collaborators from Google Research, Google DeepMind, and academia for their contributions to this work.
- Labels:
- Generative AI
- Machine Intelligence
Quick links
Other posts of interest
June 30, 2026 Introducing TabFM: A zero-shot foundation model for tabular data * Data Management · * Machine Intelligence · * Product
June 26, 2026 Accelerating Gemini Nano models on Pixel with frozen Multi-Token Prediction * Machine Intelligence · * Mobile Systems · * Natural Language Processing
June 24, 2026 Thinking to recall: How reasoning unlocks parametric knowledge in LLMs * Generative AI · * Machine Intelligence · * Natural Language Processing