VERIPHYS
从感知到智能，让每一份身体数据都可被验证。 From sensing to intelligence — every body signal, verified.
从治病走向全生命周期健康管理 · 让医院级的诊疗走进每一只日常设备From treating illness to lifelong health management — hospital-grade care in everyday devices.
从治病
走向全生命周期健康管理From treating illness
to lifelong health management
- WHERE医院Hospital
- WHEN发作期才就医Only when symptoms flare
- DATA单次门诊、主观量表One-off visits, subjective scales
- TOOLS昂贵、依赖临床医生的大型设备Costly, clinician-dependent equipment
- AFTER出院后缺少持续管理Little follow-up after discharge
- WHERE院内 + 院外，主战场在居家In and out of hospital — mainly at home
- WHEN全生命周期，每一天Across the lifespan, every day
- DATA连续、客观、可验证的数字标识物Continuous, objective, verifiable biomarkers
- TOOLS戒指、耳机、手机、毫米波——日常设备Rings, earbuds, phones, mmWave — everyday devices
- LOOP感知 → 推理 → 居家干预与用药调整Sense → reason → home intervention & dosing
以人为尺度的人体世界模型，
覆盖全生命周期的慢病与健康管理。
An anthropocentric world model of the body, spanning chronic-disease and health management across the entire lifespan.
以 COPD 为例：全球第三大死因，中国约有 1 亿 患者。COPD alone: third leading cause of death worldwide, ~100M patients in China.
慢病的关键是「预防 + 管理」，主战场在居家，而不只是医院里的「治病」。Chronic care is prevention + management — the main arena is home, not only treatment in hospital.
连续感知 → 数字标识物与筛查评估 → 居家干预、康复训练与用药调整。Continuous sensing → biomarkers, screening & assessment → home intervention, rehab and medication adjustment.
SENSE
捕捉身体信号Capture body signals
可穿戴与无接触感知，让每一个与慢病相关的生理与行为信号，都能在日常被持续拾取。Wearable and contactless sensing continuously captures every physiological and behavioral signal tied to chronic disease.
→REASON
推断数字标识物Infer digital biomarkers
数字标识物与可解释模型，把数据转成可被验证的洞见——每一次测量都经得起复核。Digital biomarkers and explainable models turn data into insight that can be verified, measurement by measurement.
→CONTROL
闭环干预方案Close the care loop
居家干预与个性化方案直达用户，并把结果送回下一次测量，形成持续的健康闭环。Personalized at-home interventions reach the user directly, and outcomes feed back into the next measurement.
以人为中心的物理智能框架Human-Centric Physical AIHuman-Centric Physical AI FrameworkFramework
从异构物理感知，到多模态感知基础模型，再到任务自适应的人体理解From heterogeneous physical sensing, to multimodal sensing foundation models, to task-adaptive human understanding
点击模态，模拟某一传感器缺失Tap a modality to simulate a missing sensor
单模态基础模型Single-modal foundation models
面向通用表征、与具体应用解耦General representations, decoupled from specific applications
多模态融合基础模型Multimodal fusion foundation model
- 相关模 / 互补关系Correlated / complementary
- 冗余 / 跨源解耦Redundant / decoupled
- 继承 / 限制规则Inheritance / constraint rules
标准化的人体中心表征，用于下游建模与推理A standardized human-centric representation for downstream modeling and reasoning
充分学习各模态的独特信息。Fully learn what is unique to each modality.
对齐 · 建模关系 · 选择 · 组合Align · model relations · select · compose
建模不同模态之间的互补与协同关系。Model complementarity and synergy across modalities.
支撑下游推理与应用。Supports downstream reasoning and applications.
光谱笔、耳机、毫米波、手环、深度摄像头、指环——六种物理感知模态，都可以是日常可及的设备。Spectral pen, earbuds, mmWave, band, depth camera, ring — six physical modalities, all as everyday devices.
Research built
under hospital light.
公司技术由 HKUST QIAN GROUP 的研究体系提供基础，在香港科技大学持续推进六年以上的 AIoT 全生命周期慢病 / 健康管理研究。Our technology rests on the research program of HKUST QIAN GROUP — more than six years of systematic AIoT research on lifelong chronic-disease and health management.
- Focus
- AIoT 赋能居家全生命周期慢病 / 健康管理AIoT for lifelong chronic-disease & health management at home
- Framework
- 以人为中心的物理智能框架 · 五层架构Human-Centric Physical AI Framework · five layers
- Venues
- MobiCom · UbiComp / IMWUT · SenSys · npj Digital Medicine
Why we can build this base.
正是多年的积累，让框架的每一层都有据可依、有术可用。Years of groundwork mean every layer of the framework is evidence-backed and ready to use.
模态与技术栈Modalities & tech stack
六大物理感知模态全覆盖，从传感器硬件到信号处理、多模态融合模型，都自主可控。Full coverage of six physical modalities — from sensor hardware to signal processing and fusion models, all in-house.
- HW传感器硬件Sensor hardware
- DSP信号处理Signal processing
- FM多模态融合模型Fusion models
临床验证与数据Clinical validation & data
900+ 名真实受试者、7+ 家合作医院，沉淀了真实世界健康数据与临床知识。900+ real subjects and 7+ partner hospitals, building real-world health data and clinical knowledge.
- 7+合作医院与机构Partner hospitals
- RWD真实世界数据Real-world data
- KNW临床知识Clinical knowledge
闭环系统经验Closed-loop experience
多个方向跑通了“评估—监测—预警—干预”全链条。The full assess–monitor–warn–intervene chain is running in several directions.
- COPDEasySpiro → AECOPDetector → DeepBreathEasySpiro → AECOPDetector → DeepBreath
- EYESDE 筛查 → Blinic 居家管理SDE screening → Blinic at home
- MSKKneeGuard · ACLGuard · RTSenseKneeGuard · ACLGuard · RTSense
Six ways to listen.
覆盖可穿戴与无接触两大路线的六大物理感知模态，构成人体的物理观测空间。选择一种模态，看看它如何“听见”身体。Six physical sensing modalities across wearable and contactless routes form the body's physical observation space. Pick one to see how it listens.
Precision, not promise.
每一个数字都来自真实被试与院内验证。Every number comes from real subjects and in-hospital validation.
多项工作在五家医院同步验证，并与合作医院推进大规模部署计划——从「做出系统」走向「用进临床」。Validated across five hospitals in parallel, with large-scale deployment underway — from building systems to putting them into clinical use.
Bringing the lab home.
用耳机替代肺活量计，用手机替代听诊器，用摄像头替代康复师。Earphones instead of spirometers, phones instead of stethoscopes, cameras instead of rehab coaches.
Care you wear.
把实验室级的感知装进你每天都会戴上的东西里。Lab-grade sensing, inside the things you already wear every day.
CuSa
CuSa 智能指环CuSa smart ring
睡眠、HRV 与血压趋势——长期佩戴的安静信号层。Sleep, HRV and BP trends — a quiet, always-on signal layer.
Band
智能手环Wristband
震颤、运动与血压，辅助帕金森管理与运动康复。Tremor, motion and BP — supporting Parkinson's care and rehab.
Buds
智能耳机Earphones
肺功能、呼吸音与语音——把大型设备带回家庭。Lung function, breath sounds and voice — big-hospital devices, brought home.