基于智能设备的精确健康监测关键技术研究

The health monitoring system utilizes various sensors to perform monitoring on the users' sleep quality, body behaviors and vital signs, which can further help users to better understand their own health conditions and timely access to the medical services. However, existing health monitoring systems are either not accurate enough or depend on dedicated sensors. In addition, few studies provide secure and efficient user verification on smart devices, which undermine the security of such health monitoring system. This proposal studies how to derive sleep related information from noisy data collected by smart device, and achieves fine-grained sleep monitoring which is capable of detecting the users’ respiratory status and sleep events simultaneously. Furthermore, our proposal designs unique features to capture the intrinsic characters embedded in each user’s body behaviors, and proposes a fine-grained behavior monitoring system by leveraging smart devices. Last but not least, our proposal investigates how to identify the segments which remain invariant within a user’s signature, and proposes a critical segment based online signature verification system to secure monitoring information on smart devices. If successful, our project will result in the development of fine-grained health monitoring system which can achieve both fine-grained sleep monitoring and fine-grained behavior monitoring by leveraging smart devices.

健康监测系统利用多种传感器对使用者的睡眠状况、身体行为和生命体征等进行连续监测，并帮助人们了解自身的健康状况，从而及时获取医疗服务。然而，现有的相关监测方案存在着过于依赖专业硬件设备和精度不够高等缺点。并且，现有基于智能设备的监测系统欠缺一种安全而高效的用户身份验证方案，给系统的安全造成一定威胁。针对现有工作的不足，本项目研究从带有大量噪声的传感器读数中挖掘睡眠呼吸和睡眠事件信息的方法，给出基于智能设备的精确睡眠监测方案。研究特征提取方法，对人们行为中的固有特点进行准确描述，解决利用智能设备实现精确行为监测的难题。探寻从签名中提取具有稳定特征笔画的方法，设计一种基于手写电子签名的用户身份验证方案，为监测系统提供必要的信息安全保护。本项目的研究成果对基于智能设备的精确健康监测系统设计具有重要的指导意义。

现有的健康监测方案存在着过于依赖专业硬件设备和精度不够高等缺点。并且，现有基于智能设备的监测系统欠缺一种安全而高效的用户身份验证方案，给系统的安全造成一定威胁。针对现有工作的不足，本项目研究从带有大量噪声的传感器读数中挖掘睡眠呼吸和睡眠事件信息的方法，给出基于智能设备的精确睡眠监测方案。探寻从签名和用户语音中提取具有稳定特征特征的方法，分别设计一种基于手写电子签名和语音的用户身份验证方案，为监测系统提供必要的信息安全保护。本项目的研究成果对基于智能设备的精确健康监测系统设计具有重要的指导意义。在基金资助期间，本项目共培养硕士生4名，发表论文8篇，其中CCF-A类顶级会议与期刊论文4篇，中科院一区顶级期刊论文1篇，CCF-C类会议论文2篇。