Smart sensors for a wearable-free and contactless virtual ward at home

该项目将支持关于保持家庭独立和家庭保健的倡议。为此，该联盟将探索使用一套微创、非穿戴式和非接触式传感器的可行性，为家庭和护理环境中的患者创建一个易于部署的监测系统。使用扩展/虚拟病房的家庭监测已被证明是应对2020-21年大流行期间挑战的有效解决方案（https://www.england.nhs.uk/nhs-at-home/covid-virtual-wards/）。虚拟病房通过为临床医生提供远程患者监测，加速了从医院到家庭和居住环境的出院。加速出院有许多好处：降低感染风险，减少失代偿（一种导致住院时间更长和预后较差的情况），以及增加医院病床容量。现有的方法是结合使用物理测量设备（如脉搏血氧仪）和电话服务，在家中管理患者并及早发现病情恶化。对于家中有其他护理人员/家庭成员，以及患者/护理人员较年轻且具有较高的健康和技术素养的患者群体，它们最为有效。核心传感器技术是基于毫米波（mm-wave）雷达，用于寻找运动和活动迹象，而无需使用侵入式摄像头或侵入式吊坠/可穿戴设备。人工智能被用来解释雷达的输出，创建居民活动的图片，并识别：他们是下床，在房间里走动，睡得很熟，还是他们可能摔倒了。它也可以用来测量心率和呼吸频率。主毫米波传感器与红外相机一起使用，用于非接触式温度和脉搏血氧测量，另外一套传感器将通过测量护理环境的状态（温度、空气质量等）来支持这些任务。时间序列算法和人工智能技术将用于解释模式和搜索传感器数据中的异常情况，以识别健康恶化。例如，一个人从床上爬起来走到浴室或厨房所花费的时间可以随着时间的推移而被监控，以报告他们的行动能力是在下降还是在改善。该项目的资金将用于在家庭护理环境（ExtraCare）中对患者和临床医生的焦点小组进行测试：这种类型的家庭监测的吸引力，最容易使用的技术以及界面的设计。这将超越人工智能行为准则。毫米波传感器的基础技术将进一步加强，以改进活动识别和生命体征检测人工智能模型，并扩展预测模型（如循环神经网络），以根据传感器输入预测患者健康变化。资金还将用于开发集成传感器所需的接口，将非接触式传感器与标准健康监测传感器（AHSN作为评估合作伙伴）进行比较，并与地方当局、NHS和护理社区的利益相关者进行接触。

The project will support initiatives on maintaining independence at home, and health within the home. To do this, the consortium will explore the feasibility of using a suite of minimally intrusive, wearable-free and contactless sensors, to create an easy-to-deploy monitoring system for patients at home and in care environments. Home monitoring using an extended/virtual ward has proven to be an effective solution to challenges during the pandemic in 2020-21 (https://www.england.nhs.uk/nhs-at-home/covid-virtual-wards/). Virtual wards accelerate discharge from hospitals to homes and residential environments, by providing remote patient monitoring for clinicians. The accelerated discharge has numerous benefits: reduced risk of infection, reduction in decompensation (a condition which leads to longer hospital stays and poorer outcomes), and an increase in hospital bed capacity. Existing approaches have used a combination of physical measurement devices (e.g. pulse oximeters) and telephone services to manage patients at home and identify deterioration early. They have been most effective for patient cohorts where there are other carers/family members at home and where patients/carers are younger and have a high level of health and technology literacy.The core sensor technology is based on millimetre-wave (mm-wave) radar, which is used to look for movements and signs of activity without the use of invasive cameras or intrusive pendants/wearables. Artificial intelligence is used to interpret the outputs of the radar, to create a picture of residents' activities and recognise whether: they are getting out of bed, walking across a room, sleeping soundly, or if they have potentially fallen over. It can also be used to measure heart rate and respiration rate. The primary mm-wave sensor is used in conjunction with an IR camera for contactless temperature and pulse oximetry measurement, and a further suite of sensors will support these tasks by measuring the state of the care environment (temperature, air quality, etc.). Time series algorithms and AI techniques will be used to interpret patterns and search for anomalies within the sensor data, in order to identify health deterioration. As an example, the time it takes a person to get up from bed and walk to the bathroom or kitchen can be monitored over time, to report on whether their mobility is degrading or improving. Funding from the project will be used to test with focus groups of patients and clinicians in a homecare environment (ExtraCare): the attractiveness of this type of home monitoring, the technologies which are easiest to use and the design of the interface. This will go beyond the AI code of conduct. The technologies underpinning the mm-wave sensors will be further enhanced to improve activity recognition and vital signs detection AI models, with forecasting models (such as recurrent neural networks) extended to predict patient health changes based on sensor inputs. Funding will also be used to develop the interfaces needed to integrate the sensors, evaluate the contactless sensors in comparison with standard health monitoring sensors (AHSN as an evaluation partner), and engage with stakeholders from the local authority, NHS, and care communities.