Understanding and predicting falls of people living with dementia

跌倒是一个严重的健康和社会护理问题，因为它们与缺乏社交(外出)、住院、髋部骨折、死亡、行动不便等有关。已公布的证据表明，认知障碍，特别是痴呆症，是跌倒的风险增加因素，因为双重任务和分享注意力的能力降低。最近的研究重点是开发基于人工智能的跌倒检测方法。然而，这项研究还没有建立在更广泛的社会和心理背景下。该项目旨在通过将计算机科学和心理学方法结合起来进行研究，试图了解跌倒的机制和人们跌倒的更广泛背景，以解决这一不足。这一新的理解应该允许制定干预策略。关键的研究问题可能包括：能否开发基于人工智能的方法来检测灵活性和稳定性的丧失，并将其部署在真实环境中？与检测到的跌倒相关的环境因素是什么，以及哪些环境因素有利于注意力转移？能否设计干预措施来缓解这些已确定的因素？成功的申请者将设计基于人工智能的方法来检测步态和姿势的不稳定性，这些因素表明跌倒的风险更高。这些方法将基于并扩展主主管在识别家庭环境中偏离学习的“正常”模式的动作方面的成功工作。他们将被用来识别家庭中有较高跌倒风险的区域。65岁以上患有早、中期痴呆症的参与者将被招募，在家里或疗养院进行研究。焦点小组和神经心理测试将有助于将检测到的易摔倒区域与摔倒风险增加的可能原因联系起来，例如注意力转移的原因。在重点小组期间也可以讨论减少或抑制这些因素的可能干预措施。迭代发展战略将有利于用户参与方法的设计和评价。这将确保所开发的方法在实际场景中可用并实现效果。HCRW将帮助获得与威尔士政策制定者的直接接触，以帮助影响政策并使其在威尔士产生最大影响。

Falls are a serious health and social care concern, as they are related to lack of social interaction (going out), hospitalisation, hip fractures, death, immobilisation, etc. There is published evidence that cognitive impairment, especially dementia, is an increased risk factor for falls due to reduced ability to dual-task and share attention. Recent research has focused on developing AI-based methods for detecting falls. However, this research is yet to be set in wider social and psychological contexts.The project aims at addressing this shortage by combining computer science and psychology methods in a study that seeks to understand the fall mechanisms, and the wider context within which people have falls. This new understanding should allow devising intervention strategies. Key research questions are likely to include:Can AI-based methods be developed to detect mobility and stability losses, and be deployed in real environments?What are the environmental factors associated with the detected falls, and that favour attention shift?Can interventions be designed to alleviate these identified factors?The successful applicant will design AI-based methods to detect instabilities in gait and posture that are indicative of higher fall risk. These methods will be based on, and extend, the main supervisor's successful works on identifying movements that deviate from a learnt 'normal' model in home environments. They will be used to identify areas in homes that present a higher risk of fall.Participants aged 65+ with early and middle dementia stages will be recruited, for studies either at home or in care-homes. Focus groups and neuropsychological tests will help linking the detected fall-prone areas with the likely causes of increased fall risk, such as causes of attention shifts. Possible interventions to reduce or suppress these elements may be discussed during focus groups as well.An iterative development strategy will favour user-involvement in the design and evaluation of methods. This will ensure that the developed methods are useable in real scenarios and realise impact. HCRW will help gain direct exposure to Welsh policymakers to help influencing policies and enable a maximum impact in Wales.