Creating adaptive, wearable technologies to assess and intervene for individuals with ADRDs

点击翻译按钮获取中文摘要

PROJECT SUMMARY / ABSTRACT Advances in machine learning and low-cost, wearable sensors offer a practical method for understanding, assessing, and intervening for Alzheimer's Disease and Related Dementias (ADRDs) in everyday spaces. We propose to create a Behaviorome research program that will create ground-breaking methods for building health-predictive models from wearable sensor data by mapping patterns of behavior using machine learning and pervasive computing technologies. This program will create innovative multidisciplinary ideas to address NIH ADRD Milestone 11.c, Embed wearable technologies/pervasive computing in existing and new clinical research. Our research program builds on a history of interdisciplinary research contributions in areas including human behavior modeling from longitudinal sensor data and design of novel assessment and intervention mechanisms. We propose to design and validate methods for mapping a human behaviorome “in the wild”, automatically assessing cognitive and functional health from behavior markers, scaling technologies through machine learning, linking health and behavior with their influences, and closing the loop with automated interventions. Similarly, our mentoring program builds on experience training students and early- career investigators to become leaders in the field of gerontechnology. We will recruit and train graduate students and early-stage researchers, including those from underrepresented groups, to grow an institutional multidisciplinary Behaviorome research program and to establish new research programs that contribute to the targeted Milestone. We will scale the impact of mentoring by establishing a webinar series and creating youtube videos that highlight and explain breakthroughs in the design and application of Behaviorome research. Results of this program will include scripts and templates to construct a behaviorome with resource- limited wearable devices, scale data and models to large diverse populations, integrate data with multiple information sources (e.g., genetics), automate health assessment and intervention, and create understandable explanations of data and models. These will contribute to existing clinical studies such as the clinician-in-the- loop smart home, digital memory notebook, and pervasive computing measures of functional performance. Furthermore, they will lead to new clinical studies that formalize connections between health and its influences, exploration of the impact of ethnicity and the built environment on health, and the design of ADRD interventions for medication adherence, task prompting, and negative interaction de-escalation. The proposed contributions are significant because they will provide insights on detecting and assessing ADRDs within a person's everyday environment using wearable sensing and pervasive computing methods that have not been investigated in prior work. Additionally, the mentoring steps will pave the way for a new generation of researchers to offer improved methods of addressing the need to understand, assess, and intervene for ADRDs in everyday settings, thereby improving quality of life and reducing health care costs.