By Anind K. Dey, Carnegie Mellon University, Principal Investigator.Here at Carnegie Mellon University in Pittsburgh, we have been working with people with Alzheimer's Disease for a few years, focusing on how to improve their quality of life and that of their familial caregivers'. A clear source of tension was episodic memory impairment or EMI. This was true for caregivers in having to repeatedly remind their loved one about a past event, and for people with Alzheimer's Disease in losing their sense of self from not being able to remember important events from their past. Our approach to addressing EMI was to capture lifelogging information (images, audio, sensors) of events they wanted to remember, and then to apply heuristics from our fieldwork to create salient summaries, to filter the vast amounts of information being collected down to a consumable and reviewable amount. This approach improved our subjects' confidence in their own memory and their ability to recall aspects of episodic memories, and reduced the burden on caregivers.
We are very excited about this approach of creating salient summaries, and for Project HealthDesign, we believe that we can apply this approach to detecting the onset of cognitive and physical decline, before an official diagnosis of an impairment such as Alzheimer's Disease. Often such a diagnosis is delayed due to a lack of recognition of change in one's abilities, and/or the extensive time between assessments. Earlier diagnosis would permit earlier interventions, both behavioral and pharmacological.
We are planning to deploy sensing technologies that can detect changes in the performance of everyday activities, such as medicine taking, sleeping, using the phone, and meal preparation. As our sensors detect changes in the quality of performance of these different activities (e.g., time taken to complete the activity, performing steps correctly or out of order, skipping or adding steps), our proposed system will create appropriate salient summaries to present to the person whose activities are being sensed, occupational therapists, physicians and other care providers who can offer interventions as necessary. The biggest benefits of our technological approach are that it can be used to monitor activities 24-7 and detect changes continuously and doesn’t place users into "test situations" where they behave differently because someone is observing them. An example of one of our sensor systems is shown here: a pillbox instrumented to detect which door a user has opened, what is placed inside each door, whether the pillbox is being shaken or turned upside down, etc. For the medicine taking task, and the other tasks, we are sensing all the steps that an individual takes to perform the task, in a privacy-preserving and comprehensive manner, from selecting the appropriate pills to swallowing them with water.
The project team
We've put together a great team for this project. Matthew Lee, my student, is a 5th year Ph.D. student in the Human-Computer Interaction Institute, and is focusing on the use of technology and salient summaries to improve quality of life of elders. Diane Collins is a professor in the Department of Rehabilitation Science and Technology at the University of Pittsburgh, and is also a licensed occupational therapist with over 20 years of clinical experience. Linda Kent is a licensed occupational therapist with many years of clinical experience at Presbyterian SeniorCare, a local network of living and care options for elders and people with disabilities, and a key partner in this project.
We are very excited about our project and hope you'll visit our webpage for updates and follow our progress through the blog!