WSU ’smart home’ keeps getting smarter
Being able to remember a simple daily routine can mean the difference between independent living and life in a nursing home for people with dementia.
New research coming out of the Center for Advanced Studies in Adaptive Systems (CASAS) at WSU could help millions of U.S. seniors cope with the memory loss associated with aging and other forms of cognitive decline to stay in their own homes longer.
Maureen Schmitter-Edgecombe, professor of psychology, and her team of graduate students are working with colleagues in electrical engineering and computer science to develop a new technology to help people with dementia remember to perform basic tasks of daily life, such as taking their medicine, preparing meals, or performing physical therapy exercises.
Designed to work with existing smart home systems, the new device would provide memory cues to help residents get back on track when they are unsure of what to do.
Schmitter-Edgecombe and Diane Cook, professor of electrical engineering and computer science, have been testing and refining the WSU smart home for the last seven years. It uses sensors on the walls, doors, and various household objects to help monitor, predict, and ultimately improve quality of life, especially in care of the elderly.
The researchers coauthored a recent study showing for the first time that computer algorithms can use motion-sensor data from the smart home to predict when cognitively impaired people are between activities and may need a prompt to perform a new task.
“If you have someone with cognitive difficulties, it is probably not a good idea to prompt them to start a new activity when they are in the middle of their physical therapy or cooking dinner. That can cause more confusion and errors,” Schmitter-Edgecombe said. “Our research shows the best time to prompt someone is between activities.”
The smart home system used in the study comes in a box and can be installed in an apartment or home relatively easily. It consists of roughly 30 infrared, motion, and vibration sensors that look at patterns of behaviors, such as whether a certain activity has happened before; when it took place; and how long it lasted and then deduce what’s going on.
Over time, the system uses this information to determine when the home’s inhabitant has completed a specified task and might need a prompt to perform a new one.
To test their system, the research team collected data from two experiments. One was conducted at the smart home on the WSU Pullman campus. The other took place at a retirement community in Seattle where smart homes have been installed in eight apartments. For the second experiment, each of the apartments housed a single older-adult who was asked to go about daily routines while the wireless sensors collected data. Every day, the researchers analyzed three hours of data starting when the participants first awoke. The system was able to accurately predict periods of transition between activities 80 percent of the time.
For example, the smart home used data from sensors on the front door and in the living room to determine when visitors arrived, how long they stayed, and when they left.
“The smart home is able to recognize the short period of time after visitors leave and before the resident starts a new activity,” said Kayela Robertson, a psychology postdoctoral candidate working on the project. “We are starting to develop a tablet or smart phone-based application that integrates with the smart home to provide a prompt to perform a new task during this transition.”
The end goal of the project is to develop an easy-to-use and portable tool that would provide reminders to perform a number of important activities. The technology could also provide smart home occupants with a list of activities they’ve done throughout the course of the day and give healthcare providers and loved ones a better idea of how their aging patients, parents, or other family members are doing, whether they live nearby or in another state.
“As a clinician, if I had an elderly patient who was recovering from some type of surgery that required performing regular exercises to ensure their recovery, it would be important to know if the person actually engaged in their activities,” Schmitter-Edgecombe said. “A system that could detect whether a person performed their necessary activities and, if they grew unsure of what to do next, could then provide them a prompt in real time would do wonders in helping them live healthy and independent lives for longer.”