Rapid advances in technology and computer science have the potential to greatly enhace healthcare. Prof. Misha Pavel is conducting research on how to make the latest technologies benefit our health in the DigiWELL project. During his scientific career in the United States, he has been involved in many technological solutions that have focused on computational models for predicting individuals‘ health states and how those predictions can be used in health technology interventions to the home. In America, he studied electrical engineering, computer science, mathematics, and psychology, successfully combining all of these disciplines in his research. His contributions span a spectrum of areas, from psychophysics and cogntive science to computer science and engineering. In our interview, he explains the importance of multiplinary teams in science, engineering and healthcare, and how various technologies can bring us major health benefits.
Misha Pavel and Marcela Ely, both participating in the DigiWELL project.
What are you currently working on in the US?
It turns out that in Europe, the USA, but especially in Japan and Germany, there is a rapidly increasing proportion of older people who depend on the support services from their community, health care system, as well as family caregiving support. We are currently working on technologies to improve the quality of life for older adults and their caregivers. Our approach includes deploying sensors that monitor individuals‘ activities in the home, for example sleep quality, eating behaviors, and activity levels. We also monitor their physiological states. Totether with computational models of behaviors we are able to detect decline in individuals‘ cogntive and physical abilities. Combining sensor data with computational modeling we are striving to detect and even prevent many injuries and potential cogntive and physical decline. We are particularly focused on specific parts of a home with high incidence of adverse events such as those in the bathroom, and other locations where people are most vulnerable. Using communication applications along with inferences from sensors data, the family can see whether grandma or grandpa is OK. Especially in the USA, relatives often live very far from their older family members. The family monitoring is useful in responding to adverse events like falls. The elderly may carry emergency push-button pendants that can call for help, but they often do not use it because they may not remember to or because they want to avoid an ambulance. Meanwhile, the family has no idea that anything is wrong. The healthcare-focused technology also impacts a nation’s economy, because the elderly are often cared for by a relative who is usually still of working age, but now cannot work full-time because of the caregiving responsibilities. We are developing aids to help with the self-sufficiency and wellbeing of seniors. Although we use the most sophisticated scientific tools, a lot of our focus is on user-inspired applied research.
This is also related to the fact that in America there is perhaps more pressure to commercialise research. Are successful spinoff and startup companies based on university research being created in the US?
I have experience in this area, but even in America successful commericalization is very difficult. There are relatively few very successful outcomes, e.g., patents that have led to profitable spinoff companies. The really successful academic spinoff companies like Gatorade from University Florida, which makes nutritionally balanced drinks for athletes, are really more of an exception. Commercialization and industry-academic collaborations successes are limited by the differences between academic and commercial environments and goals. Academics are rewarded for teaching, publications, winning research funding, teaching, and disseminating their results. In commercial settings, workers are rewarded for developing economically feasible solutions, enhancing the products while keeping their solutions secret. Successful commericalization or collaboration in heathcare technology domain benefits from a combination of the two strategies. The development of solutions require broader teams where they complement each other and work closely together towards a common goal. In my opinion, it is important to first find a group of people who are genuinely interested in a particular problem and willing to commit their energy to this cause. The key is to first focus on developing a business plan and then stick to it. It is imperative to meet all the promised deadlines and milestones.
The DigiWELL project focuses on digital technologies and their use for the benefit of health. You mentioned that you are working in the US to deploy sensors that track the movements of the elderly. What do you see as a research opportunity now within the DigiWELL project?
I believe that information communication technolgy has the potential to revolutionize healthcare. The Covid pandemic has shown us that for some kinds of help we can use video communication applications, e.g., Zoom, that reduce the necessity to go to a hospital or clinic when an online consultation with a medical ptrofessional is sufficient. But the key advance was to have this and similar services covered by insurance.
DigiWell is taking the technology-based approach a step further: it investigates whether it is possible to simultaneously improve healthcare and wellbeing, i.e., the quality of life of individuals. It is well documented that increasing physical activity improves health outcomes and may increase the quality of life for a wide range of populations. DigiWell is using monitoring devices in conjuction with subjective responses to investigate the effects of activity level, in conjunction with factors such as weather and air polution.
How specifically?
For example, people in general and older people in particular should exercise regularly to improve and maintain their heath. In one project in collaboration with UC Berkeley, we developed a set of exercises specifically designed for the elderly. Users were guided by an interactive storyboard display of our exercise coach giving tailored instructions based on data from a near-infrared 3D camera. Older adult users could also see a stick-figure representation of themselves on the side. This allowed them to perform chair exercises safely at home. This approach was helpful in achieving improved strength, flexibility and endurance, as well as getting blood flowing to the brain. We then designed more such exercises for different groups of people.
In another project we researched approaches to help people adhere to their medication-taking regimen. If you ask people if they take their medication as directed by their doctor, they will tell you that ninety-five percent of the time they take their medication correctly. But research shows that people take their medication correctly less than fifty percent of the time. We deveoped several AI-based approaches to signficantly improve the patients´ adherence to their medication administration prescribed by their doctors.
Will you also be involved in the development of some digital tools and software within the DigiWELL project?
I hope that I would be able to contribute in several ways. Most of the approaches in machine learning and AI are data-driven. These approaches have been successful for population-based inferences and interventions. Our approach is leveraging known principles from psychology and physiology to develop causal, dynamic models of individual participants. This approach combines electrical engineering, computer science and psychology to develop the models. The primary benefit of modeling individuals by developing “digital twins” is the personization of assessment and interventions. Our own research is aimed at the development of dynamic models of individuals and their behaviors that would allow us to optimize experimental designs and interventions. We also apply computational models to interpret subjective responses of indiviuals on different scales. In particular we evaluate the amount of information that an individual is able convey in a sequence of ecological momentary assessments (EMA).
So the goal is to connect computer scientists with psychologists, for example?
The goal is to improve peoples‘ health and wellbeing. To achieve this goal we need to bring together strong teams comprising engineers, computer scientists, statisticians, psychologists, caregivers and the recipents of the care. Our research is to trace people’s habits, to understand the psychological processes of many different individuals, what motivates them, and what deters them from various healthy behaviors. For example, help them to overcome “laziness” and go for a three-minute run every morning and keep it up until it becomes a habit. We want to understand the process of learning when a behavior becomes a habit. So we need to collect a very large amount of data and then evaluate it, because every person is likely to deviate from a population average. For example, people generally do not like to be told what they should do, but then comply with socially accepted norms. To discover these principles we need to collect data over a longer period of time and gradually innovate and develop models, finding out which ones work and which ones don’t. This is where the mathematics and data science come together and support our research. Because there will be a lot of data, from a large sample of the population, even those models will have several different variations based on different parameters. It’s the same as when you collect data on weight, height and so on. But in this case we will collect data on how people learn, how they react to different stimuli, what causes them stress, etc. This implies that there are trillions of parameters that we need to evaluate to understand how people think. The whole thing is a huge technical challenge for us.
Bio:
Misha Pavel is native to the Czech Republic, but currently lives and works in in Palo Alto, California as a faculty member in the Silicon Valley Campus of Northeastern University. In the mid 1960s, as a 17-year-old student, he emigrated with his parents to the United States, where he started out as a TV repairman – his hobby before he left what used to be Czechoslovakia. Since his knowledge of English was limited, he enrolled in a school for electronic technicians – RCA Institute. Overcoming his handicap, he received a scholarship to the Polytechnic Institute of Brooklyn, where he finished his baccalaureate degree in Electrical Engineering in two years. He was then hired by the exclusive research organization, Bell Laboratories, and simultaneously obtained his Master´s degree at Stanford University. His first major assignment for Bell Labs was to create a mathematical model of the telephone network that piqued his interested in the understanding the complexity of human cognitive processes. This led him to receive a PhD in mathematical psychology at New York University, aiming to represent psychological and behavioral phenomena using mathematical rigor. His initial focus was developing models of human sensory processes captured by psychophysics such as auditory and visual perception, but also began developing approaches to use modeling in support of measuring and optimizing usability of devices and systems. Equipped with this muti-disciplinary background, he obtained a faculty position at Stanford University. His research there focused on computer applications in cognitive psychology. With advances in sensor technology, he continued to continuously measure and evaluate human behavior in real life. This led to a professorship and founding chair of a new Department of Biomedical Engineering at Oregon Health & Science University. Here he delved into research on using technology to facilitate the care of older adults and improve their quality of life. His work combines artificial intelligence with psychological and physiological knowledge gained from previous research. Dr. Pavel was a co-founder of the Oregon Center for Aging & Technology, with a focus on home monitoring, mathematical models for inferring health states and activities of older adults, as well as home health interventions facilitated by new technologies.
Throughout his academic career, Misha Pavel has worked collaboratively with commercial organizations, in an effort to bring new discoveries to routine use. These organizations have included XEROX PARC, the Intel Corporation, NASA Ames, as well as many smaller companies. He has also worked with the US government, most notably starting leading the Smart & Connected Health Program at the US National Science Foundation. Currently, Dr. Pavel is a Research Professor in Computer Science at Northeastern University and is also now working with the University of Ostrava on the DigiWELL project, where he focuses on advanced mathematical data analysis and the technology design of health interventions.
