Twitter: @LMVizer

Short Version
I apply person-centered mixed methods to investigate techniques for detecting and monitoring brain conditions such as cognitive impairment and mental health. My research philosophy places the people I serve at the center, always integrating their perspective with sound research practices and appropriate technology to develop solutions that best serve their health management needs. I am particularly interested in using digital phenotyping that leverages passive markers of language and behavior to unobtrusively assess and monitor symptoms associated with trauma exposure, cancer-related cognitive impairment, behavioral health, and cognitive decline.

Long Version
I love to investigate the problems people have managing their health and then design, develop, and evaluate technology to solve those problems. In particular, I am interested in how brain health manifests in our technology interaction patterns and how we can use those patterns to passively monitor brain health. This approach is also known as digital phenotyping. I am currently investigating behavior patterns related to trauma exposure, cognitive impairment, dementia, stress, mental health, and cognitive decline. Digital phenotyping might also be useful for monitoring side effects of medications or therapies, or as a measure of comparative effectiveness. 

My area of expertise is Human-Centered Computing (HCC) applied to consumer health technologies. My thesis research investigated the use of keystroke and linguistic features of text to support the early detection of cognitive decline or cognitive stress. During my postdoctoral career, I applied this method to other conditions and gained additional training in qualitative research methods. With experience both qualitative and quantitative research methods, I am able to develop technology solutions that are grounded in a deep understanding of people’s health needs.

Brain conditions—such as mental illness, cognitive stress, head trauma, and neurodegenerative illness—can negatively affect cognitive function, leading to harmful psychological, physical, behavioral, and performance consequences, including impaired judgment and memory, sleep and motor disturbances, as well as difficulty performing self-care tasks. Psychology researchers have developed valid and reliable tests for discrete testing of cognitive function that identifies cognitive functioning outside of the normative standard, and we know that the cognitive processes measured by these tests coincide with those needed to use information technology. Furthermore, the increasing trend in self-monitoring (Fox & Duggan, 2013), partly popularized by the Quantified Self community (, offers a complement to traditional medical monitoring. However, current monitoring methods are too obtrusive, inconvenient, course-grained, and generalized to use for continuous monitoring and early detection of changes in cognitive function. Despite the ubiquity of information technology (Pew Internet Project, 2012) and the public health burden of brain conditions (Harrison, 2014; World Health Organization, 2008), we have yet to fully explore how to develop personalized models of Human-Computer Interaction (HCI) markers (Kalman, Geraghty, Thompson, & Gergle, 2012) to passively detect changes in cognitive function in order to implement appropriate patient support and interventions.

Past Work
To address this gap, I investigate how the way people interact with technology can help us characterize brain health. These behavior patters will allow us to better monitor cognitive function outside of the clinic and lead to more timely and effective interventions to reduce or prevent impacts of stress, aging, cognitive decline, or other cognitive conditions (Lupien, Maheu, Tu, Fiocco, & Schramek, 2007).

My thesis research explored how cognitive stress and cognitive decline affect linguistic and keystroke attributes of typed text (Vizer, Zhou, & Sears, 2009; Vizer, 2013; Vizer & Sears, 2015; Vizer & Sears, 2017). I conducted a study with 20 young adults, 21 older adults with normal cognition (NC), and 17 older adults with Pre-Mild Cognitive Impairment (PreMCI). Using binary logistic regression, I built models from candidate markers and used them to classify typing samples. My research shows promise in classifying typing samples as coming from stress vs. no-stress conditions (60%), younger adult vs. older adult (79%), and NC vs. PreMCI groups (79%) (Vizer & Sears, 2015). An extended analysis of some younger adult participants shows significant individual differences in the markers involved in the cognitive stress response. Tailored classification models yielded accuracies from 62 to 88% per person (Vizer, 2013; Vizer & Sears, 2017). Markers such as pause activity and affect confirm results of other research, while markers such as input rate and references to others are new. Some of these new markers are unique to interactions with a computer and could form the basis of a system for continuous monitoring of cognitive health using everyday interactions with technology.

To investigate the HCI markers relevant to other brain conditions, I collaborated with researchers at University of Washington Medicine and Sage Bionetworks—a Seattle non-profit organization—to deploy mHealth apps for active and passive monitoring of symptoms to people with depression, anxiety, Parkinson’s Disease, and cognitive issues after chemotherapy. Amy Bauer, MD deployed apps to patients with depression and anxiety to monitor symptoms between clinic appointments. The apps recorded both active measures such as the PHQ-9 as well as passive timing measures and linguistic input from surveys. With Sage Bionetworks, I helped develop apps with active and passive measurement of cognition, sleep, exercise, and mood to monitor Parkinson’s Disease and breast cancer survivors. These apps were deployed to thousands of participants through initiative including PatientsLikeMe, the Michael J. Fox Foundation, and foreign government health systems.

Current Work

Word use after trauma exposure

  • Investigating how word use is associated with symptoms in people after trauma exposure

Mobile technology and cognitive impairment in older adults

Conceptual Model of Shared Health Informatics: CoMSHI

Survivors of Adolescent and Young Adult (AYA) Cancer

  • Investigating alcohol and tobacco use in AYA survivors with DataAware student
  • Reviewing research on cancer-related cognitive impairment in AYA survivors with DataAware students
  • Investigating AYA survivors’ use of education and career resources with DataAware students

Mobile technology for assessing Traumatic Brain Injury

  • Reviewing the literature on detecting concussion and comparing to the evidence base of available apps

Future Work
The information learned during monitoring is a valuable basis for education or intervention. This could be as simple as alerts to raise awareness of cognitive stress or data visualizations showing trends of cognitive function over time, or could include a tailored program of physical and cognitive exercises as well as an interface to medical professionals. Although researchers have explored some of these issues, we need more research into alert timing, information presentation, and motivation for behavior change. As I learn more about the social and psychological issues surrounding health and wellness, I become more committed to careful human-centered design to ensure usability and usefulness of the resulting tools. I would also like to collaborate with industry partners to take advantage of cutting edge technology to promote engagement, encourage healthy behaviors, and improve outcomes.

Finally, cutting across monitoring, education, and intervention are concerns about the privacy and security of health information. Although this is an important issue for all, it is particularly important to understand the perspectives of vulnerable populations, such as people with brain conditions, and how their needs might be different than others.

In the future, digital phenotyping may be as much a routine part of a healthy lifestyle as regular physicals and laboratory tests. It is a low-cost health monitoring method incorporated behind the scenes of everyday activities that can yield valuable information to help people understand their health and make informed decisions. I expect this research to contribute to debates on personalized medicine and play an important role in shaping the discussion of health monitoring and patient engagement in the coming years.