Category: Data mining methodology

New Article: Capturing temporal dynamics of fear behaviors on a moment‐to‐moment basis

tbrick on
Detail of fear seqeunces from several children

One early indicator of psychopathology risk for small children shows up in patterns of fear behaviors. In our new paper Capturing temporal dynamics of fear behaviors on a moment‐to‐moment basis, published in Infancy, we apply discrete sequence methods to data about children’s responses to a fearful situation.

One of the very cool things about this paper is that we identify a few different clusters of child behavior. We were hoping to find a way to identify a high fear group at risk to show dysregulated fear, which has psychopathology implications. We found them to some extent, but also identified some other groups that are interesting: a group of external regulators, who use a parent to help regulate their emotions; one low reactive group, who just aren’t that scared by things; and a cool group called fearful explorers who show fear, but don’t let it stop them from checking things out. This work provides some new data points to help understand the emergence of emotion regulation across child development.

New Article: Individualized Modeling to Distinguish Between High and Low Arousal States Using Physiological Data

tbrick on
Two timestreams, warped to common timing

As Wear-IT begins to take shape in more detail, we’ve been looking at new ways to understand and model physiological data on an individual level. Recent work with Ame Osotsi, Zita Oravecz, and Joshua Smyth examines individual differences in physiological response in a paper called Individualized Modeling to Distinguish Between High and Low Arousal States Using Physiological Data, just published in the Journal of Healthcare Informatics Research.
In it, we illustrate the importance of individual modeling for physiological states, showing that high-arousal states have physiological signatures that can be detected by machine learners, but that those signatures differ enormously from person to person, and individualized modeling is really important.

GSA 2019

tbrick on

Dr. Brick and Ms. Dickens presented talks at the annual meeting of the Georntological Society of America (GSA 2019) in Austin, Texas this week. GSA is always an interesting meeting, looking at the science of aging–a vital science, especially as the global trends show our population living longer, and the proportion of older adults increasing as time goes on. Topics included methods to model and understand patterns in individual growth trajectories (like Dr. Brick’s sequence learning methods), and the relationships between sleep and well-being.

Six-state emotional day solution

Other topics focused on issues of cognitive aging, dementia and Alzheimers, and technological and methodological innovations to model and improve the well-being of older adults through improved cognitive function, better mobility, and increased independence.

Society for Ambulatory Assessment (SAA2019)

tbrick on

Dr. Brick presented at the Society for Ambulatory Assessment’s 2019 meeting in Syracuse, New York.  The SAA is dedicated to examining the applications and uses of tools like Wear-IT to clinical and research settings.  As always, a phenomenal set of scientists were there.  Penn State’s College of Health and Human Development was there in force, with a number of great presentations.

Dr. Brick presented in-process work using new statistical and data mining methods focused on real-time feature selection.  The idea is that smartphone surveys are annoying, and the longer they are the more annoying they are.  Annoyance leads to people ignoring the surveys, which means less data and more participant costs (e.g. money, energy, burnout).  The goal of this new work is to make in-the-moment decisions about which questions to ask and how to ask them, in order to minimize the burden on the participants and maximize the amount of data gathered.

industryXchange 2019

tbrick on

Dr. Brick, Ms. Dickens, and Mr. Mundie all attended Penn State’s industryXchange 2019 to talk about new developments from the lab. This year’s workshop focused on sensors and their applications, so we mostly showed off applications of wearable and passive measurement devices, and the ways that they could be applied to improve health and well-being, and assist with psychopathology.

Dr. Brick also presented some upcoming work with Dr. Jessica Menold using these same approaches to enhance workplace efficiency, reduce worker stress and burnout, and improve on-the-job learning.