Mixed Reality Brain Visualization
Traditional learning of brain anatomy typically involves the use of images, plastic models, and cadaveric dissection. These methods are either resource intensive, or can’t fully capture the intricacies and small scale structures of the human brain. Can mixed reality help students learn neuroanatomy? This project is a collaboration with Microsoft, and assesses whether a HoloLens-presented virtual brain is a useful neuroanatomy learning tool for students.
Smartphones as Research Tools
While not a line of research per se, I’m very interested in using emerging technologies as tools to improve the process of science. Many of my exercise/motion based studies utilize smartphones. One of my interests is developing smartphone apps that can utilize onboard accelerometers for gait measurement. Hopefully this will provide researchers with easier (and cheaper) access to tools that can be used to study exercise and motion – they just need to pull out their phone.
Gait and Cognitive Load
Gait (i.e. walking) is highly related to cognitive function. Does a fundamental change to cognitive state, such as imposing high cognitive load on the subject, change the dynamics of gait? An extension of this idea is the distinction between natural and urban environments. Urban environments have been shown to impose high levels of cognitive load on an individual, so does gait change as a function of walking in urban vs. natural environments?
In order to study gait dynamics outside of the lab (i.e. when they’re walking through different environment types), we need to develop and validate a new measurement tool. The goal is to develop a smartphone based application that can record gait information as accurately as traditional methods. Ultimately this will give researchers easier access to gait measurement tools, without having to resort to specialized equipment.
Measuring Physical Activity with Smartphones
Daily physical activity is typically measured using accelerometers. Smartphones have built-in accelerometers, but an issue with phone-based measurement is that smartphone location does not stay constant (i.e. sometimes it’s in your pocket, other times it is in your hand). We can leverage the power of recurrent neural networks to transform smartphone acceleration (independent of location) into a common metric, which can then be used to calculate physical activity levels. While this transformation process can introduce noise into the resulting measure, it provides a way to measure physical activity without worrying about the location of the smartphone.
Exercise and Cognitive Function
A lot of research shows the benefits of recent bouts of exercise of cognition. For example, subjects are typically asked to perform some exercise and are tested on a cognitive task immediately afterwards. However, what are the “near term” benefits of exercise? Does exercise you performed yesterday, or over the past few days, affect cognition? How do lifestyle factors such as sleep and diet affect this relationship?
Given a “near term” benefit of exercise, does it benefit only specific cognitive domains? For example, does recent exercise serve to improve attention? Or memory? Or both?
Exercise and School Performance
While many studies show exercise improving performance on computer-based cognitive tasks, does it also improve performance in a more applied setting? The main questions driving this line of research are “does exercise improve learning in a classroom environment?” and “does exercise improve performance on a school exam?”
Machine Learning and Attention
EEG data contains a wealth of information about attentional states. However, traditional methods are not able to identify large scale functional changes due to the amount of data present in collected EEG data.
This project uses an ensemble of various machine learning algorithms (ANN, CNN, RNN) to identify fluctuations in attentional state across a variety of tasks.