Machine-learning techniques are powerful tools for data analysis that allow for identifying complex (non-linear) patterns, aggregating various types of data, and are robust to overfitting to deliver accurate predictions. We offer interpretable machine learning solutions for your data that also shed light on feature importance, helping you to understand which variables most influence your outcome. We also provide a transfer of knowledge so that your team can carry out such analyses independently.

Despite its effectiveness, the labor-intensive nature of exposure therapy and the scarcity of therapists limit its accessibility, prompting a need for scalable alternatives. We address these limitations by leveraging advanced AI technologies (Model Predictive Control and Bayesian Optimization), wearable biosensors, and augmented reality to create a scalable and personalized exposure therapy solution. Our approach will provide a customized clinician assistant system and a stand-alone solution for early intervention for at-risk individuals, thus cost-effectively treating and preventing anxiety disorders.

Leveraging NLP and LLMs, we seek to harness highly relevant but underutilized data derived from verbal and written diagnostic assessments and reports, a crucial resource in psychiatry that remains challenging to quantify effectively with traditional methods. By integrating these novel textual analytics with demographic and clinical data through our interpretable machine learning models, we aim to enhance the precision and accessibility of diagnostic and therapeutic tools.

Social interaction plays a pivotal role in human cognition and well-being. Hyperscanning studies of social interaction have associated increases in interbrain synchrony with positive outcome measures of cooperation, learning, and mutual understanding. We work on developing a hyperfeedback system designed to train and enhance interbrain synchrony using functional near-infrared spectroscopy (fNIRS) to benefit human interaction.

We develop new brain imaging methods and apply them to clinical conditions. Specifically, we use functional magnetic resonance imaging (fMRI) as a non-invasive therapeutic tool. This can be achieved by using real-time fMRI and neurofeedback of brain signals to teach participants and patients to self-regulate their brain activity. With our work, we aim to translate this cutting-edge technology to the clinics to benefit patients.