Lili Zhang ins a postdoctoral researcher in Insight at DCU. Her research focuses on using computational models and machine learning techniques to understand how people learn and make decisions, particularly in naturalistic settings through taking advantage of pervasive internet-enabled devices, e.g. smartphones and sensor-based wearables. She is currently looking at how computational models might shed light on decision making in the human brain, as it relates to psychiatric disorders.
Computational models have found extensive application in the last decade, aiding in comprehending the intricacies of cognitive processes such as decision making. Such models are also referred to as neuro-computational measures. These measures hold significant potential in unveiling the dysfunctional processes thought to underlie psychiatric disorders, thus bridging the gap between fundamental research findings and clinical applications.
Nevertheless, the transition of these models into substantial advancements in clinical practice has been somewhat elusive. A primary hurdle is the absence of comprehensive assessments of the psychometric properties of experimental paradigms and model-derived metrics. To effectively characterise individual disparities and interpret personal trajectories—whether influenced by interventions or indicative of clinical progression—robust test-retest reliability is imperative.
Another significant challenge lies in making individual-level predictions as opposed to drawing inferences about group-level discrepancies in neuro-computational parameters.
With traditional lab-based methods, it is simply too difficult to predict and observe the onset of disorder-related behaviours of enough subjects for enough time for meaning analysis. New methodologies that can deliver cognitive assessments in natural settings at regular intervals or in response to state changes, e.g. Ecological Momentary Assessment (EMA), need to be invested and validated. To this end, we focus on examining and validating smartphones as a new methodology to densely sample the behavioural paradigms within the same individual and meanwhile capture clinically relevant symptoms in real time. The use of smartphone-based EMA methodologies in large-scale, naturalistic assessments may provide a much more nuanced perspective on individual trajectories in decision-making and learning and their contribution to mental health.