Harvard Psychiatric Research
Worked as a Data Science Research Assistant by supporting coding and data structuring work with Dr. Dillon and Dr. Cataldo in the Motivated Memory and Learning Lab at McLean. I worked extensively with Python to support Dr. Cataldo’s coding and modeling work for Dr. Dillon in primarily Python and R. This involved developing more efficient coding approaches and running models on patient data including fMRI. It also resulted in validating the findings of Dr. Dillon’s prior research with the soft max and drift diffusion models, culminating in my creation and presenting of a poster (Pictured on the right) at McLean Research Day in January and Harvard Psychiatry Research day in April. This provided me with great public speaking and leadership experience to represent my lab’s research to members of Mass General Hospitals.
Modeling reveals slower learning from positive but not negative outcomes in depression.
Background. Anhedonic depression may reflect dopaminergic abnormalities (Pizzagalli, 2014), suggesting impairments in signaling reward prediction errors (RPEs; Schultz, 1998) critical for reinforcement learning (Huys et al., 2013). Importantly, however, few prior studies have investigated how participants make decisions based on learned values, which the reinforcement learning drift diffusion model (RLDDM; Pedersen & Frank, 2020) can do. We therefore investigated the impact of depression on learning and decision- making by fitting the RLDDM to data from healthy and depressed adults.
Methods. Forty-three unmedicated adults with Major Depressive Disorder (MDD) and 41 healthy controls (HC) completed the Probabilistic Selection Task (PST; Frank et al., 2004). In the PST, participants use probabilistic feedback from 240 training trials to learn to select the more frequently rewarded symbols out of three pairs. The data were fit with the RLDDM, a Bayesian hierarchical model that uses Q-learning to model value assignment but models choice with the DDM (Ratcliff, 1978), providing a more comprehensive treatment of decision-making than the commonly used Softmax. Our implementation included parameters for positive and negative learning rates, the speed of evidence accumulation, and the width of decision thresholds.
Results. Both groups learned to quickly choose the high-reward image in each pair. The RLDDM did not reveal group differences in decision parameters, but it did estimate slower learning rates following rewards in MDD vs. HC (posterior probability of MDD < HC: 85.24%).
Conclusions. The RLDDM revealed slower learning following rewards in depressed adults vs. healthy controls. Although was modest, this result supports the hypothesis that anhedonic depression may impair reinforcement learning by disrupting RPEs. Moreover, this work demonstrates that, by modeling learning and decision-making simultaneously, the RLDDM provides a sensitive assessment of the negative impact of depression on behavior. In subsequent analyses, we will integrate these modeling results with functional neuroimaging.
I was accepted to work with a Lafayette econometrics professor, Dr. Adam Biener, conducting a capstone project on the economic cost of mental health resulting from the 2008 financial crisis (Paper available for download on the right). This topic is something the Fed has been investigating, and my professor’s and I sought to conduct further analysis with Python coding and clinical data e to complement his economic perspective using Fed data to publish a paper on the true economic costs of MDD resulting from the crisis.
Abstract. Poor mental health, and the symptoms associated with it, are well-established factors associated with significant economic burdens to the U.S. Additionally, numerous studies have shown that macroeconomic changes affect the quality of an individual’s mental health significantly. What remains unclear, however, is how outcomes of poor mental health can vary across sex and race/ethnicity during a negative macro-economic change such as the Great Recession of 2008. I used data from MEPS to estimate the effect of the recession on various indicators on various effects of mental health. I find that there are noticeable effects across gender and race/ethnicity on both economic and mental health-related indicator outcomes during the Great Recession.
BlockPour is a startup based out of Singapore which brings real-time decentralized data across multiple networks for you to view, compare, and track token data.
Hired as a data scientist, I am currently applying quantitative analyses to identify potential premium paid API features for traders of cryptocurrency.
Additionally, I have been applying Machine Learning algorithms to pricing models of archived and real-time data from decentralized exchanges. In this light, I have been employing factor modeling and general regression analysis on time series data to identify potential trading indicators.
Furthermore, I have been utilizing my consulting experience to built out a go-to-market licensee pricing model for API data.