Research
I develop statistical and machine learning methods for complex data from psychiatry, neuroscience, digital health and biology. Explore the themes below to see the work.
Decision-making dynamics and reinforcement learning
How do people learn from reward, and how does that differ from one person to the next? I build statistical and machine learning models that recover the latent cognitive states behind choices in behavioral tasks and connect them to brain activity such as EEG. The aim is a personalized description of decision-making in mental disorders.
Published & forthcoming
-
Biostatistics2025
HMM for Discovering Decision-Making Dynamics Using Reinforcement Learning Experiments
-
Ann. Appl. Stat.2025
Joint Modeling for Learning Decision-Making Dynamics in Behavioral Experiments
-
Biometrics2024
A Hierarchical Random Effects State-space Model for Modeling Brain Activities from Electroencephalogram Data
-
JASA2024
A Semiparametric Inverse Reinforcement Learning Approach to Characterize Decision Making for Mental Disorders
Under review
-
Preprint
Anchored Variational Inference for Personalized Sequential Latent-State Models
-
Preprint
Shared Hidden-factor Information Framework for Multiple Behavioral Tasks
High-dimensional functional data
When every predictor is a curve and there are many of them, which ones matter? This line builds theory and methods for regression with high-dimensional functional predictors: selecting the relevant curves, predicting with minimax guarantees, and identifying the right form of the model.
Published & forthcoming
-
Statistica Sinicato appear
Variable Selection and Minimax Prediction in High-dimensional Functional Linear Models
Under review
-
Preprint
Model Form Identification in High-Dimensional Functional Linear Regressions
Precision medicine and digital phenotyping
Phones and wearables now record health day by day. I develop methods that turn longitudinal, biomarker and digital health data into individualized treatment strategies: estimating the sequential effects of interventions and learning biomarker-guided policies for chronic disorders.
Published & forthcoming
-
Stat. Med.2024
Learning Optimal Biomarker-Guided Treatment Policy for Chronic Disorders
Under review
-
Preprint
Cumulative Marginal Mean Model for Assessing Sequential Effects Using Digital Health Data
Imaging and spatial omics
Computational tools that pull biological signal out of images and high-throughput measurements: segmenting overlapping plants in field photographs, reconstructing spatial omics data at higher resolution, and profiling metabolomic variation.
Published & forthcoming
-
Plant Phenomics2023
High-Throughput Field Plant Phenotyping: A Self-Supervised Sequential CNN Method to Segment Overlapping Plants
-
Plant Phenomics2021
KAT4IA: K-means assisted training for image analysis of field-grown plant phenotypes
Under review
-
Preprint
Trustworthy super-resolution reconstruction across spatial omics modalities