Multimodal Mental Health Data Scientist
I build computational tools that bridge genetics, neuroimaging, and behavior to unlock insights for neurodiversity and psychiatric research. Currently finishing my PhD at University of Iowa.
About Me
My passion: I combine genetics, neuroimaging, and language analysis to understand mental health at the intersection of biology and behavior. As an autistic researcher, I bring lived experience to understanding neurodiversity—and I'm building tools that actually serve this community.
My research integrates diverse data modalities—genetic, clinical, neuroimaging, audio, interview transcripts, and facial imagery to uncover patterns that bridge laboratory discovery with real clinical interventions.
- Clinical informatics / health data science roles in psychiatry, neurology, or genomics
- Computational genetics / statistical genetics positions with large-scale cohorts
- Data science roles in biotech, health tech, or research institutes
- Positions that value reproducible science and open-source tools
I'm wrapping up my PhD in April 2026 and actively networking. Let's chat about opportunities!
Expertise & Tools
My work spans multimodal data analysis, reproducible pipeline development, and computational tool building for psychiatric research.
Neuroimaging
7T structural MRI processing, fMRI analysis, DTI, cortical reconstruction, surface-based morphometry
freesurfer · ANTs · FSL · AFNI · nilearn
Genomics
GWAS, eQTL analysis, transcriptomics, single-nuclei multi-omics, expression-QTL integration
R · Python · TWAS · Seurat · edgeR
NLP & Behavior
Acoustic feature extraction, interview transcription, sentiment analysis, linguistic biomarkers, emotion detection
WhisperAI · GPT · lingmatch · topic modeling
Reproducible Science
End-to-end computational pipelines, QC workflows, interactive visualizations, open-source development
Git · Bash · R Shiny · Docker
Statistical Analysis
Mixed-effects modeling, machine learning, deep learning for phenotype prediction, cross-species validation
lmmSeq · scikit-learn · TensorFlow
Computer Vision
Facial landmark detection, morphometric analysis, automated QC on images
OpenCV · facial recognition · morphometry
Featured Research Projects
Gene Expression Signature of Human Brain Stimulation
Key Findings: Identified cell-type-specific genes upregulated in response to electrical stimulation.
- Engineered end-to-end pipeline for single-nuclei multi-omics (RNA+ATAC) data with bootstrapped pseudo-bulk strategy
- Applied mixed-effects models for robust cell-type-specific detection
- Cross-species validation using RRHO identified conserved gene sets
Exceptional Ability: Multimodal Cognitive Study
Key Insight: Used a 10-minute language task that effectively captures cognitive performance as a digital biomarker, demonstrating potential for scalable assessment.
- Integrated NIH-Toolbox scores, custom language task, acoustic features, interview transcription, facial landmarks, and multi-modal MRI
- Applied WhisperAI for automated transcription and GPT embeddings for linguistic analysis
- Built reproducible QC and visualization pipelines
Polygenic Drug Response Signatures
Impact: Scaled analysis to 88,000+ participants (SPARK, ABCD) with validated predictions against behavioral phenotypes and neuroimaging biomarkers.
- Integrated GWAS, eQTL, and RNA-Seq to generate personalized treatment recommendations for psychiatric disorders (ADHD focus)
- Cross-validated polygenic scores against CBCL, BPM behavioral scales and fMRI data
Brain-Wide Drug Effects: Deep Learning Prediction
Deliverable: Interactive R Shiny application mapping 838 compounds to predicted functional brain changes for phenotype discovery.
- Integrated brain-wide gene expression (Allen Institute), fMRI trait maps, and drug perturbation signatures (CMAP, LINCS)
- Built deep learning model for cross-dataset drug effect prediction
- Deployed interactive 3D brain visualizations linked to Neurosynth and Neuromaps phenotypes
Linguistic & Behavioral Patterns in Bipolar Disorder (Social Media)
Discovery: Identified distinct behavioral signatures including disrupted sleep patterns, content topic shifts, and cyclical emotional patterns aligned with mood episodes.
- Analyzed 20 years of Reddit data identifying 2,847 self-disclosed bipolar users
- Extracted temporal posting patterns, sentiment, topic shifts, and GPT-4 embeddings for emotional volatility tracking
- Validated linguistic markers against reported mood episode timelines
Selected Talks & Presentations
Beyond Yes/No: Multimodal Autism Propensity Score from Genes to Brain
INSAR Conference 2025 | Oral Presentation
Presented novel deep learning framework integrating multi-modal neuroimaging (fALFF, structural morphometry, DTI) to generate continuous autism likelihood scores. Demonstrates potential of combining MRI modalities for improved autism neurophenotyping.
Optimizing Structural MRI Processing Pipelines for 7T Data
Iowa Neuroimaging Consortium, University of Iowa | Invited Talk
Comprehensive overview of 7T structural MRI processing pipelines, comparing tools for cortical reconstruction, subcortical segmentation, and surface-based analysis. Provided practical guidance on pipeline selection based on research objectives.
Visualization Gallery
Click any visualization to view in full size. Gallery organized by research domain.
Media, Publications & Links
Learn more about my research, background, and lab environment: