Raha Dastgheyb, PhD

Co-Director, JH-CAHN Biomarker Core

Assistant Professor of Neurology

Brain Health Program

• Data-driven phenotyping & machine-learning for modeling cross-sectional and longitudinal cognition in NeuroHIV, aging, and long-COVID
• Multi-omics + advanced neuroimaging integration to discover and validate biomarkers of neurodegeneration
• Sex- and hormone-specific brain health across the lifespan, with a focus on women living with HIV
• Sleep analytics & chronobiology (actigraphy, polysomnography, wearables, self-report) as modifiable determinants of neurocognition
• Digital phenotyping via tablet-based neuropsychological testing and wearable data streams
• High-throughput electrophysiology analytics for neuronal networks (multi-electrode arrays, organoids, brain slices)
• Translational biomarker pipelines for discovery science
• User-friendly software design & graphical interfaces (e.g., MEAnalyzer GUI, SciDataReportR) for reproducible, open-source neuroscience

My group develops open-source data-science frameworks that make complex analytic methods reproducible and clinician-friendly. SciDataReportR, our flagship R package, streamlines the entire workflow for both hypothesis-led and hypothesis-generating projects - from data cleaning and metadata checks to publication-ready tables, figures, and dynamic reports.

A second pillar of my program is high-throughput electrophysiology. I co-created MEAnalyzer, a point-and-click graphical interface that wraps advanced spike-train algorithms in an intuitive dashboard. Investigators can upload raw multi-electrode-array recordings, run burst detection or functional-connectivity metrics with one click, and export a complete provenance file for transparent auditing or re-analysis.

I believe that data only become knowledge when people can see them. My team builds interactive visualizations (shiny apps, plotly dashboards, animated GIFs) that let users rotate multidimensional results, drill down to individual observations, and share insights across disciplines.  These tools lower the barrier to entry for collaborators with diverse backgrounds, ensuring that every angle of the data is explored and every discovery is both interpretable and accessible.

In the HIV neurocognition sphere with the Brain Health Program and Dr. Leah Rubin, my group has pioneered unsupervised-learning pipelines that reveal hidden cognitive phenotypes and their drivers. Using self-organizing maps coupled with hierarchical clustering, we parsed neuropsychological data from 929 virally suppressed women in WIHS and resolved nine reproducible cognitive profiles whose membership was shaped by modifiable factors such as depression, CD4 nadir and antiretroviral regimen. We then scaled this framework to a sex-comparative study, showing that women and men display distinct profile architectures yet share a common set of biological and psychosocial predictors identified via random-forest models.

Complementing these cognitive-phenotyping efforts, we have established a sleep-analytics platform in collaboration with the MWCCS, Dr. Audrey French, and Kathleen Weber that combines actigraphy, polysomnography summaries, and patient-reported outcomes to study sleep as a modifiable determinant of neurocognition. These analyses leverage our reproducible clustering and machine-learning pipelines, enabling rapid sensitivity checks across sleep metrics and demographic strata.