MIND Data Science Lab

Overview

The MIND Data Science Lab studies neurological disorders across biological, clinical, and population scales, with a primary focus on Alzheimer’s disease and related dementias (ADRD). Our goal is to elucidate disease mechanisms and improve early detection, risk prediction, and differential diagnosis.

We develop and apply statistical, bioinformatics, and AI approaches to integrate diverse data types, including multi-omics, digital pathology, fluid biomarkers, neuroimaging, and electronic health records. Our work focuses on building scalable, interpretable, and clinically grounded models that advance understanding of disease mechanisms, enable real-world cognitive phenotyping, and accelerate the discovery of novel therapeutic strategies.

Research Projects

1. AI modeling for real-world cognitive phenotyping

We develop AI models, agents, and LLM systems that leverage EHR to identify cognitive impairment, support differential diagnosis, characterize disease heterogeneity, and evaluate emerging treatments for ADRD. Our work uses real-world clinical data to enable scalable cognitive phenotyping, early detection, and disease staging.

Representative papers

• Leng Y, et al. A GPT-4o-powered framework for identifying cognitive impairment stages in electronic health records. npj Digital Medicine, 2025.
• Cheng Y, et al. High-throughput phenotyping of the symptoms of Alzheimer disease and related dementias using large language models: cross-sectional study. JMIR AI, 2025.
• West, Matthew, et al. Unsupervised deep learning of electronic health records to characterize heterogeneity across Alzheimer disease and related dementias: cross-sectional study. JMIR Aging, 2025.

2. Cellular and Spatial Omics of Alzheimer’s Disease and Other Brain Disorders

We use spatial biology, and integrative omics approaches to investigate how pathological features such as plaques, tangles, and genetic risk factors (e.g., APOE genotype) reshape the brain at cellular and regional resolution. A major focus of our work is understanding how astrocytes, microglia, and other non-neuronal cells respond to pathology and how these responses influence neuronal vulnerability and disease progression. We also investigate mechanisms across neurological disorders including Alzheimer’s disease, tauopathies, Parkinson’s disease, Huntington’s disease, and myotonic dystrophy (DM1).

Representative papers

• Kumari P, et al. Analysis of human urinary extracellular vesicles reveals disordered renal metabolism in myotonic dystrophy type 1. Nature Communications, 2025.
• Serrano-Pozo A, et al. Astrocyte transcriptomic changes along the spatiotemporal progression of Alzheimer's disease. Nature Neuroscience, 2024.
• Das S, et al. Distinct transcriptomic responses to Aβ plaques, neurofibrillary tangles, and APOE genotype in Alzheimer’s disease. Alzheimer’s & Dementia, 2024.
• Agra Almeida Quadros AR, et al. Cryptic splicing of stathmin-2 and UNC13A mRNAs is a pathological hallmark of TDP-43-associated Alzheimer's disease. Acta Neuropathologica, 2024.

3. Open Science Platforms, Models, and Agents for Neurobiology

We develop computational infrastructure that makes complex biomedical data more accessible, interoperable, and reusable. We build open science platforms, models, and AI agents that enable researchers to integrate and analyze large-scale biomedical datasets, transforming them into actionable scientific insights.

DataLENS

DataLENS is an open data analytics portal that enables researchers to explore, visualize, and share bioinformatics analyses of Alzheimer’s disease omics datasets. It processes public datasets using standardized pipelines and provides intuitive web interfaces to query and integrate results across multiple biological modalities. https://alzdatalens.partners.org/

Alz-TXPILOT

Alz-TXPILOT is an agentic AI system for drug target prioritization and repurposing in Alzheimer's disease. It uses an AI planning framework to iteratively test hypotheses across multimodal Alzheimer’s data, analytical tools, and scientific literature, providing a live interface with transparent reasoning traces. https://alz-agent.netlify.app

Alz-GNN

Alz-GNN is a multiscale graph neural network that integrates brain omics data and protein interaction networks to model cell-type–specific molecular mechanisms in Alzheimer’s disease.

Representative papers

• Chauhan et al., Multi Scale Graph Neural Network for Alzheimer's Disease, 2024.
• Noori et al., Alzheimer DataLENS: An Open Data Analytics Portal for Alzheimer's Disease Research. Journal of Alzheimer's Disease, 2024.
• Bihlmeyer et al., Novel methods for integration and visualization of genomics and genetics data in Alzheimer's disease. Alzheimer’s & Dementia, 2019.

4. Digital health, wearables, and fluid biomarker for longitudinal monitoring of aging and dementia

We develop scalable approaches to monitor health, behavior, and cognition across the lifespan by integrating digital signals from wearables, sensors, and speech with fluid biomarkers from blood, plasma, and cerebrospinal fluid. Our goal is to identify accessible markers that enable earlier detection, risk stratification, and longitudinal tracking of cognitive decline and dementia.

Representative papers

• He et al., A predictive model for cognitive decline using social determinants of health, JAR Life, 2026.
• Ford et al., Using Apple Watches to Monitor Health and Behaviors of Individuals with Cognitive Impairment: A Case Series Study, The Journals of Gerontology, 2025.
• Favaro et al., Toward Inclusive Large-Scale Alzheimer's Disease Detection via Speech and Language Modeling, Annu Int Conf IEEE Eng Med Biol Soc, 2025.
• Kivisäkk et al., Plasma biomarkers for prognosis of cognitive decline in patients with mild cognitive impairment, Brain Communications, 2022.

5. Drug repurposing, causal inference, and treatment heterogeneity

We combine observational data, causal inference, and machine learning to identify candidate therapies, estimate treatment effects, and understand who benefits most from interventions.

Representative papers

• Weinberg et al., Association of BCG vaccine treatment with death and dementia in patients with non-muscle-invasive bladder cancer, JAMA Network Open, 2023.
• Vakulenko-Lagun et al., causalCmprsk: An R package for nonparametric and Cox-based estimation of average treatment effects in competing risks data. Comput Methods Programs Biomed, 2023.
• Charpignon et al., Causal inference in medical records and complementary systems pharmacology for metformin drug repurposing towards dementia, Nature Communications, 2022.

Research Team

Sudeshna Das, Ph.D.
Principal Investigator
Associate Professor, Massachusetts General Hospital (MGH) and Harvard Medical School (HMS)

Sudeshna Das, PhD, is an Associate Professor of Neurology at MGH and Harvard Medical School. She develops AI, bioinformatics, and statistical methods to analyze omics, clinical, and real-world data to advance research in Alzheimer’s disease and related dementias. Harvard Catalyst Profile

Anna Favaro, Ph.D.
Research Fellow

Anna Favaro is a Research Fellow with a PhD in Electrical and Computer Engineering from Johns Hopkins. She develops AI methods for biomedical signal analysis, serves as Treasurer of the Harvard Medical Postdoc Association, and enjoys philosophy. Email: afavaro@mgh.harvard.edu

Huan Li, Ph.D.
Research Fellow

Huan Li is a Research Fellow with a PhD in Biology from Tsinghua University. She uses multi-omics data to understand mechanisms of Alzheimer's disease. Outside the lab, she enjoys musical theatre and lives in a tiny apartment under the strict supervision of her cat. Email: huli1@mgh.harvard.edu

Beata-Gabriela K. Simpson
Data Manager

Beata-Gabriela Simpson is a data analyst who earned her MPH at Emory University. When she is not analyzing data for the Mass. Alzheimer's Disease Research Center (MADRC), she utilizes time-tested algorithms to turn yarn into wearable works of art. Email: bgsimpson@mgh.harvard.edu

Akhila Gundavelli
Bioinformatics Analyst

Akhila Gundavelli is a Bioinformatics Analyst with an MS from Boston University. She uses single-cell data to study cellular and molecular changes in Alzheimer’s disease. In her free time, she makes friends taste-test her vegan cooking experiments. Email: agundavelli@mgh.harvard.edu

Yingnan He
Data Scientist

Yingnan holds MS degrees in Health Policy and Management and Data Engineering Analytics. She develops and evaluates AI models using EHR data to improve early detection and phenotyping of ADRD. She is also highly trained in video games. Email: yihe1@mgh.harvard.edu

Swarnali Dasgupta
Bioinformatician

Swarnali Dasgupta is a bioinformatician with an MS from Northeastern University. She studies Alzheimer's disease by day, and by night she explores the city, builds Legos, and is always on the hunt for the next whodunit to read or watch. Email: sdasgupta5@mgh.harvard.edu

Yu Leng
Data Scientist

Yu Leng is a data scientist with an MS in Biomedical Informatics from Harvard developing AI methods for multimodal biomedical data, and a classical Chinese and ballet dancer. She also curates an “olfactory dataset” of 200+ perfume samples—enough for her own “precision fragrance” system. Email: yleng2@mgh.harvard.edu

Giovanna Carello Collar
MSc, Research Scholar (Fulbright)

Giovanna Carello-Collar is a Fulbright Scholar with a Master’s in Biochemistry from UFRGS, Brazil. She investigates why some brains resist AD, while her dog Rafa (her personal reviewer 2) provides critical feedback, mainly in exchange for treats. Email: gcarellocollar@mgh.harvard.edu

Publications

View publications