Rahul Biswas

Postdoctoral Scholar · Department of Neurology · Weill Institute for Neurosciences · University of California, San Francisco
Research Data Scientist · San Francisco VA Health Care System

prof_pic.jpg

I develop causal and interpretable machine learning methods for neuroscience and medicine. I design these methods around the structure of each problem. In neuroscience this means modeling directed neural interactions that act across several time lags, and accounting for the other recorded neurons so that indirect influences are not mistaken for direct ones. In medicine it means deciding with clinicians what a model should assess and what it must handle, such as the technique errors that distort blood pressure readings, or the artifacts in everyday skin photos. I also keep assumptions light, using nonparametric tests where possible, and computation fast, using GPU acceleration, so the methods scale to large datasets and can support real-time use.

These methods produce explanations, such as which neurons drive which, or which cues drive an AI system’s decision. I then test them against independent evidence. For example, I compare connections inferred from neural activity with synaptic wiring measured by electron microscopy and, in ongoing collaborations, with the results of perturbation experiments. In medicine I check that an AI system relies on cues clinicians would assess and stays accurate under real-world conditions.

Computational Neuroscience. Applied to more than 57,000 neurons in mouse visual cortex, my method CITS inferred connections that aligned with the synaptic wiring in the electron-microscopy connectome far more closely than Granger causality or lagged correlation, and revealed how this circuitry reorganizes with arousal. Current projects include directed network dynamics of chronic pain with Dr. Prasad Shirvalkar, and AI-based reconstruction of visual experience from fMRI with the Gallant Lab (UC Berkeley).

Computational Clinical Diagnostics. With Dr. Julio Lamprea (UCSF Cardiovascular Prevention Center), I developed Coach-BP, a computer vision system that detects errors in blood pressure measurement technique from clinic video. With Dr. Maria Wei (UCSF Dermatology and SF VA), I am developing skin-lesion classifiers that stay accurate when photographs contain common artifacts.

Before joining Dr. Reza Abbasi-Asl’s lab at UCSF, I completed my PhD in Electrical and Computer Engineering (with a secondary MS in Statistics) at the University of Washington, advised by Dr. Eli Shlizerman, with dissertation Causal Functional Connectivity from Neural Dynamics. Before that, I earned a BS and MS in Statistics at the Indian Statistical Institute, Kolkata.

I also maintain open-source implementations of my methods, including CITS and TimeAwarePC.

Feel free to reach out at rahul.biswas@ucsf.edu.

Updates

Oct 2026 New preprint: Coach-BP: Computer Vision-Based Assessment and Correction of Blood Pressure Measurement Technique.
May 2026 Presenting at CCN 2026 (New York University) and CNCM 2026 (UC Irvine): State-Dependent Organization of Microscale Functional Circuitry in Visual Cortex. The CCN poster was additionally selected as a Spotlight.
Mar 2026 New preprint: State-Dependent Organization of Microscale Functional Circuitry in Visual Cortex.
Feb 2026 US Provisional Patent Application filed: “Methods and Systems for Assessing Blood Pressure Measurement Technique”, assigned to The Regents of the University of California.
Jun 2025 Visiting seminars at the University of Sydney, UNSW, and Macquarie University, Australia.
Jun 2025 Presented at the Organization for Human Brain Mapping Annual Meeting (OHBM 2025), Brisbane, Australia.
Jan 2025 New preprint: CITS: Nonparametric Statistical Causal Modeling for High-Resolution Neural Time Series.
Oct 2024 Visiting seminars at the National University of Singapore and Indian Institute of Science, Bangalore.
Jul 2024 Started as a Postdoctoral Scholar in the Department of Neurology at UCSF, working with Dr. Reza Abbasi-Asl.
Jun 2024 Successfully defended my PhD dissertation, “Causal Functional Connectivity from Neural Dynamics”, at the University of Washington.

Selected Publications

  1. medRxiv
    Coach-BP: Computer Vision-Based Assessment and Correction of Blood Pressure Measurement Technique
    Rahul Biswas, Alvin Joseph Abad, Jesuit Nugas, and 4 more authors
    2026
    Preprint
  2. bioRxiv
    State-Dependent Organization of Microscale Functional Circuitry in Visual Cortex
    Rahul Biswas, Hasika Wickrama Senevirathne, Yujing Wang, and 3 more authors
    2026
    Preprint
  3. arXiv
    CITS: Nonparametric Statistical Causal Modeling for High-Resolution Neural Time Series
    Rahul Biswas, SuryaNarayana Sripada, Somabha Mukherjee, and 1 more author
    2025
    Preprint
  4. Front. Comp. Neuro.
    Causal Functional Connectivity in Alzheimer’s Disease Computed from Time Series fMRI Data
    Rahul Biswas and SuryaNarayana Sripada
    Frontiers in Computational Neuroscience, 2023
  5. Stat. Comput.
    Consistent Causal Inference from Time Series with PC Algorithm and its Time-Aware Extension
    Rahul Biswas and Somabha Mukherjee
    Statistics and Computing, 2024
  6. PLoS Comput. Biol.
    Statistical Perspective on Functional and Causal Neural Connectomics: The Time-Aware PC Algorithm
    Rahul Biswas and Eli Shlizerman
    PLOS Computational Biology, 2022