MSL Seminar: Local Invited Speaker, Dr. Tim Murphy

Local Invited Speaker:
Dr. Tim Murphy
Department of Psychiatry and Djavad Mowafaghian Centre for Brain Health
Faculty of Medicine, UBC

Host: Dr. Xin Tang

Off-cycle from our regular monthly seminars, we have invited local researchers to give talks at the MSL to open up opportunities for interesting research discussions and collaborations. We are excited to welcome Dr. Murphy as one of these local speakers for the 2025-26 seminar cycle.

This seminar will be presented in a hybrid format. The talk will be delivered in person at the MSL Lecture Theatre (room 102). Audience members are welcome to attend either in person or via the zoom link. Those connecting via zoom will be able to ask questions during the Q&A portion using the chat function.

Zoom registration link: https://ubc.zoom.us/meeting/register/8oO0BGETSveIV__8W0s1gQ  

Talk title: Augmenting Data Analysis in Neuroscience with Synthetic Mice and Patients

Abstract: We employ generative AI to reshape how we study behavior, brain circuits, and clinical psychiatry. I will introduce an integrated set of tools spanning animal and human research. First, I present a generative pose-estimation framework that combines synthetic mouse animations, depth-estimation models, and Video-LLMs (large language models) to produce quantitative behavioral imaging with improved accuracy, modality fusion, and both inferred and calibrated 3D coordinates. These tools enhance automated behavioral tracking, dataset augmentation, and cross-camera generalization for parallel mesoscale and two-photon imaging studies, forms cellular and regional activity imaging.  We have created a common synthetic mouse body capable of amalgamating behavioral data across animals and experimental systems.  I then describe our synthetic-patient transcript engine, which uses LLM interviewer–patient loops to generate diverse, demographically balanced psychiatric dialogues for model benchmarking, safety evaluation, and privacy-preserving clinical AI research. Together, these generative approaches provide a scalable framework that links behavior, neural activity, and mental-health applications, demonstrating how AI systems can augment datasets supporting reproducible neuroscience and next-generation psychiatric research.  Such data analysis approaches also highlight the multimodal capability of current generative tools and are likely transferrable to other areas of biology and will help overcome low-data regimes.