Resources
Talks
Upcoming
-
Transformers and Language Models Explained
AI in Dementia Data (AIDD) Series, webinar “Generative AI: Diffusion and Transformers” · Online
-
Center for Neural Data Science
Stanford Center for Neural Data Science, Stanford Data Science and Wu Tsai Neurosciences Institute · Stanford, California
-
Mining Real-World Clinical Data with AI to Identify Symptom Patterns and Treatment Signals in FTD
AFTD 2027 Holloway Summit: Research to Advance Symptom Management in FTD · Orlando, Florida
-
AI Advances in the Clinic
Keystone Symposia: Neurodegeneration: From Mechanisms to Precision Diagnosis and Therapies · Whistler, British Columbia, Canada
2026
-
Clinical Readiness: AI Guided Dx
Tau Consortium Investigators' Meeting (TCIM 2026) · Irving, Texas
-
AI in Neurology
UCSF Department of Neurology Grand Rounds, with Eddy Amorim, MD, and Jon Kleen, MD, PhD · San Francisco
-
Scaling the Frontier: Large Models for Health
Frontiers in Computational Precision Health (FiCPH 2026), panel · San Francisco
2025
-
Agentic AI for Neurodegenerative Disease Diagnosis
5th Annual UCSF Dementia Day · San Francisco
-
Artificial Intelligence in Alzheimer's Disease Clinical Trials
CTAD 2025 Task Force · San Diego, California
-
Spatiotemporal Dynamics of Arithmetic Computation in the Human Brain
Collège de France symposium “Seeing the Mind, Educating the Brain” · Paris, France
2024
-
Memory & Aging Center Co-pilot: Advancing Neurodegenerative Disease Diagnosis with Generative Artificial Intelligence
UCSF Center for Intelligent Imaging (ci2) · San Francisco
Teaching
-
Voxel-wise grey matter and its normative model
deepmriprep preprocessing (brain extraction, tissue segmentation, registration, modulation, smoothing) and a voxel-wise normative model (A), step by step on one case.
Slides -
Regional brain volumes and their normative models
FastSurfer segmentation, a regional normative model (B) and a dedicated model for T1 white-matter hypointensities (C), step by step on one case.
Slides -
FLAIR white matter: lesions, regions and normal-appearing tissue
One lesion segmentation (LST-AI) and one white-matter parcellation (the bullseye) feeding three normative models: total lesion volume (D1), lesion volume per region (D2) and the signal of normal-appearing white matter (D3).
Slides
-
Introduction to using AI tools at UCSF
A practical introduction to agentic AI for researchers: what it is, the PHI-safe tools available at UCSF (ChatGPT Enterprise and the Versa API), and worked examples with skills, Codex, and MCP connectors. Presented to the UCSF Brain Health Registry by Pedro Pinheiro-Chagas.
Slides