Welcome to the Lin Lab (Bytes of Minds Lab)! We
are an interdisciplinary research group at the Icahn School of Medicine at
Mount Sinai in the Departments of AI, psychiatry, and neuroscience. Based
within the #1 hospital and the biggest hospital network in New York, we aim to
push the boundaries of Computational Psychiatry
and Neuro-AI by creating adaptive, intelligent
systems powered by multi-modal
foundation models
such as large language models that can
eventually both deepen our understanding of the mind and enhance mental health
care in a scalable way.
Our Mission
The human brain is a powerful organ, enabling us to think, feel, and interact in
complex ways. Yet, its workings and the nature of the mind remain partly mysterious.
Our mission is to unravel these complexities by developing AI inspired by the brain's
structure and function in social setting. We aim to create autonomous agent-based
systems and foundation models of the brain that learn, adapt, and reason in ways
that reflect the human mind, ultimately solving real-world challenges, interacting
naturally with humans, and supporting mental health through advanced diagnostic
and therapeutic tools.
Key Research Areas
Ongoing Research Directions
1. Computational cognitive models and neural dynamics focused on perception,
memory, and social learning
2. Neuro-inspired and neurosymbolic AI architectures for adaptive learning
3. Real-time, multi-modal, and autonomous AI interfaces for clinical care,
including AI companions, large language models, and psychometric assessments
4. Brain-computer interfaces, extended realities, and embodied cognition
5. Privacy-preserving AI solutions and societal impacts for secure
healthcare applications
What we have built
Our work runs in production, not only in papers. We deployed an LLM voice
agent for patient health coaching inside a major US hospital network,
through IRB, clinical governance and security review. We co-direct the
Therapeutic Alliance AI core of the $20M NIH IMPACT-MH consortium across 6
sites, covering 2,000+ patients over 5 years. Our methods work is published
in
PNAS, eLife, Nature Translational Psychiatry, and
presented at AI conferences such as NeurIPS, AAAI, IJCAI, ICML, and AAMAS,
and our code is open source on
GitHub.
The lab's books are
Privacy & Security in Large Language Models
(O'Reilly, 2025) and
Reinforcement Learning Methods in Speech and Language Technology
(Springer, 2024).
If you want to work with us, or on something adjacent, get in touch.
The publications can be grouped
or
(Books, Manuscripts or preprints, Journal articles,
Conference proceedings,
Extend abstracts and Patents). For other insights, please
refer to our
Google Scholar
page.