Ben Deverett, MD, PhD

AI researcher at a healthtech startup, engineering and evaluating new AI tools to enhance healthcare delivery.

Building and validating foundation models to unlock the predictive power of intraoperative data and guide clinical decisions (manuscript in revision).

Characterizing how the cognitive strategies of AI agents compare to those observed in neuroscience (at Google DeepMind), published and presented at NeurIPS.

Applying insights from neuroscience and clinical medicine to prepare for safe and ethical interactions with non-traditional consciousness.

Bayesian modeling of human inference strategies for inferring causality between observed events.

Quantifying how animals retain information in memory and use it to guide decisions, using measurement and perturbation of neural networks during learning and behavior.

Developing software and hardware to perform closed-loop behavioral, pharmacologic, and EEG-based human studies.

Noninvasively modulating brain circuits with high spatial precision to enhance the way medications influence the brain (study ongoing, ask me for details).

Running clinical studies to test therapeutic applications of psychedelics and anesthetic medications for new purposes (trials ongoing).

Identifying risk factors for neurologic injury in patients receiving cardiac mechanical support.

Demonstrating that distinct cortical signatures of ketamine dissociate under wakefulness and anesthesia.

Studying neuronal encoding rules using cellular-resolution two-photon imaging of the rodent brain.

Educating Princeton students in programming and math; and Stanford residents on intravenous anesthetics and neuroanatomy.

Collaborating with under-resourced medical centers in Guyana, Tanzania, Vietnam, and Rwanda to share clinical and research skills.

Using multi-institutional operating room data to uncover how anesthesiologist habits influence patient outcomes (in press).

Open-source Python library for high-speed video capture using PlayStation Eye cameras.

Reimagining how we interact with keyboards in iOS to improve efficiency on touchscreens.

Using Amazon Mechanical Turk to study how we can represent measures of centrality in visual graph representations.

Arduino-based MIDI musical instrument based on frontal EEG signals

Numerous original musical production projects including for an award-winning musical.