Parker Singleton

@parkersingleton.bsky.social

1182 Followers 228 Following

Senior Scientist at Penn Lifespan Informatics and Neuroimaging Center (PennLINC) studying psychedelics and the brain. sypres.io

Forrest plot of main meta-analytic model.
Screen shot of psilocybin for depression study page on sypres.io
Image of Gartner Hype Cycle
screenshot of the dashboard home page
screenshot of the GitHub repository readme for the database
Forest plot of effect sizes.
screenshot of the psilocybin for depression section on sypres.io
Image of the Gartner hype cycle
SYPRES (Synthesis of Psychedelic Research Studies) logo
scheme of simulation paradigm and simulated vs empirical traces overlaid
dominance analysis radar plot showing 2a most dominate for all 3 regional metrics
three scatterplots of regional CE metrics vs 2a density
CE over time plotted alongside LZ and intensity
traces of CE for each network under DMT and PCB
traces of global CE under DMT and PCB
schema of control theory analysis
fMRI-EEG recorded for 8 mins before and 20 mins after injection of DMT/placebo
SYPRES LOGO: a cartoonish drawing of a tree with structures resembling a brain/neurons.
A plot of the Gartner hype cycle
bag of shiitake mycelium from mushroom mountain
Abstract: As machine learning algorithms in healthcare transition from research into deployment, they face a constantly evolving environment, rife with changing clinical practices, data collection policies, patient populations, and even diseases themselves. Models that performed well in the past are liable to fail in the future, and especially in such high-stakes settings as healthcare, complacency can have consequences. In this talk, we start by empirically characterizing real-world shifts over time in medical data by examining model performance using a deployment-oriented evaluation framework (EMDOT). Inspired by the concept of backtesting, EMDOT simulates possible training procedures that practitioners might have been able to execute at each point in time, and evaluates the resulting models on all future time points. Across six distinct sources of medical data, we find varying levels of performance improvement and degradation, and we inspect surprising jumps in performance over time