←
Gabriel Loewinger45 posts
Gabriel Loewinger avatar

Gabriel Loewinger

@GLoewinger

Machine Learning Research Scientist, NIMH PhD Biostatistics, Harvard '22. Opinions are my own, not my employer's.

◷ Joined December 2019▧ 25 Photos and videos
NIH Innovates avatar
⟳ Gabriel Loewinger retweeted
NIH Innovates@NIH_Innovates·18 Mar 2025

NIH researchers developed a powerful method to track how brain-behavior relationships evolve, revealing insights that standard analyses miss. This could reshape how we study neural activity in psychiatric disorders and addiction. elifesciences.org/articles/9…

Gabriel Loewinger avatar
Gabriel Loewinger@GLoewinger·18 Mar 2025

FLMM finds effects obscured by standard analyses! For example, FLMM reveals effects that “wash out” when analyzed with AUCs. In published work, Cue Period AUC finds no effects because it averages over time- windows (1) and (2) that have opposing effects. 12/13

Post attachment
Gabriel Loewinger avatar
Gabriel Loewinger@GLoewinger·18 Mar 2025

FLMM can disentangle components with distinct temporal dynamics. It can also be used to run analogues of standard hypothesis tests (e.g., ANOVAs, correlations) at each trial time-point. Below is an example akin to the FLMM version of a paired t-test. 11/13

Post attachment
Gabriel Loewinger avatar
Gabriel Loewinger@GLoewinger·18 Mar 2025

FLMM plots can be conceptualized as pooling signal values (dF/F) at a given trial time-point (e.g., 1.7 sec) across animals and trials, correlating it with covariate(s) (e.g., Latency-to-press) and plotting the slope of the correlation. 8/13

Post attachment
Gabriel Loewinger avatar
Gabriel Loewinger@GLoewinger·18 Mar 2025

FLMM outputs a coefficient estimate plot that shows how the signal–covariate association evolves across trial time-points. 7/13

Post attachment
Gabriel Loewinger avatar
Gabriel Loewinger@GLoewinger·18 Mar 2025

FLMMs exploit autocorrelation to construct *joint* 95% CIs (light grey) that show time windows where effects are statistically significant (any intervals that do not contain 0). All you need to do is visually inspect! 6/13

Post attachment
Gabriel Loewinger avatar
Gabriel Loewinger@GLoewinger·18 Mar 2025

FLMM combines the benefits of 1) Mixed Models to account for between-animal heterogeneity, and 2) Functional Regression to model effects at each trial time-point. 5/13

Post attachment
Gabriel Loewinger avatar
Gabriel Loewinger@GLoewinger·18 Mar 2025

Solution: We propose an analysis framework based on Functional Linear Mixed Models (FLMM) that allows one to analyze signal– covariate associations at every trial time point. 4/13

Post attachment
Gabriel Loewinger avatar
Gabriel Loewinger@GLoewinger·18 Mar 2025

Problem: Photometry is often applied in nested longitudinal experiments with multiple trials per session and sessions per animal. This induces correlation, missing data, etc., that can obscure effects if not accounted for statistically. 3/13

Post attachment
Gabriel Loewinger avatar
Gabriel Loewinger@GLoewinger·18 Mar 2025

Problem: Common photometry analysis methods reduce detection of effects because, among other things, they average across trials and use 0 summary statistics (e.g., AUC, peak amplitude). 2/13

Post attachment
Micah Loewinger avatar
⟳ Gabriel Loewinger retweeted
Micah Loewinger@MicahLoewinger·24 Jul 2024

I am so excited to be co-host of @onthemedia, a show that I've loved making for over 8 years. @otmbrooke and our EP Katya Rogers have been incredible mentors. Thank you to OTM's world-class producers @eloiseblondiau @mollyfication @Rebecca_CC_ @cwango_ Thanks for listening ❤️

A note from @MicahLoewinger as we announce he is officially our co-host!! wnycstudios.org/podcasts/otm…