Paul Alexander Bloom

@paulbloom.bsky.social

Assistant professor in the Columbia University Psychiatry Department studying adolescent depression and suicide using digital tools 📱📊 he/him

Our case study takes a detailed look at language typed by 5 youth on their smartphones in the 30-day period before suicide-related hospitalization, combining both natural language processing (sentiment, suicide content, topics) and clinician qualitative review to identify potential risk factors.

NPP-Digital Psychiatry and Neuroscience (DPN)@dpn-journal.bsky.social · 6mo ago

🆕 paper from @isaactreves.bsky.social @paulbloom.bsky.social @prof-nick-allen.bsky.social et al using natural language processing to detect suicide risk. They demonstrated elevations in suicidal language & negative sentiment before hospitalization in adolescents. www.nature.com/articles/s44...

The fourth annual Suicide Research Symposium (#SRS25) takes place THIS WEEK, Wednesday (April 23) through Friday (April 25)! Registration is still open for our free, virtual event. Links to register and see the conference platform are on our website: www.suicideresearchsymposium.com

SRS 2025

We are delighted to announce that the fourth annual Suicide Research Symposium will be taking place April 23-25, 2025. Like previous years, this Symposium will be free and take place virtually. We ...

suicideresearchsymposium.com

Wanted to share news of the publication of this research paper with @auerbachlab.bsky.social and others. Using the EARS mobile sensing platform, we found that being homebound more than usual was associated with increased risk for suicidal behaviors in high-risk adolescents.

Detecting Risk of Adolescent Suicidal Thoughts and Behaviors With Smartphone GPS Data

This case series study examines whether an app installed on adolescents’ personal smartphones could identify risk of next-week suicidal events and clinically meaningful suicidal ideation by using geol...

jamanetwork.com

HEY ACADEMICS – I’ve been invited to review a paper for the first time! I’m excited and hope I can contribute meaningfully to the field. What are your tips/advice for reviewing (and avoiding becoming reviewer #2 😬), especially for newbies??? 📝📚

1. Evidence from @bttyeo &al supports the idea that you need some reasonable amount of data per subject (e.g., 20 min) to all but maximize within-sub reliability— then you get much greater gains by increasing number of subjects Paper in card, thread here: x.com/bttyeo/statu... 2/3

MRI economics: Balancing sample size and scan duration in brain wide association studies

bioRxiv - the preprint server for biology, operated by Cold Spring Harbor Laboratory, a research and educational institution

biorxiv.org

Are there studies that specifically talk about measuring the feasibility of specific measures designed for EMA? Working on a construct validation study and I could use some references for how you might think about whether a measure works in an EMA context.