Dan Zeltzer

Just published in @jpube.bsky.social:

"Impacts of home-care subsidies: Evidence from quasi-random assignment"

By Yuval Ofek-Shanny, @avnershlain.bsky.social, & @dzeltzer.bsky.social

www.sciencedirect.com/science/arti...

#econsky #publiceconomics

Thanks! :)

I'm happy to share that I will be joining Berkeley School of Public Health as an associate professor this summer.

Grateful for the opportunity and sad to leave the Berglas School of Economics at Tel Aviv University, my home for years as both student and faculty.

When physicians with access to A.I. are compared with A.I. for virtual urgent care visits
The A.I. performed better @annalsofim.bsky.social
www.acpjournals.org/doi/10.7326/...

This work is with a team of medical and technology experts I am lucky to collaborate with: Zehavi Kugler, Lior Hayat, Tamar Brufman, Ran Ilan Ber, Keren Leibovich, Tom Beer, Ilan Frank, Ran Shaul (KHealth) Caroline Goldzweig, and Joshua Pevnick (Cedars-Sinai Medical Center).

Natural follow-up questions: causal impacts of AI on decision quality, speed, and outcomes. Best ways to combine human and algorithmic expertise. Trust, over/under reliance, continued learning. Lots of more work to do!

Physicians had to scroll to see recommendations, and we don't know if they did. We didn't study AI impacts. Still, results show well-calibrated models can do very well - more than the average doctor - on
common symptoms that account for the bulk of virtual urgent care cases.

AI excelled at guideline adherence (e.g., no antibiotics for viral URI) & incorporating EHR data (e.g., noticing recurrent UTI history that physicians missed). Physicians did better when patient narratives evolved or when visual findings were important.

Overall scores were equal in 68% of cases, better for AI in 21%, and better for physicians in 11% (these are actual physician decisions in urgent care cases with common symptoms - not medschool exams or vignettes)

AI recommendations were rated optimal in 77% of cases vs 67% for physicians, and potentially harmful in 2.8% vs 4.6% for physicians.

Experienced physician adjudicators viewed full case info and independently rated AI recommendations & physician decisions on a 4-point scale. If two independent adjudicators disagreed, a third was brought in, but we didn't force consensus and weighted final ratings equally.

We sampled 461 adult visits from one month in 2024 with respiratory, urinary, vaginal, eye, & dental symptoms - which in previous work showed had high AI-clinician diagnostic concordance; symptoms represent 2/3 of clinic volume. www.mcpdigitalhealth.org/article/S294...

The AI, commercially developed by KHealth, is an ensemble of discriminative models trained and tested on clinical EHR + med guidelines. To be reliable, the model withholds recommendations when confidence is low (about 1/5 of cases).

The setting is CS-Connect, a telemedicine clinic where AI first automates patient intake via structured chat. AI predicts the differential diagnoses and recommends Rx, labs & referrals. Physicians then conduct a video visit.

New @annalsofim.bsky.social paper: In 461 @cedarssinai.bsky.social virtual urgent care cases with common symptoms, AI recommendations were rated optimal more often than physicians' (77% vs 67%), & less often potentially harmful (2.8% vs 4.6%).

Agree! While the focus was AI accuracy, this study is from a clinic where (calibrated, discriminative, ensemble) models already support physician work by expediting intake and providing (when confident) diagnostic predictions and care recommendations. And we plan to study collaboration impacts next!

⛽ For an academic study, we (@akraft.bsky.social and I) wish to interview US Gas Station owners to learn more about their business practices. Please share or send anyone you know to Andreas or me. Thanks!

Mississippi State Univ. is hiring a tenure-track Econ Assistant Prof w/ August start!

Open field; pref. for applied micro & teaching PhD Micro I & II (Micro II this fall).

Apply: MSU site & www.aeaweb.org/joe/listing.php?JOE_ID=2025-01_111475768

Please repost—it's an off-cycle search. Thanks!

AGI is coming in 2 years max

Looks like MC>MB from just scaling...
open.substack.com/pub/garymarc...

ishaniganguli.bsky.social @davidmcutler.bsky.social Ateev Mehrotra et al.

New DiD paper on US telemed adoption finds higher adoption assoc. w/ slightly more visits but lower spending (incl low value). Mirrors our findings from Israel, suggesting robustness across payment and delivery systems! US: doi.org/10.1001/jamainternmed.2024.8354
ISR: doi.org/10.1093/jeea/jvad035

Raphael is a disabled Army veteran who worked at the VA in Seattle and played a critical role in processing servicemembers' disability benefits.

He's received nothing but stellar performance reviews.

Elon Musk fired him with ZERO justification. Help share his story.

After being alerted about possible misconduct, the I4R are reproducing published papers that use data from a specific NGO (GDRI). This thread releases the first 2 reports and provides more information about the work and responses/statements from authors journals and journals. 🧵

Brief break from hawking The Onion to tell everyone I just subscribed to Wired. Please support publications who do brave work right now.

www.wired.com/v2/offers/wi...

PLEASE RT :)

Submit to the annual Network Science and Economics Conference, the largest US event for network theory and empirical research in and adjacent to economics

Deadline Jan 30 (soon!)

Conference Apr 11-13 at Stanford

Sad about David Lynch. Beyond his famously great stuff, I remember really enjoying "The Intervew Project", an experimental documentary where his team traveled for weeks all over America and filmed random people they met along the way.

For those who didn't see Janet Currie's AEA Presidential Address live, here's the recording of her talk titled:

"Investing in Children to Address the Youth Mental Health Crisis"

www.aeaweb.org/webcasts/202...

Mostly raw (btw asking it to comment code and make it nice is also something I do, separately)