←
Hiesinger Lab65 posts
Hiesinger Lab avatar

Hiesinger Lab

@HiesingerLab

Innovation at the intersection of cardiothoracic surgery and machine learning.

◷ Joined August 2022▧ 4 Photos and videos
Joseph Cho avatar
⟳ Hiesinger Lab retweeted
Joseph Cho@joseph_cho1·Jul 28

Excited to share that EchoAI-Peds is now published at Circulation! @CircAHA ahajournals.org/doi/10.1161/…

We are excited to announce 🔥EchoAI-Peds 🔥, the first multi-task deep learning model for pediatric #echofirst analysis. It's been a pleasure to lead this alongside @mrudangm14 under the guidance of @hiesingerlab and in close collaboration with @jolleylab. Link below⬇️ [1/n]

Post attachment
Rohan Shad, MD avatar
⟳ Hiesinger Lab retweeted
Rohan Shad, MD@rohanshad·Mar 25

Happy to report that our work on building foundation models for Cardiac MRI has finally been published in Nature Biomedical Engineering! @natBME nature.com/articles/s41551-0…

I’m excited to finally showcase our work on developing a generalizable deep learning system for cardiac MRI. This is the culmination of over 3 years of research spanning my time as a postdoc in @HiesingerLab and now as a cardiac surgery resident at @pennsurgery, and I couldn’t be happier with the finished product. Weights to be released for academic use (coming soon!). All built with @pytorch and @pytorch_lightning. Lots of really cool stuff to unpack, but here are few main results: Contrastive learning with cine-sequence CMR studies from multiple different views (short axis, 2-chamber, 3-chamber, and 4-chamber views) with long-form CMR free text reports is really hard. A good sign for us was how the video transformer embeddings evolved with each epoch. We train for about 600 epochs taking about 2 weeks. In a zero shot fashion, the model can then do some really interesting things: separate out different diseases (ACDC dataset), and different genders, age groups, and ejection fractions (UK BioBank) with no explicit supervised instruction. No such behaviour with kinetics-600 initialized weights (see preprint)! Finetuning over our system yields results superior to baseline approaches on downstream tasks of interest, with 10x and sometimes 100x less data. Here’s a quick figure showing how we achieve superior results for UK BioBank LVEF estimation with just 1% of the available data. Better yet, that same model when tested on a Kaggle dataset does exceptionally well (dashed line = kinetics-600 performance on left; lower is better!). Focusing on some disease diagnosis tasks, we label a separate dataset of some 6000 individual patients. This is real-world representative unenriched dataset. Our methods show massive improvements in AUC compared to baseline methods. Validation set shown here, internal test set (Stanford / Medstar / UCSF), and external test set (UPenn) figures are in the paper.

Post attachmentPost attachmentPost attachmentPost attachment
Hiesinger Lab avatar
Hiesinger Lab@HiesingerLab·21 Mar 2025

A huge congratulations to super-🌟 med student and @HiesingerLab member @dhamank24! 🥳

✨Congratulations to our incoming @StanfordCTSurg integrated residents! 🎉 We're excited to have three new interns joining our program: Dhamanpreet Kaur of @StanfordMed, Arian Mansur of @harvardmed, & Alice Zhou of @HopkinsMedicine- welcome to the team!🌲#WeAreStanford #Match2025

Post attachment
Hiesinger Lab avatar
Hiesinger Lab@HiesingerLab·27 Aug 2024

🔥New preprint alert🔥 We're excited to share an early preview of 𝗦𝘂𝗿𝗚𝗲𝗻, a text-diffusion model for generating surgical videos! This model can generate videos of higher quality (720x480) & longer duration (49 frames) than the current SoTA. Great work by @joseph_cho1!

This tweet is unavailable

Cyril Zakka, MD avatar
⟳ Hiesinger Lab retweeted
Cyril Zakka, MD@cyrilzakka·25 Jul 2024

Two years ago to the day, I joined the @HiesingerLab at Stanford Medicine, and now it’s time for me to wrap up this chapter of my life! 🎓 Reflecting back, I see a tapestry of efforts—my family's unwavering support, my friends’ continuous encouragement, my own growth—and some luck along the way. It truly does take a village, and I'm grateful for every part of mine. It goes without saying that I’m grateful to both @rohanshad and @HiesingerLab for believing in me every step of the way, and always encouraging me to lead my own projects and find my own voice, despite many of the obstacles we faced. I’m also lucky to have met some truly impactful people whose work over the years has heavily influenced my own research direction, many of whom have also taken it upon themselves to guide me along the way, namely @Dr_ASChaudhari @RoxanaDaneshjou @curtlanglotz @DrAalami @pranavrajpurkar @euanashley @ardalal_MD and many more! Excited for what’s next!

Post attachment
Cyril Zakka, MD avatar
⟳ Hiesinger Lab retweeted
Cyril Zakka, MD@cyrilzakka·1 Jul 2024

In case you missed it, our team in collaboration with @Dr_ASChaudhari, @RoxanaDaneshjou, @DrAalami, and @vishnuravi defined a framework for autonomous EHR systems, and implemented the first Level 1 EHR system with retrieval and extractive capabilities! Check it out!

Super excited to showcase our newest work Almanac Copilot an EHR agent capable of answering questions about your patients and placing orders for you across any modern EHR system. Background: Nearly 75% of clinicians with burnout symptoms pinpoint EHRs as a source due to poor usability or workflow integration. Methodology: We train a 33B LLM to perform open-ended QA on patient information, as well as order placement using the FHIR interoperability standard (cc @zakkohane , @AdamRodmanMD) Result: Almanac Copilot obtains a success rate of 74% across 300 common EHR tasks based on MIMIC-IV. Your very own personal EHR assistant!

Post attachment
Cyril Zakka, MD avatar
⟳ Hiesinger Lab retweeted
Cyril Zakka, MD@cyrilzakka·18 Jun 2024

I’m giving a few talks on GenAI in Health in the next few months to technical founders and practicing clinicians across the US. Would love to highlight some open source medical AI projects in my slides and am looking for some suggestions. /cc @Michael_D_Moor @katieelink