Thread by @NightingaleOS

We'll be live-tweeting from our launch at #NeurIPS2021 today

Follow-along here!

@ericschmidt – Former CEO of Google, technologist, entrepreneur, and philanthropist:

“Nightingale is incredibly important. It is the first large database of images that is being organized around healthcare. We saw how well this worked with ImageNet in 2011.”

@ericschmidt “I believe, with the Nightingale team, this repository of never-before-seen images, tied to outcomes with labeled data, will lead to revolutionary new approaches.” - @ericschmidt

@ericschmidt Ari Robiczek talking about March 2020 and the origin story of the Covid dataset, which you can find here:

docs.nightingalescience.org/psj-covid-xray…

@ericschmidt Judy Gichoya @judywawira from @EmoryRadiology, on the opportunities and challenges of finding and building meaningful health datasets, in the US and abroad

@judywawira is of our amazing early research partners: nightingalescience.org/partners

@ericschmidt @judywawira @EmoryRadiology Matthew Lungren @mattlungrenMD from @StanfordAIMI on the origin of the Fractures dataset, which you can find here:

docs.nightingalescience.org/fracture-aimi-…

@ericschmidt @judywawira @EmoryRadiology @mattlungrenMD @StanfordAIMI Elaine Nsoesie @ensoesie from @BUSPH draws our attention to social determinants of health -- one piece of advice she'd give a multi-funder coalition (like the one we announce today!) on what data needs to be built up

@ericschmidt @judywawira @EmoryRadiology @mattlungrenMD @StanfordAIMI @ensoesie @BUSPH Now a series of Spotlight Talks moderated by @MattBMcDermott

@ericschmidt @judywawira @EmoryRadiology @mattlungrenMD @StanfordAIMI @ensoesie @BUSPH @MattBMcDermott @RoxanaDaneshjou @StanfordMed shares her project on fair and generalizable algorithms for dermatology, w/ call to action for more image datasets with diversity across skin tones

"Disparities in Dermatology AI: Assessments Using Diverse Clinical Images"

@ericschmidt @judywawira @EmoryRadiology @mattlungrenMD @StanfordAIMI @ensoesie @BUSPH @MattBMcDermott @RoxanaDaneshjou @StanfordMed Sharmita Dey, from @uniGoettingen, shares a data-driven model of gait:

It allows us to craft prostheses that *learn* from how people actually walk, rather than *tell* people how to walk in a hard-coded way

cc: @2plus2make5 @oziadias et al. on knee pain in @NatureMedicine

Daniel Gedon @danigedon, from @UU_University, shares a project related to missed heart attacks and ECGs:

"ResNet-based ECG Diagnosis of Myocardial Infarction in the Emergency Department"

Reminds us of

https://twitter.com/reverendofdoubt/status/1470529659767070732?s=20

by @smithECGBlog et al.

@danigedon @UU_University @smithECGBlog Now a panel on Medical Mysteries, where @kevin_volpp and @EricTopol even discuss their own experiences as patients with @oziadias

Here is a video of Kevin Volpp's incredible story:

heart.org/en/news/2021/1…

@danigedon @UU_University @smithECGBlog @kevin_volpp @EricTopol @oziadias As @oziadias points out, we have two exciting datasets related to the heart:

One that could help the emergency team figure out why the heart stopped, what they can do about it in real time

docs.nightingalescience.org/arrest-ntuh-ec…

@danigedon @UU_University @smithECGBlog @kevin_volpp @EricTopol @oziadias We also have a dataset on Diagnosing ‘Silent’ Heart Attack, which could help identify patients who urgently need drug regimens that prevent their next heart attack

docs.nightingalescience.org/silent-cchs-ec…

@ericschmidt correct link: docs.nightingalescience.org/covid-psj-xray…

@EricTopol — Founder and Director, Scripps Research Translational Institute @ScrippsRTI:

“We welcome Nightingale Open Science, a new non-profit resource that will make large medical image datasets available to the research community.”

@EricTopol @ScrippsRTI “Nightingale OS will unquestionably help advance and accelerate AI in medicine, just as ImageNet did for deep learning several years ago. I am thrilled to serve as an advisor to the team.” - @EricTopol

@EricTopol @ScrippsRTI @kevin_volpp shared about surviving a heart attack. We hope our datasets will help save more ppl

@EricTopol had a knee replacement. We're hopeful about a future dataset on knee pain (w/ @judywawira)

@oziadias missed his own appendicitis for 4 days --still asking "how?!"

@EricTopol @ScrippsRTI @kevin_volpp @judywawira @oziadias Now a panel on Data Science for Healthcare in Academia and Government

Our managing director @khay gets insights from Kate Baicker (Dean of @HarrisPolicy), @aneeshchopra, and @jenniferchayes

@EricTopol @ScrippsRTI @kevin_volpp @judywawira @oziadias @khay @HarrisPolicy @aneeshchopra @jenniferchayes @aneeshchopra – Former U.S. CTO, Co-Founder & President, CareJourney:

"We need an 'all hands on deck' approach to transform our care delivery system towards higher value and the work starts with open data platforms for learning.”

@EricTopol @ScrippsRTI @kevin_volpp @judywawira @oziadias @khay @HarrisPolicy @aneeshchopra @jenniferchayes “I'm grateful for the launch of Nightingale Open Science and am eager to celebrate the community of entrepreneurs, innovators, policymakers and front-line care teams putting those insights into practice for the betterment of the public.” - @aneeshchopra

@EricTopol @ScrippsRTI @kevin_volpp @judywawira @oziadias @khay @HarrisPolicy @aneeshchopra @jenniferchayes Jennifer Chayes @jenniferchayes, Associate Provost @BerkeleyDataSci mentions computationalhealth.berkeley.edu as a way to build the next generation of domain experts and data scientists to build the future of health

Kate Baicker, Dean of @HarrisPolicy, reminds us that lack of data is a big barrier to field advancement, and so is a lack of training---mentions this new degree program as one example of how to start addressing this: capp.cs.uchicago.edu

Now a panel on Data Opportunities, where @oziadias talks to @BarzilayRegina, Bin Yu, @2plus2make5 and @MarzyehGhassemi about unsolved medical problems and where new data can help

@oziadias @BarzilayRegina @2plus2make5 @MarzyehGhassemi @BarzilayRegina, Distinguished Professor @AIHealthMIT @MITEECS, shares about bringing mammogram interpretation into the digital age, translating these data into better outcomes, and the roadblocks she's faced in trying to share data

@oziadias @BarzilayRegina @2plus2make5 @MarzyehGhassemi @AIHealthMIT @MITEECS @BarzilayRegina's work really resonates with one of our datasets, on Identifying High-Risk Breast Cancer, which you can find here:

docs.nightingalescience.org/brca-psj-path.…

Bin Yu, Chancellor's Distinguished Professor @UCBStatistics and @Berkeley_EECS, highlights interpretability---how valuable it is to be able to talk to doctors, ask the right person the relevant question

Our datasets are created together with health systems and they stay engaged

Emma Pierson @2plus2make5, Assistant Professor of CS @cornell_tech, on paper in @NatureMedicine on knee pain---using medical images linked to ground truth (not what a radiologist said about the x-ray)

nature.com/articles/s4159…

Our datasets are all linked to ground truth outcomes

@2plus2make5 @cornell_tech @NatureMedicine The research above, led by @2plus2make5, showed that AI was able to detect causes of knee pain in Black patients that physicians miss.

Extensions of this work--new datasets--will be featured on Nightingale

There's a great @NPR @scifri episode about it: sciencefriday.com/segments/algor…

@2plus2make5 @cornell_tech @NatureMedicine @NPR @scifri Marzyeh Ghassemi @MarzyehGhassemi Assistant Professor @MITEECS--and a pioneer for ML in health--talks about how getting data is a crucial blocking point in advancing this work, and having it benefit the wider community

See one of her recent papers here:

https://twitter.com/MarzyehGhassemi/status/1469387389818785796?s=20

@2plus2make5 @cornell_tech @NatureMedicine @NPR @scifri @MarzyehGhassemi @MITEECS Now we break for a Poster Session, featuring posters from Aakanksha Rana, Allen Ding, and Diana Mincu

@2plus2make5 @cornell_tech @NatureMedicine @NPR @scifri @MarzyehGhassemi @MITEECS Now a panel on What Problems Get Funded in Computational Medicine, with our managing director @khay and Daniel Yang from @MooreFound--an early supporter of Nightingale

@2plus2make5 @cornell_tech @NatureMedicine @NPR @scifri @MarzyehGhassemi @MITEECS @khay @MooreFound Gordon & Betty Moore Foundation:

“The Nightingale OS platform and this collaborative funding call will support health systems that serve under-represented populations to ensure that public datasets and clinical AI tools are more inclusive of all populations.” - @MooreFound

@2plus2make5 @cornell_tech @NatureMedicine @NPR @scifri @MarzyehGhassemi @MITEECS @khay @MooreFound Another early funder is Ken Griffin — Co-founder and CEO, Citadel:

“Nightingale Open Science enables the world’s leading data scientists to apply powerful machine learning and predictive analytics to solve some of medicine’s most important and urgent challenges.”

@2plus2make5 @cornell_tech @NatureMedicine @NPR @scifri @MarzyehGhassemi @MITEECS @khay @MooreFound “This pioneering initiative advances the field of computational medicine and has the potential to unlock lifesaving breakthroughs for people around the world.” - Ken Griffin

@2plus2make5 @cornell_tech @NatureMedicine @NPR @scifri @MarzyehGhassemi @MITEECS @khay @MooreFound Announcement:

We’re so pleased to partner with @SchmidtFutures @MooreFound and @PJMFnd to launch a new funder collaborative to support building and hosting critical medical datasets on the Nightingale platform

@2plus2make5 @cornell_tech @NatureMedicine @NPR @scifri @MarzyehGhassemi @MITEECS @khay @MooreFound @SchmidtFutures @PJMFnd This funder collaborative will have a specific focus on sourcing datasets from regions and health systems that serve under-represented populations

We're now accepting new funders to join the collaborative until April 1, 2022

Please get in touch with @khay if you're interested

Now a platform video demonstration, by Nightingale's CTO Josh Risley

We hope you’ll join us and sign up for data access here: app.nightingalescience.org

Closing out this exciting launch day with remarks from @oziadias on the datasets featured on the platform (docs.nightingalescience.org) and why we're *so thrilled* to share these with you

Researchers, health systems, funders, please join our mission!
nightingalescience.org 🚀🎉