Tomorrow at #NeurIPS we’re launching Nightingale Open Science, a computing platform empowering researchers to access massive new health imaging datasets
We hope Nightingale will help solve some of the biggest medical problems of our time
A 🧵on how to get involved (1/16)
We hope Nightingale will help solve some of the biggest medical problems of our time
A 🧵on how to get involved (1/16)
Launch workshop features product demo + panels w/ top minds in CS, tech, medicine
@EricSchmidt @BarzilayRegina @JenniferChayes @AneeshChopra @EricTopol @MarzyehGhassemi @kevin_volpp @judywawira @mattlungrenMD @2plus2make5 @ensoesie
nightingalescience.org/conferences-20… (2/16)
@EricSchmidt @BarzilayRegina @JenniferChayes @AneeshChopra @EricTopol @MarzyehGhassemi @kevin_volpp @judywawira @mattlungrenMD @2plus2make5 @ensoesie
nightingalescience.org/conferences-20… (2/16)
Just as ImageNet jump-started ‘machine vision’, we want to help build a new field of ‘computational medicine’
Nightingale’s mission is to bring together researchers, incl. computer scientists and clinicians, around questions pushing the boundaries of medical science (3/16)
Nightingale’s mission is to bring together researchers, incl. computer scientists and clinicians, around questions pushing the boundaries of medical science (3/16)
But you don’t need to come to #NeurIPS to join our mission
Anybody can sign up to access the data here: nightingalescience.org (4/16)
Anybody can sign up to access the data here: nightingalescience.org (4/16)
Nightingale is a non-profit, and our open platform is free to use
We fund and collaboratively build imaging datasets—curated around critical medical questions—with health systems around the world
We then make de-identified data available on a secure cloud platform (5/16)
We fund and collaboratively build imaging datasets—curated around critical medical questions—with health systems around the world
We then make de-identified data available on a secure cloud platform (5/16)
Why are we doing this? Humans have made huge strides understanding how the body works & how it fails
But deep mysteries remain:
Sudden cardiac death kills 300,000 Americans yearly, but even in the rear-view mirror, doctors can find no identifiable cause for the majority (6/16)
But deep mysteries remain:
Sudden cardiac death kills 300,000 Americans yearly, but even in the rear-view mirror, doctors can find no identifiable cause for the majority (6/16)
On cancer, we’re finding more small tumors with screening mammograms and colonoscopies, but have hardly moved the needle on rates of late-stage diagnoses or death
Doctors are missing something. Could AI complement human intelligence in the race to save our own lives? (7/16)
Doctors are missing something. Could AI complement human intelligence in the race to save our own lives? (7/16)
The key to solving these problems will lie in the massive volumes of complex, high-dimensional data produced every day: electrocardiograms, x-rays, CT scans, digital pathology images, and more
A spotlight of some of the data featured on the platform: (8/16)
A spotlight of some of the data featured on the platform: (8/16)
Diagnosing ‘Silent’ Heart Attack: 49,000 ECG waveforms linked to results of cardiac ultrasounds, to visualize scars in the wall of the heart formed by prior heart attack
This could help identify patients who urgently need drug regimens that prevent their next heart attack (9/16)
This could help identify patients who urgently need drug regimens that prevent their next heart attack (9/16)
Identifying High-Risk Breast Cancer: 175,000 digital pathology images from 11,000 patients, linked to patient outcomes (stage, metastasis, mortality)
This will allow algorithms to identify patients at high risk of poor outcomes (10/16)
This will allow algorithms to identify patients at high risk of poor outcomes (10/16)
Emergency Triage of COVID-19: 7,000 chest x-rays from Covid-19 patients, linked to data on pulmonary deterioration (need for a ventilator) and mortality
This could help doctors make critical triage decisions—whether patients are safe to go home or need to be monitored (11/16)
This could help doctors make critical triage decisions—whether patients are safe to go home or need to be monitored (11/16)
Predicting Fracture Risk: 64,000 chest x-rays linked to data on past and future fractures all over the body, as well as data on diagnoses of osteopenia and osteoporosis
This could help target preventive care and medications to those at highest risk (12/16)
This could help target preventive care and medications to those at highest risk (12/16)
Subtyping Cardiac Arrest: 24,000 ECG waveforms from ER patients who suffered cardiac arrest, linked to data on cause of the arrest and what happened to the patient
This could help the emergency team figure out why the heart stopped, what they can do about it in real time (13/16)
This could help the emergency team figure out why the heart stopped, what they can do about it in real time (13/16)
In our current health system, these type of imaging data are interpreted by humans
But machines have new ways of ‘seeing’ signals and patterns in the data that humans cannot (14/16)
But machines have new ways of ‘seeing’ signals and patterns in the data that humans cannot (14/16)
By focusing on data that link medical images with real patient outcomes, Nightingale enables the creation of algorithms that learn from nature—not from human judgment (15/16)
Nightingale is co-founded by @oziadias, a physician–scientist @UCBerkeley and @m_sendhil, a computational and behavioral scientist @UChicago
And generously supported by @SchmidtFutures @MooreFound Ken Griffin @PMJFnd
Join our mission: nightingalescience.org (16/16)
And generously supported by @SchmidtFutures @MooreFound Ken Griffin @PMJFnd
Join our mission: nightingalescience.org (16/16)
Correcting the handle for @PJMFnd!
• • •
Missing some Tweet in this thread? You can try to
force a refresh
