Estimating the feasibility of symptom based classification of COVID-19 at medium.com/@nigam/estimat…

Excited to announce our first code release of tools for weakly supervised learning using clinical text github.com/som-shahlab/eh… We've been using SnorkelML to rapidly classify clinical information for +COVID patients using emergency room notes #COVID19Pandemic

Estimating hospitalizations indirectly via case count led to 8-10x overestimate of bed needs, and $600m spent on unused field hospitals. As we try various policies for reopening, we need to measure hospitalizations to guide our reopening - tinyurl.com/MWM-hospitaliz… Stanford Medicine

#DataScientists Alison Callahan, Ethan Steinberg, and Jason Alan Fries estimate the efficacy of using presenting symptoms for prioritizing testing for COVID-19 and doing symptom-based surveillance Stanford Medicine, nature.com/articles/s4174….

I'm excited to share a preprint of our work on characterizing the limitations of fair machine learning techniques for clinical predictive modeling with Agata Foryciarz and Nigam Shah. paper: arxiv.org/abs/2007.10306 and code: github.com/som-shahlab/fa…. A thread:

I love being part of a lab of fellow data nerds who are as excited as me about our latest IRB protocol approval!

141, Idealism & related systems & doctrines. I am an idealist (definitely the same thing) so it goes well!

This is an excellent thread, and yet another example of the complexities in interpreting pandemic data.

It’s been an amazing effort to get us to the launch of Atropos Health, with incredible work by Saurabh Gombar, Nigam Shah, Brigham Hyde and the whole team. I’m honored to be a part of it!

Bare feet, in nature, with dog. Three of my favorite things.

Not cool, Photos, not cool at all.

She felt “this internalized sense of shame and judgment — that a county board felt that I couldn’t make this decision, and they would make it for me,” she said. 😠 #CovidVaccine

As a proud UofT MISt and erstwhile librarian, how could I resist?

ACE: the Advanced Cohort Engine - a tool that combines a unique query language for time-aware search of longitudinal patient records, rapid electronic phenotyping, cohort extraction, and exploratory data analyses doi.org/10.1093/jamia/… Nigam Shah Jose D. Posada, PhD Alison Callahan Saurabh Gombar

Excited to share our new paper on weakly supervised medical #NER in Nature Communications nature.com/articles/s4146… We explore combining multiple medical ontologies to more easily train and share clinical NER models without requiring hand-labeled training data

"The road to progress can be paved, in part, by improving our nation’s maternal health data infrastructure by collecting, sharing, & linking appropriately safeguarded, comprehensive data on the characteristics of individuals, the care they receive, and their non-clinical needs."

Texas #SB8 represents a clear attack on the practice of medicine in our country and on decades of established legal precedent. Read and share our joint statement from @ACOG and MeandMyOBG: acog.org/news/news-rele…