Adler Perotte (@aperotte) profile picture

Asst. Prof. of Biomedical Informatics at Columbia University

Joined March 2011
Photo by Adler Perotte (@aperotte)Photo by Adler Perotte (@aperotte)Photo by Adler Perotte (@aperotte)
Out today in @JACCJournals, our ValveNet study shows a ML model can detect mod-severe left sided valve disease from ECG. This model is currently running live in our health system daily. Now let me tell you everything wrong with it! A brief 🧵 bit.ly/3boQBRs
20
55
14
278
Come join us at the poster session 7:30-9pm New York Time (EST) at #neurips2021 Poster Session 2, Location A2! nips.cc/virtual/2021/poster/…. Joint work with @han_xintian along with @lycanduo, Thomas Wies, @aperotte, and Rajesh Ranganath!
1
1
Come learn about ML in cardiology at our #AMIA2021 workshop! Discussing work/collabs of @euanashley @StanfordAIMI @David_Ouyang @ColumbiaCardio @ColumbiaDBMI @mattlungrenMD. Will share the best resources we've found to get you started. Sold out but show up and we'll make it work!
3
2
26
In our latest News and Views: @nicktatonetti and @noemieelhadad discuss new data on the use of fine-scale genetic ancestry as a potential new tool for #precisionmedicine nature.com/articles/s41591-0…
11
21
Congratulations to 2021 PhD graduate Tian Kang, who led the study "A neuro-symbolic method for understanding free-text medical evidence” recently published by #JAMIA. @AMIAinformatics @nychunhua @aperotte ➡️ academic.oup.com/jamia/advan…
1
7
We looked at ECG abnormalities of #COVID19 patients, out in @JAHA_AHA. Particularly alarming was afib/flutter pts had 59% mortality vs 21% in those without (HR 2.1 in Cox model). Many thanks to @ElaineWanMD, @NirUrielMD, and @ColumbiaCardio team. bit.ly/3hpBd5X
2
12
I'm honored that @aperotte's and my poster -- #NORAModel -- was named best contribution for #MethodsResearch at #OHDSI2020.🥲There were so many wonderful presentations this year! If you're interested in learning more about this #Bayesian #machinelearning model, DM me!
The 2020 Best Contributor for #MethodsResearch is @AJAveritt (a 2019 winner as well 👏). Her work is "Noisy-Or Risk Allocation: A Probabilistic Model for Attributable Risk Estimation" It was also presented at a recent community call: ohdsi.org/2020-global-sympos… #JoinTheJourney
2
3
13
Adler Perotte retweeted
Now, what if the confounders were specified as a function of interventions of interest: h(T)? Like population structure in a GWAS problem? We call this problem Estimation with Functional Confounders (EFC): nips.cc/virtual/2020/public/… Work with @aperotte and Rajesh Ranganath.
1
1
3
A great reason to apply for a postgrad at DBMI - our faculty! 👀 the video below & apply (gsas.cuimc.columbia.edu/appl…) by the Dec. 1 deadline! @noemieelhadad @nicktatonetti @aperotte @nychunhua @SarahRossetti47 @JAMIAEditor_Sue @PracticalHIT @ciminoj ➡️ youtu.be/CtB2un1rf4I
3
11
My paper got an #honorablemention in the #AMIA2020 Year in Review! It uses #epidemiology+#informatics+#EHRs to study the #opioid crisis. Read it here 👉 bit.ly/3nvgoaQ @BenSlovis @aperotte
4
1
26
Attributable risks (ARs) support inference of whether an outcome is caused by an exposure. At #OHDSI2020, I will present my #NORAModel- a #Bayesian method of AR estimation that rectifies evidence at the population & individual-level. DM me! #CausalTwitter youtu.be/_FZWDbiNi4I
In our @MayoProceedings paper on pts admitted with #COVID19, we discuss findings helpful for early triage to floor vs ICU. Takeaway: EKG in ED is important--look for afib/flutter, RV strain, ST segment changes. Big thx to @NirUrielMD and @ColumbiaMed team! mayocl.in/31ZoteT
3
1
13
I am ready to get to work leading @nychealthy, the finest health department in the country. Public health is a calling I've answered throughout my career—I invite you to read the remarks I shared earlier about my story. Please follow me at @NYCHealthCommr medium.com/@NYCMayorsOffice/…
44
39
21
310
To keep up to speed with recent work on abstractive summarization, I wrote a (little) lit review called "Sending out an SOS (Summary of Summaries)". link.medium.com/nyveXXbcG8
3
6
did somebody call a doctor?! Today I successfully defended my dissertation, entitled Machine Learning Methods for Causal Inference with Observational Biomedical Data. Many thanks to my advisor @aperotte and my @ColumbiaDBMI family! #machinelearning #causalinference #PhDone
13
6
1
81
Researchers from @Columbia have have launched #CovidWatcher, an app that tracks #COVID19 NYC hot spots & surveys users about their COVID exposure, symptoms, access to medical care, and impact on daily life. @noemieelhadad @nicktatonetti @lena_m ➡️ dbmi.columbia.edu/covidwatch…
Adler Perotte retweeted
The first #preprint from the #OHDSICOVID19 study-a-thon, now posted on @medrxivpreprint, shares the risks in combo use of #Hydroxychloroquine and #Azithromycin. These are preliminary findings & not certified by peer review. #JoinTheJourney ➡️ medrxiv.org/content/10.1101/…
2
26
16
38
Grateful for the work that all of these 3D-printing folks are doing, and especially @MadihaZahra. Thank you from all of us at @ColumbiaMed! (my first tweet in 4 years, so you know it was worth saying) #GetMePPE
As hospitals are running dangerously low on protective gear, 3D-printing is helping protect health workers.
1
3
2
18
Are you a rising HS senior or undergrad interested in data science or medicine? Looking for a #summerFellowship or #summerInternship? The deadline to apply is this Friday! #stem #medicine #datascience #machinelearning #DBMIsummer #diversity #inclusion
Our #SummerFellowship provides opportunities for HS seniors/undergrads to gain knowledge, hands-on skills, & #research experience in #biomedicalinformatics. PLEASE SHARE with anybody who might be interested! 🚨 DEADLINE IS THIS FRIDAY! 🚨#DBMISummer ➡️ dbmi.columbia.edu/summer/
5
2