Honored to receive an ✨NIH Director's Pioneer Award✨ to develop a framework for the social impact of algorithms in health care! This work is at the intersection of machine learning, decision science, economics, policy & health equity. I'm excited to get started! #NIHHighRisk

Woot woot, our generalizability & transportability review is out, co-authored with @sherrirose! If you'd like to learn more about how to assess and address external validity bias, take a look: https://t.co/0CKgv9hgJq, https://t.co/UGvplWo1U2

Our review of generalizability & transportability led by @IDegtiar is now published in @AnnualReviews: https://t.co/kafhz8AEWL (arXiv: https://t.co/hcfhK4g1M1) It synthesizes work across statistics, CS & health while proposing a framework for addressing external validity bias

What does statistics bring to machine learning & AI? New piece w/@mark_vdlaan on why machine learning cannot ignore the lessons of maximum likelihood estimation https://t.co/pog30I4BtV Many ML algorithms aren't suited for statistical inference by having deviated from sieve MLEs

How to find cause and effect in real-world randomness? A Tiny Lecture, #NobelPrize edition!

Our new paper, led by @IDegtiar, develops machine learning estimators for generalizability with observational & randomized data https://t.co/Rx42T9AuFH These methods were motivated by our interest in assessing plan-specific effects on 💲 in Medicaid Code https://t.co/ZgzgbN2ba5

Congratulations to HAI faculty member @SherriRose for winning the Mortimer Spiegelman Award – the highest recognition for outstanding contributions to public health statistics. Read more about her work:

Welcome to the Farm, Sherri Rose, an expert on statistical machine learning, AI and economics. "I love working on the computational health economics tools I develop because they may have a direct impact on individual lives in the health-care system." https://t.co/xhxDrvUhI5

Registration is open for ACM CHIL 2020! Keynote speakers include Yoshua Bengio of MILA, Sherri Rose of Harvard (@sherrirose), Nigam Shah of Stanford, Ruslan Salakhutdinov of CMU (@rsalakhu), and Elaine Nsoesie of Boston University (@ensoesie). Check out

Our new @NBERpubs working paper is out! https://t.co/VtDsmrtU39 Fixing undercompensation for several groups in risk adjustment improved fairness for *many other groups and overall fit* We did this by bringing together: 1️⃣fair regression 2️⃣ML for variable selection 3️⃣reinsurance

2020 NBER Conference on the Economics of AI — Submission deadline June 1.

This summer, @sherrirose will receive the @PennCausal Mid-Career Award for achievements in the development and application of innovative causal inference methods. Dr. Rose will deliver an invited award lecture during the Causal Inference Summer Institute: https://t.co/bPPd6L2Tuy

At a minimum, a #machinelearning model should be reproduced, and ideally replicated, before it is deployed in a clinical setting

Our paper on fair regression is now forthcoming in @Biometrics_ibs! Biometrics link https://t.co/51JZeoyYn4 ArXiv https://t.co/8w9iogBCVj Code https://t.co/Hp2xNLEzQh Discussed next steps needed to bring this work to practice at the recent Int’l Risk Adjustment Network meeting

Happy to announce our keynote speakers for CHIL 2020: Yoshua Bengio, Sherri Rose (@sherrirose), Nigam Shah (@drnigam), and Ruslan Salakhutdinov (@rsalakhu). A reminder that the deadline for papers is in *just over a month* (13th January), see our CFP:

Many machine learning papers have critical errors and they are accepted anyway in journals, says @sherrirose; she wrote a nice piece @JAMANetworkOpen on Machine Learning for Prediction in Electronic Health Data. https://t.co/VofZUtkdwR @HMSHCP @theNASEM

Nice talk by @sherrirose: Towards Standards in Machine Learning. And she is emphasizing the need for teams that cross disciplines, leveraging experience from different areas. Understand applied problem. Respect the analysis. Think about the application.

Important shift in the clinical literature. @NEJM releases new guidelines for statistical reporting of p-values. https://t.co/QHV78aKtIM