🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Prediction model recipe for disaster: Have 40 events. Screen 2000 SNPs, retain p<0.1. Step-wise slctn to throw SNPs+clinical vars into a model. Get AUC 0.87 (pic 1). Publish! The story often stops there. Thankfully, they validated, and it, unsurprisingly, fell apart (pic 2).

⚠️! "While many researchers may understand the AUC as an intrinsic measure of a clinical prediction model's (CPM) quality, in fact AUC is an extrinsic property of a CPM that emerges only when a model is applied to a specific population."

Scale is all you need. Once the training dataset contains Aleph-0 observations, AGI will be able to multiply all the integers.

LLMs have complex joint beliefs about all sorts of quantities. And my postdoc James Requeima visualized them! In this thread we show LLM predictive distributions conditioned on data and free-form text. LLMs pick up on all kinds of subtle and unusual structure: 🧵

Most people don't know this but AI is just statistics. It's a special kind of statistics where you can use data without permission, spend billions of dollars of resources on a single model, and you aren't personally responsible when it fails in idiotic ways and/or hurts people.

I understand that CI, from the 1970s until around 2010, was mostly focused on the challenge of moving from OBS to EXP worlds and controlling for confounding in this sense. However, it's an oversimplification to think about CI as solely about observational studies, as the

Common statistical tests are linear models (or: how to teach stats) lindeloev.github.io/tests-as-linea… I wish I were taught like this.

You know it's U.S. News and World Report Voting Season when random division chiefs reach out to be your best friend on Doximity.

👀!

New chapter in Regression Modeling Strategies #rms: ordinal regression generalizes Wilcoxon, log-rank, Kaplan-Meier, Cox PH model, and all common survival time analyses. Semiparametric regression is a unifying concept. hbiostat.org/rmsc/ordsurv #rstats #Statistics

One of the most important results of the LLM revolution: I'll never have to learn TikZ

This is a fun interactive article explaining the intuition behind Gaussian Processes. Gaussian Processes allow you to construct a distribution over functions with a continuous domain. They are incredibly useful for applications like Bayesian regression.

👀!

We are excited to host Miguel Hernán in Houston MD Anderson Cancer Center next week to give this year’s Melvin L. Samuels lecture (hybrid; zoom info in photo). Looking forward to the synergies that will emerge from his visit to advance rigorous causal inference methods across oncology.

Timeless advice

Very interesting paper on causal inference from statistical models; perhaps more actionable than most causal inference papers: tandfonline.com/doi/full/10.10… #Statistics