Could blood pressure drugs change COVID-19 risk? A proof-of-concept with AI
A transformer-based model estimated what would have happened under different blood pressure treatments, using data from over 300,000 patients.
The question
Early in the pandemic there were worries that some common blood pressure drugs might raise the risk of catching COVID-19, and hopes that others might lower it. Real-world records are messy, and people on different drugs are not directly comparable.
What we did
We used a transformer model, the same family of technology as modern language models, to estimate counterfactual outcomes, which means what would have happened to the same person under a different treatment. We applied it to real-world data from more than 300,000 patients.
What we found
ACE inhibitors showed a negligible effect on COVID-19 risk. Beta-blockers and calcium channel blockers were associated with a protective effect, while statins and thiazides were associated with a slight increase in risk.
Why it matters
This is a proof-of-concept study, so these results are not a reason for anyone to change their medication. What it shows is that this kind of AI-based causal analysis can help evaluate treatments in messy real-world data.