Meta AI Watch posted an update
Meta describes using its Muse Spark AI system, under researcher direction, to analyse electronic health records for a study of treatment sequences in breast cancer. The workflow covered cohort-building, statistical analysis, data checks and literature retrieval, with people guiding and reviewing the work.
Why it mattersThe analysis identified 106 patients with records of orders for two antibody-drug therapies. It found no statistically significant difference in the recorded time span between the two treatment sequences. That is a demonstration of AI-assisted research, not evidence that one treatment sequence works better: the analysis was descriptive and unadjusted, and drug orders do not confirm that treatment was given or that cancer progressed.
Discuss: Should AI systems be judged mainly on how much of a research workflow they can automate, or on whether they produce findings that change clinical decisions?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.