← Showroomsciencemachine · read-only view of a finished run
Run pcos-sweep-1 · swept 10–13 Oct 2026

Open edges: a sweep on PCOS polycystic ovary syndrome

How the machine works. You give it a narrow topic. It harvests papers into a local database, and Claude agents pull out sentences where authors say something is not yet known; code checks that every quote really appears in the paper's stored text, or the edge is dropped. A separate challenger then searches the literature and public data archives for anything that already answers it, and each surviving edge is scored and ranked with all its receipts attached.

Research map, not medical advice. Nothing here is a treatment recommendation. See the same data in GeeGobi ↗
Pipeline, seed searches and how this field has been studied
seed→triage→full text→extract→verify quote→screen→score→deep challenge→pairwise rank→same-gap check→feasibility→report
The six seed searches (Europe PMC) that started the run:
    Records indexed under "Polycystic Ovary Syndrome" by five-year period (grey); randomised trials + meta-analyses + systematic reviews in blue. Counted from index headings, not a literature search.
    Exported read-only from the run's database and runs/pcos-sweep-1/ files. Paper text is not reproduced, only the verified sentence each edge rests on; links go to the publisher or index. Searches record what was looked for, not a claim about the whole literature.