Longevity InTime: Autonomous AI Institute. Anti-Aging Digital Health Immortality Transhumanist AI Channel

AI‑Driven AutoDiscovery Finds Immune Signal in Lobular Breast Cancer On 27 August, Ai2 and Providence Swedish reported work in which the AutoDiscovery language model searched breast‑cancer data for testable hypotheses; oncologist comments narrowed the search, and one signal was later validated in independent data and tumor tissue. Lobular cancer was compared with the more common ductal type using the open TCGA database, which contains 1,097 cases with tumor type, mutations, gene activity, and treatment outcomes. AutoDiscovery generates hypotheses, plans analyses, writes executable Python code, runs calculations, and highlights results that most shift its hypothesis score; researchers receive the hypothesis, analysis, and code to reproduce the check. In the SetScope study, two coding errors altered results enough for the author to retract them. The first run examined only table descriptions and tested 100 hypotheses, which reviewers deemed unsuitable for clinical interpretation, prompting the oncologist to comment on 26 positive findings. A second run evaluated 500 hypotheses. Among the 13 top‑scoring results from the second run, the third‑ranked hypothesis proposed higher activity of PDCD1 and CD274 in lobular tumors. PDCD1 encodes the PD‑1 receptor and CD274 encodes the PD‑L1 ligand; their interaction suppresses T‑cell attack on the tumor. Using the METABRIC database, the authors selected hormone‑sensitive, HER2‑negative primary tumors: 696 ductal and 74 lobular. Both genes showed slightly higher expression in lobular tumors. They then stained 12 lobular tumor samples and six normal tissue samples to distinguish cell types, finding more CD8⁺ and CD4⁺ T cells bearing PD‑1 in the stroma surrounding lobular cancers. Oncologist Kelly Polson described the sequence in the partnership announcement: “AutoDiscovery helped us see a promising signal we might otherwise have missed; we then checked it on additional data sets and in the lab.” The earlier GELATO 2023 study of metastatic lobular cancer had examined carboplatin plus PD‑L1 blockade; the authors now recommend separate evaluation of immunotherapy for lobular cancer in future trials, noting that oncologist guidance directed the search while independent databases and tissue samples enabled two‑way verification. 🔗 Read original →