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

Insilico Medicine launches O3DC consortium with 115 benchmarks for AI drug discovery On August 24, Insilico Medicine announced the launch of the Open Drug Discovery & Development Consortium (O3DC), an open alliance aimed at evaluating AI in drug discovery. Its first resource, the Benchmark Index, gathers open benchmarks to assess model performance under defined conditions. The Benchmark Index launched with 115 open benchmarks spanning ten categories and includes contributions from 15 consortia and initiatives. Each entry lists the task, custodian, code, repository status, and a known weakness that could skew results. In July Insilico also released the DDD Benchmark, which compares models across individual stages of drug development and reports the score a model achieved on a specific task. The Benchmark Index, by contrast, shows the conditions of the test itself: what is measured and which methodological quirks might make a score misleading. The Index highlights how data splitting influences outcomes, noting that random splits can let chemically similar molecules appear in both training and test sets, inflating performance. It recommends splitting by molecular scaffold, as advised by resources such as the Therapeutics Data Commons and cautions noted in the MoleculeNet benchmark about small datasets and noisy labels. 🔗 Read original →