AI-Generated Genomes, Retinal Implants, and Palomar’s Mystery Lights Explained

We open with a University of Toronto paper using a generative model (“OncoGAN”) to create synthetic cancer genomes for safer, faster precision oncology; jump to a New England Journal of Medicine clinical trial where a wireless, sub-retinal photovoltaic implant (PRIMA) restores central vision in advanced AMD; and close by pulling apart new analyses of the Palomar Sky Survey “multi-point transients” with a first-principles look at alignment statistics and an “Earth’s shadow” test. Hosted by Lester Nare and Krishna Choudhary.

Summary

• Precision oncology without the privacy bottleneck: how a University of Toronto pipeline (OncoGAN) can synthesize ultra-realistic cancer genomes to democratize model training.

• Vision restoration, not just slowdown: the wireless, sub-retinal PRIMA implant and results from the PRIMAvera study (NEJM).

• Are those “mystery lights” real? Palomar POSS-I plate transients, alignment significance, and the decisive Earth’s-shadow control.