Today we are releasing a first glimpse into the work we've been doing internally at NOETIK and sharing some details around OCTO, the first world model of cancer and the tumor microenvironment. OCTO is a massively multimodal, transformer-based model trained on proprietary data from thousands of patients’ tumors purpose-built for self supervised learning at NOETIK. It is truly multimodal, ingesting data that includes paired H&E, spatial protein, spatial transcriptomics, and whole exome sequencing for each patient. We've desigend OCTO to operate as a simulation engine to power diverse drug discovery tasks from target identification, to biomarker discovery and patient stratification, even prediction of therapeutics response populations. To find out more about OCTO, read our first technical report: https://www.noetik.ai/octo
Awesome work Ron.
Amazing to see so much progress in such a short time! Keep up the good work!
🤯🤯🤯
super cool ron and co - i look forward to seeing where y'all take this
It sounds amazing. Always available if you need help training / refining OCTO on Oracle Cloud GPUs. Best, Rick.
Exciting! Great to see the start of foundational models for oncology. Thanks for pushing the field forward 👍
This is really a breakthrough.
That's really awesome!
❤️
Machine Learning Engineer (ex-IBM Research)
1moWill this model be released publicly? I am interested in trying it, in a research setting