What can #AI do for you (and your #GWAS)? Two new papers show how automated optic nerve head labeling can improve downstream genomic discovery.

Today we share how #ML models can be trained to accurately predict phenotypes, and how these predictions can be used to identify novel genetic associations that lead to more accurate predictions of disease predisposition. Learn more ↓ goo.gle/3gYoLtX

Release of DeepVariant v1.2. Refactor by Maria Nattestad modularizes multi-sample components, improving DeepTrio accuracy. Accuracy improvements for PacBio HiFi, especially for newest PacBio chemistry. Updates to python, TF libraries improves speed. github.com/google/deepvar…

Very excited to present our work on DeepConsensus, a transformer model for PacBio HiFi sequencing. The model operates on a MSA of subreads + draft consensus and was trained using an alignment-based loss. Collaboration with many amazing folks Google and PacBio. Congrats, team!

Excited to share new phenotyping methods for COPD, improving COPD GWAS Collaborative work with IU and BWH, co-authors Justin Cosentino, B Besaz , @babak_alipanahi, Z McCaw, last author Farhad Hormozdiari, eng manager C McLean. Paper: medrxiv.org/content/10.110… Code: github.com/Google-Health/…

Today we’re highlighting a method for training accurate #ML models for genetic discovery of diseases, even when using noisy and unreliable labels. Read how we applied this to better characterize lung function and chronic obstructive pulmonary disease ↓ goo.gle/3LANzss

We’re excited to describe new Google Health work that makes phenotyping more scalable. An unsupervised deep learning method, REpresentation learning for Genetic discovery on Low-dimensional Embeddings (REGLE), to perform GWAS on high-dimensional clinical data.

At ICML and interested in LLMs and health? We'll be at the ML4MHD workshop today to present our recent work on HeLM: Multimodal LLMs for health grounded in individual-specific data. Come by at 4pm for the talk or during one of the coffee breaks to chat! arxiv.org/abs/2307.09018

Excited to share new methods for cardiovascular phenotyping from accessible inputs (ECG and PPG) and scalable unsupervised methods. Co-authors: Yuchen Zhou @cosentiyes, Ted Yun, last author B Besaz & Farhad Hormozdiari, eng manager C McLean. Paper: medrxiv.org/content/10.110…

Today on the blog, read about the latest from our two new research papers on how AI, particularly fine-tuned Gemini models, can create personalized health experiences that cater to individuals’ unique health journeys. goo.gle/3RnwHbl #AI #healthcare #personalizedhealth

Google’s research on a Personal Health Large Language Model (PH-LLM) is now published in Nature Medicine Magazine!