Finding answers with a ‘Genomics AI’ community

AI developer communities on speech recognition, natural language processing and object recognition have propelled AI into our daily lives. In a similar way, communities on genomics and computational biology could develop and play a key role in discovering pathways to disease if a shared infrastructure and tools were provided.

Given access to data and applications that can be queried by research experts for the relationships between cause and effect, for example, if a genetic variation increases or decreases with another biomarker or vice versa. This iterative approach and way to interrogate the data by researchers allows to compare against machine learning models and drives data-driven interpretations.

To develop communities, Block23 will provide researchers data access with its own querying tools to conduct ‘what-if’ assumptions. Results can be rendered on graphical interfaces (e.g. heat maps) using linear models, decision trees, and random forests that enable researchers to easily drill down into the data and visualize how attributes (variations) are organized in priority.

The end goal is to improve machine learning accuracy by having expert researchers contribute by evolving interpretive models and testing assumptions in live trials.

Block23 Decision Tree with Event Probabilities 
Block23 Genomic and Health-Lifestyle Risk Model
Block23 Biomarkers active in colorectal cancer