Members of the Google Brain team today announced that they have crafted computer vision for the identification of protein crystallization, claiming accuracy rates around 94 percent. Protein crystallization determines the shape of cells and can play a role in discovery of drugs to treat various illnesses.
“Hundreds of experiments are typically run for each protein, and while the setup and imaging are mostly automated, finding individual protein crystals remains largely performed through visual inspection and thus prone to human error,” Google Brain principal scientist Vincent Vanhoucke said in a blog post today. “Critically, missing these structures can result in lost opportunity for important biomedical discoveries for advancing the state of medicine.”
To train the AI model, Google researchers worked with the Machine Recognition of Crystallization Outcomes (MARCO) initiative, a partnership between pharmaceutical companies and academics.
Big tech companies and startups alike are increasingly entering the health care space, particularly to provide imaging services.
Baidu Research announced last month it has created an algorithm that is better than human pathologists at identifying tumor cells in breast tissue. Early results also show AI has become more accurate than people in the detection of skin cancer.
Also last month, Google researchers published in the journal NPJ Digital Medicine the results of work with predictive deep learning models that are able to determine whether a patient is likely to be readmitted, the length of their stay, and whether they are likely to die during their hospital stay.
Earlier this year, Google shared a computer vision model that can detect diabetic retinopathy, a key indicator of cardiovascular disease.
Verily Life Sciences — like Google, a property of Alphabet — is also interested in the area of drug discovery. Separately, Verily announced Wednesday it will partner with ResMed to explore sleep apnea, an affliction that can lead to weight gain or heart failure.