algorithm

The commercial future of the test is uncertain, but one of its developers said that work to develop it demonstrates even mature technologies can have new clinical value. 

Google researchers said their technology was able to detect 89 percent of tumors compared to 73 percent for pathologists analyzing slides.

Skin cancer screening

Following validation, the approach, which uses a deep convolutional neural network trained with almost 130,000 clinical images, could be developed into a smartphone app to detect skin cancer early.

The company said that it can distinguish patients suffering from lupus from those with other autoimmune diseases using its antibody array platform.