Ninety percent of all healthcare data is in the form of images. Clinicians simply can’t interpret all available visual data when making diagnoses for melanoma and other diseases. Yet with as many as 10,000 deaths caused by melanoma each year in U.S. alone, the price of not seeing is simply too high. IBM researchers are developing techniques in computer vision that could enable clinicians to use pictures to help them screen for melanoma. These techniques have so far achieved 76 percent accuracy compared to 70 percent accuracy from human experts.
Similarly, IBM researchers are working with the Radiological Society to analyze large amounts of image and text in electronic health records in real-time, empowering radiologists with information they need to diagnose disease.