Cancer and many other medical conditions are diagnosed by using imaging techniques like MRI, X-ray, Ultrasound, and others, or by taking a biopsy to be examined under a microscope. About 5.5 billion such images are created every year in the industrial west. Up to 30% of diagnoses are wrong, leading to costs of over 20,000 dollars per patient in extra testing and needless treatment. Doctors spend up to three quarters of their time in tasks not interacting with patients but analyzing data and dealing with paperwork.

Artificial intelligence can help by automatically and instantly analyzing images with over 99% accuracy. Techniques developed in areas like autonomous vehicles have now been adapted to healthcare to detect and diagnose conditions. The one big obstacle remaining is the effort and time required to produce a clean and annotated dataset. The annotation process is now largely automated by AutoLabel by Samsung SDS cutting the cost to make medical AI by 90%, leading to medical AI models operating in the cloud to revolutionize healthcare.

свернуть