Fran Blandy in Phys.Org:
In the developing world, basic healthcare is often a challenge—let alone expensive medical screening or tests for easily treatable, preventable illnesses. TEDGlobal, an annual conference devoted to "ideas worth spreading" taking place in Tanzania this week, heard of new technologies that could revolutionise healthcare for the poor. Infectious diseases are fast being overtaken by afflictions such as cancer as the biggest health problem in Africa, where some countries have only one pathologist per one million people.Sierra Leonean roboticist David Sengeh believes training more specialists is not enough, and is working with his team at IBM Africa on artificial intelligence (AI) algorithms that can predict a cancer's progression. AI software can be trained with a database of images to detect colour changes inside the cervix that point to patients at high risk for cervical cancer, which can be treated if caught in time, but which kills 60,000 women in Africa a year. Addressing a similar problem, Pratik Shah of the Massachusetts Institute of Technology (MIT) has developed a system to use simple cellphone or camera pictures—instead of expensive MRI or CT scans—to identify biomarkers that point to oral cancer. He told AFP that while AI systems typically need tens of thousands of data points to function, he has found a way to use only 50 images to train algorithms to identify a specific disease. "We believe our approach could be used to massively reduce the amount of data an AI algorithm currently consumes, and empower physicians to diagnose patients using simple images," he said. In the developing world, basic healthcare is often a challenge—let alone expensive medical screening or tests for easily treatable, preventable illnesses.
More here.