Can you trust Google with medical data?
Artificial Intelligence can do a lot in medical research and treatment, but this is probably at the expense of our privacy. It is already having an impact that Artificial Intelligence can help in medical cases. IBM's deep cloud intelligence, Watson, has been helping doctors at several cancer hospitals in the United States with diagnostic consultancy for five years. It does what doctors do. Knowledge about a patient's medical history is combined with their symptoms and recent history. This is done to identify the most likely cause of illness. It is such a success that it is expanding to hospitals in India and China, and soon to European hospitals.
What AI also does is analyze new findings in new editions of approximately 290 journals, 200 textbooks and a total of 12 million pages of text. About the health care that all humanity produces. Engineers approach healthcare as fundamentally a text analysis problem. And machines can stay up to date in a way that an overworked ER doctor cannot.
But what are the minus points?
But reports are not the only thing that hospitals will pass on to AIs. Big data is the computer analyzed revelation of patterns and trends. These will not be immediately visible to human researchers. This also means using patient records as sources. But whatever the benefits, should we be comfortable with robots snooping through our most personal data?
The patient files that are handed over to the hospital and, if they use AI, also to the Google DeepMind deal. These include minute details beyond just pathology examinations, names and addresses. Down to the locations that patients stayed in each hospital and the times when family and friends came to visit them. It is not anonymized, or even pseudonymised. These are medical records, perhaps our most personal information. At a time when there is so much data to be leaked or lost and stolen. To gain the benefits of machine learning in healthcare, medical authorities are allowing private companies and their robots into some of the most sensitive databases available.