NURSES鈥� notes on patients are a gold mine of information for medical researchers 鈥� after they鈥檝e been painstakingly stripped of any personal information. Fortunately, new software can already match the accuracy of people working in pairs, in a fraction of the time.
A team led by Roger Mark and Gari Clifford at the Massachusetts Institute of Technology have developed algorithms that use context to identify names that have not been encountered before. For example, words that follow 鈥淢r鈥�, 鈥渉usband鈥� or 鈥渄aughter鈥� are removed. It also deletes phone numbers, place names and dates (BMC Medical Informatics and Decision Making, ).
鈥淎lgorithms tend to find lots of instances of protected health information that humans might miss,鈥� says Clifford.
Advertisement
