Google is working on new predictive technologies that may be able to predict when a hospital patient will pass away.
The research is being done by Google’s “Medical Brain” team. And in at least one case, the neural network trained by that team was able to more accurately predict a patients’ chance of dying than the hospital’s own systems, Bloomberg reported.
The system was highlighted in new research published by Google in May. The report focused on the health care potential of artificial intelligence and machine learning.
By Google’s account, the neural network analyzed 175,639 data points and offered that the patient had a 19.9 percent chance of dying. The hospitals’ computers gave a 9.3 percent chance. The patient died a few days later.
Bloomberg noted that what “impressed” doctors and medical experts the most was the fact that the trained systems could sort through data that was particularly hard to take into account — such as notes buried in computer documents or hand-scribbled on filed charts.
The system gave a prediction that was more accurate in much less time than existing platforms used by hospitals. It even pinpointed where it mined certain overlooked data points.
Of course, this data — along with all of the information that Google already collects — could pose a privacy concern. For its part, Google told Bloomberg that the data is “anonymous, secure, and used with patient permission.”
But the privacy aspect could become murkier as the technology progresses. Especially since we’re not sure if or how Google’s medical data will remain separate from the data it collects online.
Like other major technology firms, Google seemingly has some broad ambitions in the health care and medical fields. Along with the noted system, the Mountain View company is also working on platforms that can detect existing conditions or a person’s risk of developing diseases.
It’s still early to see how any of this pans out. Even top Google officials say that the technology is still in its nascent phase. But at the pace of current technological progress, it’s probably safe to venture that neural networks and AIs could become a common sight in doctor’s offices and hospitals across the globe.