- Google Paper Stirs Interest, but Not Seen as Transformative Politico by Darius Tahir — By throwing some of its best engineering and medical minds at vast stores of clinical data with the help of powerful computers running for hundreds of thousands of hours, Google appears to have produced a model that accurately predicts patient deaths, hospital readmissions and other health-related events. The model won't transform medicine. But academics and other technology mavens think the methods described in the paper serve as a prototype for future work in predictive models, in areas like end-of-life care. The researchers analyzed seven years of de-identified clinical data from hundreds of thousands of hospital patients at the universities of Chicago and California. They discovered indicators of health events with a degree of accuracy that "outperformed state-of-the-art traditional predictive models," according to the paper. Still, as much as the Google work might improve the efficacy of prediction, it seems that prediction has limited value in our health system. A doctor can be told a patient will die tomorrow, but it doesn't mean the doctor can do anything about it.