Incorporating accurate prognostic information into clinical decision making could advance evidence-based, person-centered healthcare by more effectively targeting healthcare services to those patients most likely to benefit. Here we describe the deployment of predictive models for five year life expectancy of patients, built on electronic health records (EHR) of nearly 7,500 patients aged 50 and above, with one or more visits to a large, academic, multispeciality hospital in a year, exploring more than 75 modeling configurations. The online web-tool takes a non-redundant subset of 24 patient attributes as input and generates a patient-specific prediction of 5-year survival. The online five year life expectancy calculator is available at http://info.eecs.northwestern.edu/FiveYearLifeExpectancyCalculator.