Stanford Algorithm Can Estimate Chance of Mortality and Improve Palliative Care by Examining Patient Records

November 2017 - In a paper published in arXiv by researchers at Stanford University, a new algorithm based on deep neural networks can examine patient records and estimate mortality with a high probability of accuracy three to twelve months into the future. The software examines the patient’s health records for important indicators such as number of days spent in a hospital, severity of diagnosis, and medications prescribed. The algorithm is then able to prepare a report that the patient’s doctor can receive. 

 

As many as 80% of Americans say they want to spend their final days at home, but fewer than 20% do. As the population continues to expand and age, the need for quality and cost-effective palliative care will be exacerbated. Though still in a trial phase, the researchers hope their algorithm will help more people receive the palliative care of their choice and help inform medical professionals with higher accuracy and warning of the health needs of their patient.