Google recently announced expanded partnerships with three top-level medical research hospitals: Stanford Medicine, UC San Francisco, and University of Chicago Medicine. The goal of these partnerships is to explore how machine learning can use data mined from electronic healthcare records for improved results in clinical settings.
Google Brain researchers noted that machine learning is now entering the stage where it can accurately predict medical events such as when a patient might need to be hospitalized, how long they will stay, and when a normal course of treatment might not prove effective. In this way, machine learning could be used to anticipate the patient’s needs before they arise. Specific medical conditions cited included urinary tract infections, pneumonia and heart failure, as well as a general goal of preventing healthcare-associated infections, mistakes in medication, and hospital readmissions.
Google also aims to improve the way data is communicated and distributed. Following Health Level 7’s FHIR standards, Google intends to open pathways of communications between hospitals and researchers, automating standardization and data exchange. Researchers assured that patient privacy is a top priority in the process. Data is first de-identified before sharing, and then secured using Google Cloud’s infrastructure in a manner compliant with HIPAA privacy rules.