A Google patent application published on Thursday by the US Patent and Trademark Office gives a few more details on a predictive EHR system first highlighted by the tech company last May. The application was filed by Google in July 2017, and has not yet been granted.

As described in the application, Google’s system can aggregate and store EHRs for a diverse population, while compiling each individual patient’s records into a single chronological document. A computer or computer system running deep learning models would then use this collection of data to guide predictions of future health events, and to better contextualize the collected data from an individual’s record to highlight pertinent past events on an EHR. Each of these would be displayed to the provider on a desktop, tablet or smartphone display, helping them identify areas of concern or intervene prior to future events.

Along with an outline of its major components, the application also suggests that the system would be capable of aggregating records from multiple institutions and data formats. Examples of future clinical event predictions cited in the application included an unplanned transfer to intensive care unit, ER visits or readmission within 30 days of discharge, inpatient mortality and atypical lab results related to a number of conditions.

A representative from Google told MobiHealthNews that the tech company was not providing any new comments on the system, but pointed back to the company’s prior statements in a May blog post.

Here, Dr. Alvin Rajkomar and Eyal Oren — both of whom were among the patent application’s coauthors — highlighted the results of a Google, UC San Francisco, Stanford Medicine, and University of Chicago Medicine investigation that employed 46,864,534,945 retrospective EHR data points collected from 216,221 adult patients hospitalized for at least 24 hours at two US academic medical centers. From these data, the team’s deep learning models were able to predict upcoming in-hospital mortality, 30-day unplanned readmission, prolonged length of stay, and all of a patient’s final discharge diagnoses with an accuracy that outperformed traditional predictive models across the board.

Why it matters

“There is a need for systems and methods to assist healthcare providers to allocate their attention efficiently among the overabundance of information from diverse sources, as well as to provide predictions of future clinical events and highlighting of relevant underlying medical events contributing to these predictions in a timely manner,” the patent application’s authors wrote in the document. “The present disclosure address a pressing question facing the physician in the hospital, namely which patients have the highest need for my attention now and, at an individual level, what information in the patient’s chart should I attend to?”

Both the application and the results from last year’s study suggest that work on the system is still ongoing. However, Rajkomar and Oren noted in their blog post, its heavy focus on accuracy and scalability (i.e. interoperability) are an especially promising roadmap for machine learning applications in healthcare.

“Doctors are already inundated with alerts and demands on their attention — could models help physicians with tedious, administrative tasks so they can better focus on the patient in front of them or ones that need extra attention? Can we help patients get high-quality care no matter where they seek it? We look forward to collaborating with doctors and patients to figure out the answers to these questions and more,” they wrote in the post.

What’s the trend

A handful pilots and products focused on predictive health analytics have cropped up over the past couple of years. Atlanta’s Grady Health System detailed an example of the former in October, noting how it was able to save $4 million in prevented readmissions using a combination of AI and patient-level community interventions. As for the latter, Silicon Valley startup Lumiata offers an analytics platform that helps providers and plans identify at-risk patients before the cost of their care rises.