Google’s deep learning model to manage EHR data
Google is exploiting deep learning and machine learning techniques over traditional methods to regulate better healthcare outcomes.
The Google I/O 2018 Conference revealed that a Google deep learning model has reportedly surpassed traditional methods to predict healthcare outcomes, through its efficient artificial intelligence and machine learning.
The move by Google is primarily to prove its functioning in healthcare with its model and algorithms. The company has teamed up with Fitbit, to combine the Twine Health platform with Google’s Healthcare API aiming at connecting user data and health records to improve the management over chronic diseases, as reported by Health Care Dive. It’s clear that the company is definitely interested in health.
The company explained how it is going to be effective in electronic health records storage, as reported by a paper published during the Google I/O Conference.
In order to explore the methodologies of deep learning models which could be applied to hospital patients, Google teamed up with Stanford, UC San Francisco and The University of Chicago. The data required is very essential for the process.
Recently, there have been various efforts to make use of AIs and machine learning to make advances in healthcare. To predict cardiovascular risks, Google teamed up with Verily, one of its healthcare subsidies to develop an algorithm which can enable doctors to analyze a patient without the need of a blood test. The software can easily deduce the data acquired with the patient’s age, blood pressure levels, and figure out if they are a smoker to make an estimate if they are in risk of suffering any cardiac event.
And Google is stepping its foot in the realm to make its impact. Hence, it’s diving into using Google cloud along with data pipers for providers as well as payers. Google’s parent company, Alphabet is already all over healthcare referring itself as an AI and machine learning venture, as reported by CB Insights.
Deep Variant, which was launched last December by the company, was established as an open-source tool utilizing AI to form a genetic blueprint of an individual by sequencing data acquired.
The major goal is to use the information to create a systematic guide to a patient’s diagnosis and treatment further. The tech giant also partnered with Moorfields Eye Hospital NHS Foundation Trust in the UK, with DeepMind to mine Big Data in the aim to learn if early detection and treatment degenerative eye diseases are possible by ocular scans that can be analyzed by machine learning.
Google’s deep learning models were able to scale a system capable of predicting unexpected readmissions, inpatient mortality, and discharges, as reported by ZDNet.
The study conducted involved EHR data from 216,221 adults who were hospitalized for at least 24 hours. And the combinations recorded up to 47 billion data points.
By hypothesizing how the techniques like deep learning approaching in the entire EHR could be translated to be beneficial for healthcare. That includes free-text notes to get an estimate of a broad range of clinical problems and predictions over the traditional methods.
It aimed towards formulating a system by harmoniously simulating various inputs and predict events concerning health. Although the company has made it clear that this is just an initiation, which requires a lot of effort considering the amount of processing required. By using retrospective data, it is functional to create predictions and utilize the deep learning models.
Image credit: ai.googleblog.com
