At the end of 2018, one of Google Cloud Healthcare’s executive advisors – Dr. Toby Cosgrove, former CEO of the Cleveland Clinic – took to the airwaves to herald the cloud healthcare trends of 2019. In his interview, he underscored interoperability, security, machine learning and telemedicine as major near-term opportunities for the healthcare industry in the cloud.
As SADA’s CEO recently mentioned in Tech Republic, 2018 saw increased adoption of the secure public cloud by healthcare organizations in the form of Google Cloud and G Suite, among others. Part of the initial hesitancy centered around security of protected patient information, an issue that Google Cloud has addressed in full force with its HIPAA compliant BAA offering and Google Vault service.
Rather than focus on past HIPAA-related concerns that kept providers from adopting Cloud, the question now becomes – what are the primary drivers to the cloud?
Interoperability Through Google Cloud’s Healthcare API
Last year, Google announced its Cloud Healthcare API, a suite of tools geared around securely ingesting and transmitting data from an to FHIR, DICOM, and HL7v2 standards in order to allow cloud data to communicate with third party systems – both modern EHR and legacy platforms.
Not only does this allow providers themselves to glean insights from massive data sets using BigQuery, Google Cloud AI, and data visualization tools in Google Cloud, it also allows breaks down data silos between organizations. As Dr. Cosgrove mentions in his interview, many providers are agreeing to de-identify data as it moves to the Cloud to make it easier for third parties to add value to the healthcare ecosystem by building big data solutions, such as machine learning models to aid with clinical decisions. It’s difficult to create tools that positively impact the healthcare industry as a whole if data is only shored up by provider.
Although healthcare standards such as FHIR are still emerging – many major providers have only adopted selected layers of the FHIR standard – Google’s healthcare API is a commitment to an evolving standard of data access and interoperability.
Breaking Down Interoperability Barriers Allows for Big Data Analysis in BigQuery, Data Visualization
In order to be able to draw insights from medical data, providers need to overcome problems with scale, security, speed and anonymity. By moving data off physical on prem servers, providers are able to overcome issues of speed and scale. With de-identified data, they are able to run analysis en masse without violating strict HIPAA rules such as purpose-driven minimum necessary use of data (only access a patient’s record if there is a business need to do so).
With powerful, user-friendly tools like BigQuery, analysts can combine hypotheses with actual encounter-specific data and unleash trends both clinical and financial. This data can be seamlessly presented for near-real time insight in Data Studio or another supported BI tool like Tableau.
Google Medical Brain Builds AI Tools from De-Identified Patient Data
Along with troves of medical data comes the Google Medical Brain team, a new branch of Google’s Brain team focused on aiding medical decisions using predictive AI models. While many medical decision are made in the moment on limited pieces of data present, the Medical Brain team’s initial research claims to be able to pick up on data points that are often lost, such as scribbled medical records. One goal is to eventually make medical decision models available through Google Cloud as a “diagnostics as a service” model similar to other tool suites on the platform.
As Dr. Cosgrove also mentions, the healthcare industry is trending towards a new focus on telemedicine. Not only would providers running telemedicine platforms be encouraged to build atop IaaS platform like Google Cloud, the medium could also lend well to automated medical scribing, a issue already said to be a point of interest for Medical Brain. While certain companies like Augmedix already use Google tech in the form of Google Glass to create a system where the doctor is unburdened from cumbersome dictations, AI-powered scribing could be an even more cost effective evolution of the transcription bottleneck.