Colorado Center for Personalized Medicine capitalizes on Google Cloud scalability for rapid discovery during pandemic | C2C Community

Colorado Center for Personalized Medicine capitalizes on Google Cloud scalability for rapid discovery during pandemic

  • 12 July 2022
  • 1 reply
  • 85 views

Userlevel 2

https://edu.google.com/intl/ALL_us/why-google/customer-stories/ccpm-googlecoud/ 

 

CCPM capitalizes on Google Cloud scalability for rapid discovery during pandemic

With extensive analytical tools and a 6M patient record research data warehouse on Google Cloud platform, the Colorado Center for Personalized Medicine delivers “no-barriers” data and analytics under unforeseen scenarios like the COVID-19 pandemic.

In March 2020, as the novel coronavirus began to spread across the country, the University of Colorado Anschutz Medical Campus and UCHealth, a system of 12 hospitals and hundreds of clinics, suddenly needed to prepare for an expected surge of patients with advanced respiratory illness. How many ventilators would they need, and when? Needing urgent answers, the pandemic planning team approached the Colorado Center for Personalized Medicine (CCPM), a joint project with UCHealth, for help.

Dr. Michael Kahn, Emeritus Professor of Informatics and Data Science in the Department of Pediatrics, had spearheaded the Health Data Compass warehouse at CCPM for scenarios like this one, where clinicians require comprehensive, reliable data analysis with a fast turnaround. Since 2015, Compass has integrated anonymized clinical information from the electronic medical records at UCHealth and Children's Hospital Colorado, provider billing data from CU Medicine, and -omics data from CCPM. It collects and standardizes health information from six million patient records across the state—from genomics to annual physicals.

When the UCHealth system reached out to CCPM they were ready with the data and resources for prompt results. Initial data sets were extracted within 36 hours of the initial request; a working model was available within the first week that aided clinicians in identifying patients’ therapeutic needs in the coming days and weeks. “It was spectacularly quick from initial conception to useful first-version models,” Dr. Kahn reports.

What medical directors are saying

 

We want to change the model for personalized care and improve clinical practice, not just research.

Dr. Michael Kahn, Director of Health Data Compass and Emeritus Professor of Informatics and Data Science in the Department of Pediatrics, University of Colorado Anschutz Medical Campus

 

From building capacity to driving impact with the cloud

In 2017, CCPM began migrating their Compass data analysis and storage from on-premises servers to Google Cloud’s cost-effective, scalable, and secure infrastructure for cloud computing. As Compass’ data grows, Google Cloud automatically scales to support it. Sarah Davis, Product Development Manager, says, “by moving to Google Cloud we no longer had to manage our infrastructure maintenance and service. Even security and HIPAA-compliance is built into Google Cloud. So we were able to focus on developing research tools and formatting the data to make it as easy to use as possible.”

In March, Davis and her colleagues at CCPM worked closely with UCHealth and CU faculty to refine their model based on ever-changing clinical and capacity conditions. The hospital collected patient data in real time and updated demand for ventilators every 72 hours; the model created risk profiles for four days, seven days, ten days, and fourteen days in advance. Dr. Kahn says that they had daily conversations about whether to convert other medical units into ICUs, and having predictive models made a difference. Davis adds that “during a very stressful period, having a model that produced a score for risk helped keep everyone’s emotions in check. And we showed that we could deliver results in a crunch.”

 

More compute, more data storage, and over $2.9M in savings with Google Cloud

Now CCPM and Compass are building the user-friendly tools and workflows affiliated researchers need to mine this comprehensive resource for clinical insights. Using Compute Engine and BigQuery they developed the Eureka platform, which lets CU and UCHealth faculty quickly and easily search for therapeutic insights to better serve their patients. Working with BC Platforms, a leading provider of bioinformatics platforms for personalized medicine and drug discovery, the CCPM team has also streamlined their workflows to produce genomic data that is more accurate and detailed. BC Platform’s tailored solution manages Compass’ bioinformatics pipelines, ensuring that data is properly certified, secure, and reliable at every stage of the process. Anoop Yamsani, Solutions Architect and Data Engineer for Compass, says, “the technology stack helps us a lot. With Google Cloud we leverage their security team and backend so we can focus on user access. We can implement and test immediately. We can get results in days, not years. That gives researchers flexibility.” Davis adds that it also costs less: “you can spin up your resources and only pay for what you used.” Based on initial analysis, CCPM estimates that they were able to save more than $2.9M over their previous on-premise solution while also being able to consume 150% more compute and 277% more storage capacity.

Compass has already delivered over a thousand completed analyses to researchers who have queried the data. Dr. Kahn is excited to keep moving forward: “we intend to continue to exploit the Google healthcare ecosystem to customize and adapt tools dynamically, like Google’s Healthcare API. We’d like to incorporate images like X-rays and CTs that are crucial for COVID-19 patients, as well as Natural Language Processing and text identification.” Davis agrees on the potential ahead for the field as a whole: “No one academic institution can improve healthcare for everybody. It takes the collective, and as we have learned, these types of advanced technology give us the jump start. The faster we can transfer biomedical research into medical practice as an industry, the better we can deliver improved personalized care to those in need." “There’s a last mile gap we were trying to solve,” Dr. Kahn adds. “It takes a long time to transfer biomedical research into medical practice. We want to change the model for personalized care and improve clinical practice, not just research.”

What data engineers are saying

 

With Google Cloud we leverage their security team and backend so we can focus on user access. We can implement and test immediately. We can get results in days, not years. That gives researchers flexibility.

Anoop Yamsani, Solutions Architect and Data Engineer, Health Data Compass

1 reply

Userlevel 7
Badge +29

What an amazing and detailed article, @scano! Thank you for posting this!
 

I will also tag @wesley83@PRANEETHRACHUMALLU and @tuliani, as I remember you are also working in healthcare related fields! :)

Reply