In many industries, we’re sold solutions based on business needs. In healthcare, these solutions may help us manage population health or the claims process, for example. At Maven Wave, we argue that the platform is of primary importance and set up in a scalable fashion to quickly provide several solutions based on needs. We previously shared how data enabled healthcare systems to get ahead of COVID-19, but now we will dive into the specifics of how to build an effective a data platform.
The foundation of the data platform is always security. Some are resistant to using cloud platforms, fearing that data housed there is less secure than on local servers. We find that combatting this fear is a matter of education: education about how the controls and capabilities work, and education showing the differences in cloud security versus on-premises. This is imperative to understand because cloud platforms are key in organizing and analyzing data in a way that’s accessible and scalable to many people.
Point solutions can only be developed and used for specific use cases. We’ve found that 80% of the process in developing solutions is ensuring security needs are met. Using the right platform allows that to happen more efficiently. The data must flow in secure network pipes, be encrypted and use secure API protocols.
In healthcare analytics, choosing the best use cases can follow this process:
Develop a case prioritization heat map: This includes the list of recommended projects mapped by ease of implementation and return on investment (ROI). Which projects can you do quickly and easily that provide the best value for the outcomes you seek?
Current state of analytics architecture: What is your state of cloud readiness and current capabilities? Do you have the data to use in the analyses? Do you have the ability to access data quickly and execute projects to gain consistent wins?
Organizational delivery maturity evaluation: This is the change management piece. How ready is your organization to take on these projects, in terms of knowledge about the data you maintain, and knowledgeable employees to evaluate it?
Actionable roadmap: Document your roadmap, including timeline and milestones, cost and schedule.
How accessible is your data?
One population health use case is understanding needs for those with health disparities, who have greater likelihood for poor outcomes with COVID-19. During the pandemic, a fair number of people are not accessing healthcare when needed, or accessing it in different ways than expected. That means they are at high risk of needing help at a later point than they would otherwise. The data sources might be ambulatory clinics or urgent care centers, where health systems don’t necessarily have access. These facilities traditionally are not connected to the EHR of the hospital or don’t share data through an exchange.
Telehealth is another example where obtaining data can be difficult. Some devices used for telehealth aren’t connected to the records systems, making the data difficult to access from a research perspective. Pulling data from clinical settings can provide important clinical information for developing COVID-19 treatment guidelines, but for studies that aren’t approved by an institutional review board (IRB), that makes it difficult.
To provide treatment based on current trends and evidence, researchers want to gain real-time clinical insights to cycle those back into clinical guidelines. Clinicians treating COVID-19 patients make decisions based on low oxygen saturation levels, deciding whether to use ventilation or high pressure oxygen, based on COVID-19 data. Getting access to the data and using it in a way that’s best for patient outcomes can be difficult.
How to build the data platform
After understanding the requested use cases, determining what data should be targeted, and addressing security needs, it’s time to build the data platform. The platform would first ingest and integrate data, then transform and store it, before it is analyzed.
Ingest data: This is a newer term than its predecessor, data acquisition. Data is pushed or pulled from other sources, with experts determining what formats and tools are needed to support this process. Multiple levels of data security are needed, along with multiple touchpoints. Data should be identified including cordoning off protected health information (PHI). A variety of services help with data ingestion, whether streaming or batch level. It is important to ensure that the data comes from the right time and place so it can be used properly.
Transform data: Data quality rules need to be matched to the use cases. Again, good data is better than perfect data to move forward. In these situations, we can’t spend weeks purifying the data as the disaster continues. Using the 80/20 rule, ensure that 80% of the data is in great shape, and that may be good enough. Data in cloud storage is not the same as a data lake. Users can choose the data form, including raw data. Cloud storage is inexpensive and configurable, and is the foundation of any data project. Security is involved in data transformation as well, as that can’t be compromised. The Data Loss Prevention (DLP) tool in Google’s API helps identify sensitive data like PHI and social security numbers.
Analysis: This is the stage for gathering data insights. With rich data like we’re getting during COVID-19, any analyst can discover insights in the data when given free reign using data integration tools.
A cloud platform like Google Cloud Platform provides a secure foundation to collect, manage and analyze data needed for COVID-19. Whether a healthcare system is fully on the cloud or just thinking about it, Maven Wave can help create an architecture to help healthcare systems quickly and accurately get the best insights from their wealth of valuable data to make it truly actionable. Contact us to connect with our experts and learn more about how we can help build an effective data platform for your healthcare system.
Get the latest industry news and insights delivered straight to your inbox.