The partnership builds an end-to-end solution to access and transform clinical data from 300+ million patients into analysis and AI-ready data in order to materially improve care delivery.
Healthcare organizations (HCOs) are increasingly using data to make critical, often life-changing decisions. But two main challenges have stood in the way of more HCOs effectively leveraging data at scale: 1) Data Access and 2) Data Transformation.
Today we’re incredibly excited to announce a strategic partnership that solves both of these challenges, with a single open-source data pipeline that accesses data via Metriport’s clinical data network and transforms it via Tuva‘s Data Factory, into quality-tested data that’s ready for analysis and AI.
HCOs routinely have a difficult time accessing all the medical records and clinical data for their patient population. Patient medical data typically resides in disparate silos, fragmented across various healthcare providers and systems, each employing their own formats, making compatibility and standardization major challenges to interoperability. Throw stringent privacy regulations like HIPAA and complex technical barriers into the mix, and most HCOs are stuck with limited access to patient information, often resorting to less-than-ideal methods such as fax, phone calls, and lengthy patient surveys to retrieve patient records. These slow and cumbersome practices often result in valuable patient information being missed, and greatly increase the likelihood of retrieving incorrect information — both with potentially grave consequences.
Metriport is solving the Data Access problem via a single, open-source API that gives HCOs access to over 300+ million individuals’ clinical histories. Metriport takes the basic demographic information of a patient — name, DOB, and address — and returns comprehensive, consolidated medical data, scouring the entire US for the patient’s full clinical history. By standardizing, de-duplicating, consolidating, and hydrating the data with medical code crosswalking, Metriport’s FHIR-native software ensures clinical accuracy and completeness of medical information, so providers can get a rich understanding of their patients’ health.
Once HCOs have access to the right patient data, a second problem emerges: Data Transformation. Raw healthcare data requires a tremendous amount of transformation before it can be used to answer even simple questions (e.g. how many patients had a visit to the ER in the past month), let alone more advanced applications, like training a model to predict patients at-risk for an ER visit in the future. And doing this transformation requires a laundry-list of healthcare-specific tools and skills. (Nikhil Krishnan does an excellent job describing these challenges in a recent blog post on Tuva).
Tuva Health is solving the Data Transformation problem via the Tuva Data Factory. Data Factory preprocesses and unifies an HCO’s raw healthcare data sources into a core data model, enriches that data with data marts that are designed for analytics and AI, and provides in-depth data quality intelligence. The data is then ready to be used by clinicians at the point of care (e.g. displaying patient care gaps and risk gaps), by analysts for provider benchmarking and care navigation analysis, and by data scientists for building patient risk identification and stratification models, to name just a few use cases.
How It Works
With this partnership, eligible HCOs can get turnkey access to analysis-ready data on their patient population through a single agreement with Tuva. Here’s how it works:
- The HCO signs up for Tuva Data Factory.
- Tuva Data Factory connects to any existing data the HCO already has (e.g. claims or medical records) and queries Metriport’s clinical data network for additional relevant patient records on the HCO’s patient population.
- Tuva Data Factory unifies the patient data from Metriport’s clinical data network with any existing data from the HCO, and normalizes, enriches and quality-tests the combined data.
- Tuva Data Factory then transmits the transformed and quality-tested data directly to the HCO’s data warehouse where it can be used for analysis and AI.
Today, eligible HCOs must be either providers or technology companies that directly serve providers to access Metriport’s clinical data. In the future, this will expand to include other non-treatment use cases, such as, payment, operations, and individual access services (IAS).
Why Open Source
If you’ve worked in healthcare data for a long time like us, you’ve seen countless companies build and rebuild the same tools to solve the same problems over and over. These tools are typically black-box and you have no idea how they work. This is a major problem, because small differences in how data is transformed can cause massive differences in downstream analysis (you can literally get opposite answers that are statistically significant).
We believe that HCOs should have full control of and transparency into their healthcare data without having to build from scratch. That’s why we’re both building open-source.
For example, Metriport’s record location service code that connects to thousands of endpoints to retrieve patient data is open-source. Customers and users can see exactly how it works and suggest changes that can improve the match rate for a given end point. You can check out all of Metriport’s code on GitHub and explore their developer docs for a detailed guide to their API.
Another example are Tuva’s data marts, which enrich data with measures, groupers, care pathways, and risk models. Customers and users can see exactly how these measures and other concepts are defined in code and suggest changes or fork (customize) if they need a different definition. You can check out all of Tuva’s code on GitHub and explore their Knowledge Base, which is an open source book on healthcare analytics.
Interested in learning more? Contact the team at Tuva Health here.