An Open Letter on Health Data
What we're building and why
At Automate Medical, we believe that one of the biggest problems in health care is data interoperability. Despite decades of digitization of medical records, most medical data remains shockingly rudimentary and siloed. Patients have delayed access to care when different clinics and providers cannot quickly and properly synchronize their records. Public health and new medical research are stymied by unnecessary complexity.
Automate Medical is on a mission to liberate closed silos of health data and make data exchange easy for everyone. At Automate Medical, we build open source developer tools for health care standards like FHIR with modern languages and frameworks. The future of health data is interoperable.
Interoperability Is Finally A Priority
After many decades of digitization, starting from scans of ticker-labeled manila folders, medical records have evolved. Today, both public health authorities and payers are increasingly demanding standardized interfaces for medical records. Interfaces that work across health care provider settings, regions, and networks.
COVID unearthed systemic data sharing problems at the institutional level. Significant public interest in health data has emerged. At the same time, value-based purchasing and care programs are growing rapidly. Both health care payers (public and private alike) and their patients increasingly demand that outcomes be factored into payment models. Fee-for-service contracts are a declining segment of the market.
Reporting patient outcomes beyond just the services rendered is an important component of all alternative payment models. The full gamut of patient data from discharge events all the way to readmission rates for specific diseases has to be queryable to meet this demand. In many cases, this isn’t cheap. The core problem is that health data is spread across multiple systems, providers, and formats. Getting health data into one place, and making it interoperable, is a really big deal.
A McKinsey report from 2020, The math of ACOs, suggests that organizations with value-based care as a core component (like ACOs), may spend “0.5 to 1.5 percent of the total cost of care [per patient]” on analytics and data needs related to reporting.
The Value-Based Market Keeps Growing
In the United States, health care is being structurally transformed by the creation and growth of value-based purchasing by both public and private payers. HCP LAN’s 2019 APM Report suggests that ~60% of all US healthcare spending is at least performance-incentivized for value, and 5% is reimbursed on a full-capitation basis (i.e. periodic, per-patient payments from the payer, despite how many times the patient comes in for treatment or how many services are needed).
Similarly in Canada, the same movement towards value-based health care is playing out. The overall share of spending via alternative payment models more than doubled In Canada between 2000 and 2018. In some provinces like Alberta, the government is even going a step further and exploring expanded capitation agreements.
Lack of Interoperability Creates Nightmare Scenarios
When we’ve spoken to people working in health care, we’ve been shocked by just how rudimentary many medical record exchange processes really are. We’ve heard stories of medical records being faxed in large PDF bundles from one clinic to another. Administrative assistants act as “human glue” between systems and manually check the upcoming patient visit calendar to determine whose records are missing.
Health care providers are kept awake at night worrying about nightmare scenarios with data exchange. Like that of an oncology patient in a rural community, who could drive for hours for a major visit at the only regional hospital with an oncology clinic. Weeks of worry turn to anger and tears when they, and the local medical team, learn on arrival that their medical records were never transmitted. Worse yet, even if they were transmitted, the documents have been exchanged as a 70-page PDF requiring careful, manual specialist attention to read.
Lack of interoperability creates problems at other stages of the patient journey as well. Clinical decision support systems are only as good as the data they receive. Many decades of promises of “computable medicine” have gone by without delivering the goods.
When health care providers “burn-out”, they cite interoperability and data problems as a key reason:
A 2018 study by the National Institute of Health (NIH) concluded that more than half of U.S. physicians are experiencing burnout, citing “treating the data, not the patient” and “electronic health record woes” as primary causes.
Public Health and Novel Medical Research Depend Upon Improved Access
There are major issues for health data at a macro level. The scope of advances we can make in the field of medical research is increasingly bounded by the scalability and portability of health information systems. We hear so much about how important data is, but the practical realities of health information systems today make realizing its potential difficult, tedious, and expensive.
Medical research teams universally describe painful processes of “data consolidation” - unifying common data elements between multiple health information systems. The time cost of this activity ultimately limits the evolution of diagnostic medicine and the rate of new medical discoveries.
The same can be said for any public health monitoring activity, including our ability to respond to emerging health threats like pandemics. There is currently a great deal of focus on vaccine credential management for a post-COVID world. If governments wish to pursue such projects seriously, they will need to contend with the realities of limited interoperability in existing systems.
To Get Interoperability, We Need Better Developer Tools
We believe that having substantially better developer tools for health data is the key to interoperability winning.
These systems can only become easier to build if the architectures, frameworks, and tooling we use to build them are developer friendly. It turns out that we don’t need to boil the ocean to get there. Better interoperability solutions already exist and are growing quickly. HL7, the health care data standards organization, maintains a specification called FHIR - Fast Healthcare Interoperability Resources.
FHIR is an openly documented, detailed description of a way to structure, query, and exchange the kinds of data that might exist in a health information system. FHIR enables new levels of interoperability by providing a modern “lego blocks” approach to structuring health data and defining exactly how the exchange of health data should work. It does so by using standards and best practices familiar to developers as lessons learned in other industries. This reduces the overall complexity of health data projects - making them faster to deliver and quicker to see results.
We believe that Automate Medical can accelerate the adoption of FHIR by building new developer tools using modern languages and frameworks with an open source license. Automate Medical will be building in the open with the belief that a rising tide will lift all boats. Our first release will be a brand new set of TypeScript definitions, programmatically built for every released version of FHIR via the blueprints (JSON-Schema) that HL7 provides. We plan to build a FHIR resource validator, and eventually an entire server implementation in TypeScript.
We’re building with TypeScript because it has the fastest-growing developer audience amongst the most popular languages on GitHub. Health data is important enough that the best developers in the world should be able to use the best tools possible.
Our vision is to develop FHIR tools focused on advancing clinical decision support (ex. first-class support for CDS Hooks and CQL). We'll take advantage of new infrastructure options like being able to run the REST interface of FHIR in serverless functions like AWS Lambda (one-click developer experiences). Our FHIR server will be easy to run locally, easy to build from, and well documented.
Building better developer tools means raising the bar on health data for everyone. As health data becomes more interoperable, everyone benefits. Patients get access to better care outcomes. Medical professionals get to go back to treating patients, not data - a big reason for burn-out. Payers get to deliver on the promise of value-based purchasing. Public health authorities and researchers get to develop new population-level health insights in critical fields like pandemic response management.
We’re extremely excited about the possibility of using developer tools as a beachhead to push interoperability forward as hard as we can. If you’re excited too, please subscribe to hear more from us.
p.s. You can also email us directly at firstname.lastname@example.org with your ideas, questions, or to offer to lend a hand. We couldn’t be more excited about where things are (finally) moving in health tech.