Design principles for healthcare interoperability
As healthcare systems become increasingly more complex, interoperability between products and services as well as which architecture to pursue are critical decisions to make early in the process. The different solutions and ecosystems vary to a large degree but I believe a few fundamental principles are critical to success.
User Experience
- Patient-centric: The patient is the one common denominator across the majority of system interactions, put them at the center of your architecture.
- Patient-control: As governing bodies move more towards releasing ownership of data back to patients we’ll see more regulation around ensuring patients control how it’s shared, think GDPR. Make sure your solution takes this into account.
- Authentication and Authorization: a mechanism to allow patients to regulate access to their data is critical. OAuth 2.0 is an industry-leading option, simple to manage, Facebook allows sharing and revoking of data with apps and 3rd parties in this manner.
Data Sharing
- Share data freely: The concept of hoarding data and creating data silos is one that I don’t believe will age well. The solutions that will survive will recognize the value is in sharing data and that the well-integrated solution is greater than the sum of it’s parts.
- Open APIs: To free this data, follow an Open API playbook centered around industry standard message structure and transmission protocols.
- HL7 v2: event based messaging, been around for decades and isn’t going anywhere, fundamental to healthcare interaction.
- FHIR: standard gaining traction due to speed and ease of implementation allowing more complex bundles of information to be shared, lightweight and ideal for mobile.
- REST: the dominant transmission protocol for sharing data quickly and reliably.
Architecture Models
- Value-based architecture: Every architecture decision should take into account this question - What value does this add to the end user? For example, does an end user truly care that the information model houses data in a Federated model utilizing IHE standards to interact with data? Provided that the data is securely stored, probably not, but a user will care if performance suffers as a result of this decision.
- KISS: Keep things simple, store as much information centrally and share and keep all components in-sync via a publish and subscribe mechanism.
- Modular over Monolithic: Data being shared elegantly and seamlessly between services allows for a flexible solution. Choose the best of breed for each component and let each vendor focus on their value-add as opposed to asking one product to deliver everything at a mediocre level. If the data can’t be shared well then accept a lower quality product with less vendors.
What principals would you add?
Founder, TopCoverHealth - Healthcare design and Global Patient Advocacy.
6 年The discussion below has 'evolved' into a discussion on pt centric or clinician centric or hybrid solutions to Patient Records.? First we have to solve the technical parts to interoperability.? Once the rail tracks are in place the data formats can be worked on.??
Clinical Systems Engineer at Renown Health
6 年Agreed. In my opinion the most important part of this solution is Open APIs: To free the data, follow an Open API playbook centered around HIMMS standard message structure and transmission protocols. NOT an OEM driven standard.
Interesting to see no mention of payload, e.g. terminologies, classifications and unstructured information need to be explicitly considered and specified in solutions or true interoperability will not be achieved.? These typically may include SNOMED CT, International Classification of Diseases, LOINC (particularly in North America) and profession specific terminologies (although theses are diminishing in favour of multidisciplinary ones.
Founder, TopCoverHealth - Healthcare design and Global Patient Advocacy.
6 年In the year 2018 it is strange that we still haven't agreed and standardised a simple (4 or 5 fields) data set across the industry.? That would be a start.? Technically the solutions for interoperability exist but an agreed data set is a good start.....begin with the end in mind, define the purpose and requirement of that data for clinicians and patients.? The reasons for failure to agree and implement technical interoperability are more complex and probably relate to competitive forces.? Perhaps patient-centric 'control' or ownership of records may be part of the answer but some access or break-glass mechanism required.? Good discussion anyway...it's an important topic.? Blockchain may add some new qualities by bringing an ecosystem approach.
AVP Business Intelligence at Care Design NY & Partners Health Plan
6 年Understanding that patient knowledge is vastly different than professionals’, it seems that design would be the key to supporting a patient-centric solution. If trained healthcare professionals guided the design to identify what data would best inform their decisions, the value would be increased plus the mere experience of patients engaging in their own health tracking holds tremendous power in general.?