A simple use case: a care provider needs a new clinical application. Using a traditional approach and depending on the complexity, this could take months or even years of development and testing, and a lot of developers, QAs, UX designers, and clinical domain experts. The front-end development timeline shortens with classic low-code platforms, but ensuring compliance and functionality remains challenging, and you need domain knowledge experts. This is where the data-first approach comes to shine.
In the dynamic world of software development, three primary approaches stand out: traditional coding or code-first development, low-code development or UI-first, and an emerging, more advanced method of domain-driven development or data-first. Each has its strengths and weaknesses, particularly in the context of healthcare. We explore why the data-first approach offers unparalleled benefits for healthcare application development and how to amplify it by combining it with low-code tools.
Traditional development: Code-first
The traditional approach involves writing code in Java, C++, Python, and others. First come application logic and business rules, and then the development of other components from that code, such as database schemas, APIs, and front-ends. This method, while powerful and flexible, demands significant expertise in coding and a deep understanding of healthcare-specific requirements. Development cycles can be lengthy, and ensuring data interoperability and standardisation is a complex, resource-intensive task. In most cases, these have proprietary data models.
Low-code development: UI-first
Low-code platforms aim to simplify and accelerate the development process by allowing users to build applications through visual interfaces. The screen and what the user sees is the start, and then the business logic is reverse-engineered, and the data model is auto-created – by default, proprietary to that application. These tools are advantageous for industries like telecommunications or banking, where business lgic can be relatively straightforward, and data models are simpler than healthcare. However, in healthcare, the need for more standardised data models and extensive domain knowledge can hinder the effectiveness of these platforms. Applications built this way often need more structure, interoperability, and compliance with healthcare standards.
Domain-driven development: Data-first
There is a concept called “Headless EHR”, which was until recently more theoretical than practical. Today, however, we are witnessing the arrival of true headless EHRs – both those designed to be headless from the ground up (platforms) and existing EHR systems swiftly opening up their core functionality through a wide range of APIs. A headless EHR is a type of EHR system that separates the back end (data and functionality) from the front end (user interface). This approach allows developers to create custom applications and interfaces or integrate with other systems using the EHR’s data and capabilities without being tied to a specific user interface.
However, with the above-mentioned options, developers still need to develop the application, or domain experts need to define the models for the application and define the business rules. But is there a better way, somehow combining it all?
Data-first low-code development
Unlike traditional low-code platforms, this method starts from the data model and the domain knowledge within. A similar concept to headless EHR – but requires no coding. While we can integrate to the backend via APIs and build applications on top, here we have a drag-and-drop development environment as well, which allows us to take that domain knowledge from the data and put it on the screen.
Comprehensive data model ensures all necessary fields, clinical validation of the model, and use of data standards are included. To achieve this, you need a robust framework, and here, openEHR and FHIR come into play. One of the core advantages of the openEHR approach is the clear separation of clinical domain expertise from software development. Using openEHR, clinicians and clinical informaticians are empowered to define and manage clinical knowledge and workflows. At the same time, the application development teams can focus on building robust applications with the best user experience.
There are two significant benefits of this approach. Leveraging the pre-built, standardised data models, such as those offered by openEHR and FHIR, clinically verified and interoperable applications can be built from the outset. This introduces an entirely new term in data exchange, INTRAoperability, meaning seamless data exchange within the data persistence layer itself and not on the level of interfaces and integration engines. Building a clinical application doesn’t require development knowledge, as the low-code tools significantly reduce development efforts. Moreover, clinical knowledge is already embedded into the data model, further accelerating development.
Advantages of data-first low-code development
- Standardisation and interoperability out of the box: Using standardised data models, applications built on platforms based on openEHR and FHIR ensure consistent data structuring and seamless interoperability across different systems. This approach enables the delivery of an active Shared Care Record, as it is set up on a national level in Slovenia, with 98% of the shared national health data stored on a platform.
- Acceleration of development: Traditional low-code platforms claim to improve development speed by 7-10 times. Combined with low-code tools, the data-first approach can double or even triple these improvements. Starting from a robust data model, it eliminates the guesswork and trial-and-error typically involved in designing healthcare applications. EHA Clinics from Nigeria report that building simple use cases can take 1 to 6 hours, while the most complex ones can take 3 to 4 days.
- Built-in domain knowledge: Healthcare applications require specific medical knowledge. The data-first approach incorporates this knowledge within the models, enabling developers to drag and drop actual data points onto their application canvas. This ensures that the application logic is inherently aligned with clinical requirements. There are almost 1000 archetypes on openEHR CKM available to download for free.
- Ease of use: While traditional coding requires extensive expertise and low-code platforms necessitate understanding the business logic, the data-first, low-code platforms democratise development. They empower business analysts and UX designers, working hand in hand with healthcare professionals, to not only participate in but actively run the application development process.
- Lower maintenance and development costs: Adopting a data-first approach with standardised data models combined with low-code tools and contextual launch of applications within the existing systems massively decreases maintenance and new use-cases development and rollout costs as you develop once and maintain one instead of many. This is well described in the paper about London’s Universal Care Plan, which is built upon these technology principles.
At Better, we exemplify our commitment to innovation and quality in healthcare IT with our Better EHR Studio. It showcases our data-first and low-code approach in a single tool. By leveraging openEHR and the Better digital health platform, we are accelerating the development process and ensuring that our solutions meet the highest standards of interoperability and compliance.
