Why is healthcare
software so hard to create? healing

I think existing healthcare software isn’t that great. The user experience is terrible, the functionality is disappointing, and overall, it misses the potential that healthcare software has.

I’m not the only person to think this - many people have tried to build better software for healthcare. However, even with immense funding, political influence, institutional backing, and experience in healthcare - smart people have tried and failed to change Health IT. Why?

I explore this question in my thesis, and argue three reasons why healthcare software is so hard to build:

1: Healthcare is complex, and varied

Modern care is complex. What makes it worse is every institution, and region does it slightly differently.

2: Healthcare is risky

Your software fails - tragic things happen.

3: Healthcare software is hard to distribute

Once you’ve built some fantastic software - how do you actually get it into real world use?

Proposing:

Barnett

An open, extensible, distributed electronic health record system

Barnett is a fictional piece of software aimed to design around the issues I discuss in my thesis. It is a general purpose EHR system. It’s not institution-oriented, but problem-oriented - the problem of record management. This means it has a very limited scope of functionality:

Functionality

Store health records

Represent healthcare information in a consistent, but extensible format.

Share health records

Patients are surrounded by a team of healthcare professionals - they should all work on the same page.

Interact with health records.

Patients, software, and medical professionals all need to interact with health records - through both code & UI.

And nothing else

In my thesis, I argue that existing healthcare software is too complex. For example, patient management systems in primary care will easily handle all of a medical practice’s needs - appointment management, patient records, invoicing, claims, analysis, etc. The result is the software does everything, but nothing that great.

Barrett aims to do one thing well - manage healthcare records. Regardless of the context - primary care, hospitals, specialists, anything. It’s designed to be used in tandem with other systems which can specialise in their area of responsibility. This aims to reduce distribution costs - if an institution wants to try a new tool, they don’t have to replace their entire software system.

Design Principles

Adaptability

Every part of Barnett is modular, so they can be replaced and improved - allowing it to adapt to any workflow, context, or individual.

These improvements are not provided by a single company - anybody can create an improvement, with some simple coding. For example, say you dislike part of the interface - write an extension to replace it. And then, share your extension, so that everybody can benefit from your work!


Mix and match the community's work to configure Barnett to match your needs. Together, Barnett will grow and evolve.

Faster Distribution Cycles

Upgrading software is a huge challenge for a healthcare institution, due to the sociotechnical change. Institutions have to balance the value created by a new system, against the cost & risk of changing.

Instead of trying to create value with excessive features, Barnett focuses on reducing the cost & risk of change. Changes are designed to be smaller, faster, and less risky.

This aims to create better software - though a faster distribution cycle of designing, testing, and improving.

How does it work?

Format

Health records are represented as a collection of modules. They are small, consistent, scoped pieces of health information. They are essentially schema's - they provide structure to health information

What is unique about modules is that anybody can define new modules. All you have to do is write a new format, and add it to the online registry for others to use.

Write a module for diabetes management

Patient’s needs vary, and the information they need tracked varies. Why should a young mother and an elderly cancer patient have the same record? Their doctors, families, and themselves - care about very different information.

Defining what information should be part of a health record now, or in the future is a tricky task. This format would allow users, institutions, and specialities, to easily redefine and adapt records as their needs change. Or even better - find existing formats which fit their needs.

Network

Barnett is a distributed application - there is no central service or database which could fail. The network emulates a single record which is accessed by institutions, but in reality, the storage of the record is distributed across all institutions who interact with that patient. Barnett’s network serves three main purposes:

Clusters

Barnett is self-hosted. Institutions run their own servers, which are organised into clusters. This provides incredible robustness and accessibility.

All of an institution’s records are stored on its cluster, independent of any other institution.

Distributed Network

Clusters are then connected to create a distributed network. Clusters will share information if they both have the same patient, and same module.

For example, our patient Alice visits three institutions. They all need to know her medication list - so when her list is updated at one of them, they all sync the changes.

She has a chronic condition, which is managed by both her GP and her specialist, so they share her dermatology module. However, her detailed visitation notes are private to just her GP.

Why use a Distributed Network?

Distributed systems make a lot of sense for healthcare. The primary reason is robustness - they are fantastic at mitigating failure.

Secondly, they are scalable. Healthcare is mostly a regional issue - you will most likely need to access your data close to your existing healthcare providers. Distributed networks have been proven to have the potential to scale to a worldwide health network.

User Interface

Fantastic software can often be let down by a bad User Interface. However, what counts as ‘Bad’ for one user can be fantastic for another, and vice versa.

A user’s need of their UI can change dramatically between different workflows, context, or environments. Put simply, different users need different User Interfaces.

Solution?
Extensible User Interfaces

refresh

3

Eliot Slevin, 22

Visitations

Details

Medications

Allergies

Lab Results

Probable Strep Throat

is simply dummy text of the printing and typesetting industry. Lorem Ipsum has been the industry's standard dummy text ever since the 1500s, when an unknown printer took a galley of type and scrambled it to make a type specimen book. It has survived not only five centuries, but also the leap into electronic typesetting, remaining essentially unchanged.

Routine Checkup

is simply dummy text of the printing and typesetting industry. Lorem Ipsum has been the industry's standard dummy text ever since the 1500s, when an unknown printer took a galley of type and scrambled it to make a type specimen book. It has survived not only five centuries, but also the leap into electronic typesetting, remaining essentially unchanged.

Ankle Injury

is simply dummy text of the printing and typesetting industry. Lorem Ipsum has been the industry's standard dummy text ever since the 1500s, when an unknown printer took a galley of type and scrambled it to make a type specimen book. It has survived not only five centuries, but also the leap into electronic typesetting, remaining essentially unchanged.

Viral Illness

is simply dummy text of the printing and typesetting industry. Lorem Ipsum has been the industry's standard dummy text ever since the 1500s, when an unknown printer took a galley of type and scrambled it to make a type specimen book. It has survived not only five centuries, but also the leap into electronic typesetting, remaining essentially unchanged.

Sleeping Issues

is simply dummy text of the printing and typesetting industry. Lorem Ipsum has been the industry's standard dummy text ever since the 1500s, when an unknown printer took a galley of type and scrambled it to make a type specimen book. It has survived not only five centuries, but also the leap into electronic typesetting, remaining essentially unchanged.

Viral Illness

is simply dummy text of the printing and typesetting industry. Lorem Ipsum has been the industry's standard dummy text ever since the 1500s, when an unknown printer took a galley of type and scrambled it to make a type specimen book. It has survived not only five centuries, but also the leap into electronic typesetting, remaining essentially unchanged.

Every part of Barnett’s UI can be re-written, and replaced with extensions. Don’t like the immunisation tab? Remake it! Want a sidebar? Write it! Want more keyboard shortcuts? Add some! Similar to modules, extensions are shared online via registries. Share your hard work with other users - and mix and match extensions to fit your needs, and how you want to work.

Similar ideas have worked for Atom Text Editor, Sublime Text, and Sketch.

Software Interface

In the same way users (patients, medical staff, and family) interact with medical records, software needs to interact with records. In fact, Barnett treats software interaction just as important as human interaction.

OAuth Logins

Access records with OAuth style login flows - perfect for consumer facing products. Storing healthcare info yourself is risky, and complex - this way you just act as a client.


Example use: A diabetes tracker, where the data can be viewed and discussed by both the patient, and the doctor.

Infant mortality rate, United States 1915 - 1997

Statistical Analysis

Data science has more opportunity than ever to improve healthcare, through advancements in Machine Learning and data collection.

Barnett provides automatic anonymization and data collection tools. This makes it easier for government, institutions, and communities to open up their data for public research.

FAQ
Wait, who are you? I’m Eliot Slevin, designer and software developer. This project is my thesis, which I completed for my Masters in Design Innovation at Victoria University of Wellington, New Zealand.
Where can I read your thesis? Right here! My thesis is part of this website, hit the hamburger menu in the top right to get started.
Barnett looks good, where can I download it?Nowhere currently, this is just a design proposal
What would it take to actually make this? Apart from funding and a very talented team, what it really requires is a group of regional institutions to commit to using and supporting Barnett.
This would never work because X I’m seriously interested in any and all feedback. I’d love to know your thoughts - please email me at eliot.slevin@gmail.com. But, make sure you've read my own critical review first.
Why is it called Barnett It’s named after Dr Octo Barnett, who was involved in the creation of MUMPS, one of the earliest medical record storage systems.