“The problem with large centralized national repositories for health data is, that connecting literally thousands of providers, they together create thousands of points of access to this centralized database. This can create security and privacy issues. In Australia, with uneven network coverage, not all providers are in the network, creating a system that doesn’t provide much value to patients or providers. It may work for some people, but not others and in essence becomes an unjust system,” said Marie Johnson about My Health Record.

Marie Johnson is the CEO of the Centre for Digital Business. She is a writer, commentator, and international speaker on artificial intelligence, digital transformation, cyber, technology, ethics, and the human experience.

Marie’s expertise is the human interface in complex servicing systems. She led the collaborative development of Microsoft’s global e-government strategy, led business authentication, business digital identity, and professional digital credential initiatives, was Chief Technology Architect (CTA) of the Australian Government Health, and Human Services Access Card program, among other things.

With her experience as a patient and expert in complex systems, she is skeptical about My Health Record - the centralized repository of health data in Australia. Mostly because it seems too complicated to work in practice. “There is a paradigm in Australia and elsewhere across the world that digital health equals some sort of centralized health record held mostly by governments. That whole paradigm falls down on a whole range of different levels. On the spot in the ambulance, where paramedics don’t have time to call up a health record. Even if they did, it probably doesn't have everybody's updated information. And even if a did, even if it had the latest information from a doctor's visit a month ago, it doesn't have what's happening on the spot at that point in time,” she mentions.

That’s why she believes that going forward personal health repositories held by individuals are going to prevail.

“Health data is not just about what medications are received or what conditions I have. It also includes indicators of wellness and lifelong good health. So I think that's a paradigm shift. In Australia, there have literally been trillions of dollars spent on what's called My Health Record - the paradigm of having a central control, which actually doesn't deliver value to the patient. On the other side, companies are building solutions enabling patients to have their health data with them on their smartphones. That is a far more empowering and democratizing use of their own health information than a centralized health record. I think that's going to change going forward. I don't know whether governments will have an appetite to change the paradigm. But many others are coming into this space because there was a realization that the centralized control does two things: it doesn't serve the patient and it turns the health practitioner into a data entry operator, which is an incredibly wasteful use of a health practitioner’s time,” Marie Johnson says.

The role of co-design in healthcare

Marie Johnson is the co-creator of Nadia, the first AI digital human for service delivery, and the creator of the AI digital human cardiac coach. She sees co-design as the key to the advancement of AI and digital solutions for healthcare. As she explains, co-design works with the common patterns that already exist.

Regardless of the health condition, people have common patterns of conversation. Co-design starts from the perspective of a person and brings that person into the situation of being provided a service. It's not theoretical but grounded in real-life practice. “ For example, cardiac health information is generic across the world. Mostly we all have one heart and there's a specific range of medications and treatments and surgical interventions. But without health literacy, people don’t understand concepts. An example is an information about having a certain volume of fluid. If you don't understand the concept of volume, how can you understand recommendations? The process of co-design builds up the data set. How do people understand volume? They understand it through analogy. A cup or two cups, for example. So co-design is not about building up a data set of breadth as you might have with Alexa or Siri who can answer questions on any topic. Co-design builds up the depth in a particular domain and adds to that that some concepts don't exist in other languages,” she elaborates. Another example is “disability”. In the Australian indigenous community, the concept of disability does not have a direct translation. This illustrates that just working on a purely technical, bureaucratic corpus, a good solution can’t be created, but co-design is about these bounded domains. 

Listen to the full interview:

Some questions addressed:

- How would you describe the role of AI in your life? 

- You worked in the business for several years: led the collaborative development of Microsoft’s global e-government strategy, led business authentication, business digital identity, and professional digital credential initiatives, was Chief Technology Architect (CTA) of the Australian Government Health and Human Services Access Card program. What led you to focus more on healthcare in the end? 

- What has the work with the government taught you about how change on national levels is achieved? 

- How do you see Australian healthcare today?

- You are working on several solutions to increase health literacy. For starters: why is health literacy a problem?

- Can you briefly describe the two solutions you co-created - Nadia and digital human cardio coach? 

- What kind of datasets were used to create them? (What does that tell us about a broader understanding or reality of having digital coaches for various conditions? Every disease category has its own complexity and it's difficult to imagine the creation of digital coaches for all conditions) 

- You've been in the field of AI for 15 years. How do you see the field evolved in real life? (Many of the same ideas that have been popular 20 years ago regarding what AI will be capable of, are still with us today as if they're new which makes it seem that we're not really moving very fast with AI development.)