Why we need global Health Data Networks?

9 min readJun 2, 2018

In popular culture, the future of healthcare is often portrayed with physicians looking at their patient’s records on holographic screens, just like this documentary from 2007 :

It’s 2018. We have almost figured out how to make holographic display. Yet, we’re not even remotely close to figure out how to get the health records on the screen in the first place. A world where everyone can benefit from the immediate availability of health data must have some kind of network where health information can transit safely and quickly. Such technology is not universally available yet.

The wealth of health data

Health data can be virtually anything. It comes in different shapes and dimensions : clinical metrics, socio-economic determinants, behavioral information, financial transactions or even Instagram Photos. It’s traditionally collected at the point of care, also increasingly as “secondary data” such as research or surveys, and, more often than not as the by-product of healthcare providers’ operations. This includes billing as well as clinical activities.

The emergence of digital health has expanded the breadth of health data by creating opportunities for patient-generated health data (PGHD).

Along with variety, there’s an exponential inflation of volume, thanks to increasing EHR systems’ penetration and use of digital clinical hardware. Let alone the proliferation of smart devices, which generate a swath of information.

The typical Phase III clinical trial protocol now collects more than one million data points, double the level observed 10 years ago. Our research at Human API consistently shows that EMR data (Electronic Medical Records) has been increasing in density at a faster pace in the last 5 years.

Occurrences of 6 health data types in a random sample of patients in the US, from 2008 to 2017 — Courtesy of Human API.

Why does data liquidity matter ?

So, we have health data collected and digitized more then it ever was, yet not effectively used at scale. For example, only 19% of hospitals in the US said they often used data from outside providers, even when it’s available electronically. There’s a chronic data liquidity problem in healthcare.

You have probably experienced poor data liquidity when you had to fill the same form, again, at the waiting room in your new doctor’s office…

Some patients experience poor data liquidity as medical errors, which can be very serious and are preventable if the care provider had the right information at the right time...

Care providers experience poor data liquidity as unpaid bills charged to the wrong insurance, or denied for transcription errors…

Insurance Payers experience poor data liquidity when they can’t predict future utilization patterns and estimate changes in claims costs.

Perfect data liquidity is when the information is collected or input only once, then made available immediately downstream for other systems or users. Having to input the data again is not only slow but can be ineffective or even harmful, as every retranscription introduces a risk of errors and omissions.

“We are in the midst of a Third healthcare revolution driven by citizens, knowledge, and the Internet” — Pr Muir Gray

Professor Muir Gray identifies 5 major challenges facing healthcare services worldwide in the 21st century. All of them are due to inefficiency and ineffectiveness.

Inefficiency comes as waste, inequity, unwarranted variations in outcomes and generally low value/cost ratio.

Ineffectiveness, comes as consistent failure to prevent preventable disease and, sometimes, patient harm, even with high quality care.

Those are very hard problems to solve. Fortunately, we are at a major turning point where we can apply new science and technologies.

The scientific understanding of diseases and care pathways is experiencing an exponential growth, notably thanks to translational medicine and other interdisciplinary approaches which leverage Big Data to consistently derive medical knowledge.

“Healthcare is knowledge-rich; yet healthcare knowledge is largely under-utilized at the point-of-care and point-of-need.”

The lack of information when it’s needed may cause serious health issues. It’s also inefficient, more so that healthcare economy is shifting towards a Value-Based paradigm, where multiple entities share financial responsibility for the same patient’s long-term outcomes.

In such context, the continuity of care is the continuity of data, and thorough communication between care providers is an absolute necessity to avoid duplicated services and mistakes as the patient moves across the care continuum.

The anatomy of a global Health Data Network

This is not an attempt to outline my vision of the future, or anyone else’s for that matter. Because, as William Gibson puts it so brilliantly, “the future is already here, it’s just not evenly distributed”.

As a matter of fact, regional and national health data networks are currently being implemented all over the world : NHIN and, EHDN or the Health Information Exchange initiative are just a few examples worth mentioning. Let alone data banks and privately operated networks, such as the one we’re building at Human API. As they merge and connect to each other, they’ll become hubs within a global health data network with small world properties. One can look at the successes and shortcomings of what is being done today to extrapolate how the future should look like.

The global health data network will be a decentralized constellation of interconnected heterogeneous networks, with 4 things that matter :

Comprehensive

All kinds of health information can transit on the network.

As I mentioned above, health data comes in different shapes which highly complement each other. We learned it first hand at Human API, as the idea of “Complete Picture” underlies every customer’s Use Case. Everyone has their own definition of what the Complete Picture should be, yet they all agree that it’s multi-faceted and relies on more than one type of health data.

Failure to support the widest breadth of health data will drive a fraction of users away, leading to inefficiencies and inequity at the boundaries of the network.

Patient-centric

Patients control their health data.

By force of law, the patients own their health data, but ownership is not control. Individuals should be empowered to exercise effective control on their health records. Multiple studies have actually documented their interest in doing so. A true HDN will not only include the patient as a first class citizen, but also put them in the driver’s seat, as the main protagonist of their health.

Also, it’s now widely acknowledged that a more patient-centered, collaborative approach is needed to foster patient engagement. Research has clearly established improved patient satisfaction, behaviors, and health status in response to patient-centered practices. One such practice is increasing patients’ access to timely and accurate information as a method for improving the quality of healthcare delivery.

Rich space of modalities

There are always multiple ways to transact.

Let’s draw a parallel with transportation networks. There are always multiple ways to travel from a point A to a point B. It all depends on how fast, cheap, comfortable, green or spiritually fulfilling you want the journey to be. A single mode of transportation doesn’t solve everyone’s problem, but a combination of modes will.

Similarly, a health data transaction involves a request, patient consent, the exchange, storage and retention of data, all of which can be done in different ways (or modes). For example, the patient can consent to share their data with a third party using an electronic signature, biometric identification, oAuth and so on.

Economically viable

The network organically supports itself.

Many health data networks struggle to get traction because they lack a proper business model. In the US, an evaluation of the HIE program outlines challenges to secure the necessary funding to continue operating in the short term — and failure to demonstrate long-term financial sustainability to stakeholders. In France, a similar study documents poor adoption of “Meaningful Use” programs by providers due to the lack of financial incentives.

Challenges ahead

A global health data network is an inevitable outcome. The question is not whether we’ll get there, but when we’ll get there, which depends on this generation’s will to fix trust issues, give patients access to their data and achieve real interoperability.

Trust

Indicators of patients confidence in electronic health data exchange did not substantially improve. A 2017 survey by HealthIT.gov suggests that 66% of Americans have concerns regarding unauthorized access to their private health information during exchange between providers. At least 10% withhold health information from providers because of privacy concerns.

Trust is beyond security and compliance with regulations. A health data network won’t see wide adoption unless all the parties deem it trustworthy. When you use your credit card, you trust the Network, so does the merchant you’re paying, so does your banker, the merchant’s banker, the credit card terminal manufacturer and so on.

“Your network is the people who want to help you, and you want to help them, and that’s really powerful”
— Reid Hoffman

Patient access

In March 2017, 67% of all healthcare providers in the US reported using an EHR, a 1% increase over September 2016.

As of 2018, about 93% of hospitals in the US provide some form of online access to patients, but the figure is much lower for private practices which are the bulk of the provider landscape, especially in rural areas.

The figures are significantly lower in developing countries, such as my home country, Tunisia, where EHR penetration is below 50%.

A shared meaning of things

While 43% of hospitals in the US reported that data from outside providers was available electronically when necessary in 2015, more than one-third reported that they rarely or never used it. In fact, just 19% of hospitals said they often used data from outside providers.

The most common barrier these hospitals reported to using outside information was that their clinicians could not see it embedded into their own software. That’s because interoperability is NOT liquidity. If you consider a network of fax machines, it has high interoperability but offers poor liquidity, because someone has to input the data, again.

Liquidity is the highest level of interoperability, when data encoding, syntax and semantics are well known and unambiguous to everyone. Common standards tackle the syntactic aspect of health data and care less about the semantics. While standards like HL-7 and DICOM prescribe how data should be packaged, there’s no consistent shared meaning of things. The addition of layers of semantic models (such as RIM and FHIR) and coding systems (such as ICD) doesn’t solve the issue completely. That’s why medical coding is an actual job.

What now?

Health data is being collected and digitized more than it ever was. As it becomes easily and safely available from anywhere, the entire healthcare landscape will change forever and physicians will browse medical records on holographic screens. That future, however, is not some place we are going, but one we are creating. It takes a whole generation of determined and smart people to make that possible, but it’s only a matter of time.

I work at Human API, which strives to empower those people. We are building and operating the first consumer-centric network for health data, now the largest of its kind. We have a long way to go, but we are constantly galvanized by the positive impact in our user’s lives when they can access and share health records in real time. We want to bring that faculty to everyone. Give us a try, or join us.