I recently read a paper by Jingshan Li and Pascale Carayon published in the Transactions on Healthcare Systems Engineering journal (Health Care 4.0: A vision for smart and connected health care). They had some fascinating things to say about how the evolution of healthcare mimics that of industry. However, I thought their discussion focused too much on technology as a driver of change, rather than simply an enabler. And I think much more should be said about how important the quality, not just availability, of data will be to move us to their vision of Health Care 4.0 as an interconnected system driven by big data.

What is Health Care 4.0?

Before we can talk about the forces driving and enabling Health Care 4.0, we should define what it is. Li and Carayon give an excellent description, which I summarize, below.

Since the beginning of the Industrial Revolution, industry has gone through four major evolutions:

Li and Carayon chart a similar trajectory for healthcare:

Healthcare is in a transition stage between Health Care 3.0 and Health Care 4.0. EHRs are firmly established: the Centers for Disease Control and Prevention reportsthat 89.9% of office-based physicians use an EHR and the Office of The National Coordinator for Health Information Technology pegs hospital use at nearly universal (96%, using 2016 data). But the ability to share data across the health ecosystem of patients, payers, providers, labs, public health departments, the federal government and others is still a challenge. Lack of consistency and the siloed way in which clinical data is collected impede our efforts, as does our arcane nationwide lack of a unique patient identifier.

Li and Carayon do an excellent job of identifying the emerging technologies that are moving us into Health Care 4.0 (e.g., AI techniques, wearables), in parallel with Industry 2.0. (e.g., the Internet of Things, Big Data and AI). However, while technologies like wearables, and increasing computing power enabling analytics and AI, are very impressive, these are all simply enablers, not drivers of the transition. And Li and Carayon miss one critical enabler that is unique to healthcare: a foundation of high quality, interoperable data.

Drivers: Demographic Forces

If advances in technology are the enablers, then what is really driving the evolution to Health Care 4.0?

While short term accelerators, like COVID-19, are putting the proverbial pedal to the metal, the real drivers are deeper demographic trends facing the U.S. and the world. Two of these trends highlight the need for high quality data we can trust: the aging population and a shortage of healthcare professionals.

As the population ages we expect to see an increased load on healthcare, simply because older people require more care. The sickest 5% of the population (including the aging population) already consume 50% of health care spending (Pearl & Madvig 2020). The aging of baby boomers is predicted to affect costs even more as the proportion of the population over 65 increases to almost 20% after 2030 (Schreck 2019).

Ironically, the availability of a larger number of sophisticated diagnostic tests and treatments – while giving us a longer lease on life – is also contributing to the complexity and cost of care. In fact, the use of new technologies and drugs may be the largest single factor increasing health care costs (Schreck 2019).

This increased pressure on healthcare will come up against the shortage of healthcare professionals to deliver it. The Association of American Medical Colleges “projects that the United States will face a shortage of between 54,100 and 139,000 physicians by 2033.” (Boyle 2000). This is partially due to the aging population, with medical professionals retiring in droves. But there are other factors. Similar to Industry 3.0 pushing out jobs because employees were not able to work at the wages provided, Health Care 3.0 has created a strain on medical professionals being asked to do shift work, see more patients in less time, and deal with increasing administrative burdens. Many healthcare workers are looking to get out of clinical practice and into more attractive and less stressful jobs. Looming shortages of medical professionals will create increasing demand for innovative and data-driven solutions to drive workforce efficiencies.

These trends also elevate the need for preventative care, and active engagement by patients in their own care. The federal government has taken a leadership role in giving consumers timely access to their own health information, with promulgation of the Centers for Medicare and Medicaid Services Interoperability and Patient Access Final Rule. This requires use of Fast Healthcare Interoperability Resources (FHIR) Application Programming Interfaces (APIs) to deliver Medicare and Medicaid members access to their records, and we expect the commercial payers to follow suit. But addressing the fundamental issue of data quality – data that can be shared and understood by both humans and advanced technologies like AI, is the critical enabler of Health Care 4.0.

Data Quality is Key to Health Care 4.0

But consider that patients do not have the benefit of years of medical training. They may simply be unable to tell when two medications are actually the same, only coded differently. They may also be overwhelmed by repeated data within their record. They may not know that, because of how coding is done for billing purposes, sometimes their immunizations might appear under their list of procedures. Medical professionals waste a great deal of cognitive load dealing with these issues, but many patients simply do not have the knowledge needed to correct for them.

AI, another key characteristic of Health Care 4.0, also depends on high quality data. Li and Carayon talk extensively about “stratification and classification” as a first element of Health Care 4.0 and focus on identifying cohorts of patients to power “… risk stratification tools for stroke, cancer, and delirium.”

Let’s say we are interested in finding patients with diabetes. In order to train our AI, we will use patients already labeled with diabetes as a condition. With this AI we can then detect patients that have not been properly coded as having diabetes, but probably should. These are the patients we want to reach out to make sure they control their blood sugar.

Now let’s approach the problem without clean and enriched data that is coded correctly. We may have multiple ways that the drug, Metformin, is coded. Metformin should be a strong indicator in our AI, as it is often used to treat diabetes. However, its association will be diffused by all the different ways it was coded. This could result in the AI treating it as a weak indicator.

If, instead, we have data with semantic correction that standardizes all the ways that Metformin can be coded to one. then the AI will have little problem spotting the correlation. The resulting AI will treat Metformin as a strong indicator and be much more successful identifying patients who have diabetes.

As described in the examples above, semantically normalized, enriched and standards-based data is the key to moving us into Health Care 4.0. While advancements in science, wearables, and AI will only move us closer to Health Care 4.0, high quality data will get us all the way there.

Subscribe to our blog to keep up with issues related to clinical data quality.