By Adair Chesley, PhD, CCS, CPC, CPMA, CPCO
“If the hare makes too many missteps and has to keep adjusting, the tortoise wins. If the tortoise spends too much time planning each step, the hare wins.”
—Paul Kalanithi, When Breath Becomes Air (2016)
The beauty of a gardener dedicated to their craft originates from their experiences with the colorful interplay of nurturing, designing, and understanding the necessary components to cultivate a healthy, ecologically diverse environment that provides for its users (soil, plants, insects, and human joy in the spirit of subsistence or aesthetic pursuits). To discover that reward between toil and blossoms, a gardener must “know thy soil” and the governing laws of biodiversity.
Come to think of it, it is not entirely too dissimilar from the role of a health information management (HIM) leader. The blended, requisite knowledge of privacy and security, data exchanges, the full revenue cycle, medical coding, and key performance indicators for C-suite executives readily compares itself to the art of gardening.
Most gardens—or in this case, the health system—are in different stages of growth and dormancy, which is why it requires a dependable commitment and upkeeping to maintain a robust and sustainable garden. And yet there is another layer: An excellent gardener is one who constantly learns, adapts, and applies their gained knowledge in both the experimental and experiential.
If you reframe your HIM mindset akin to a gardener, then perhaps the growing interest and recognition of artificial intelligence (AI), machine learning (ML), and robotic process automation (RPA) will serve as areas of exploration rather than a predetermined, irrevocable notion of decimation to your HIM endeavors.
Traditional Applications of AI
To start exploring in the potential value of AI, which circumscribes the fields of ML with its subsequent subfields such as deep learning—by way of broad strokes—an HIM expert can strive to learn and understand the foundational vernacular of these fields and its para-fields (see Figure 1).
Figure 1: Overview of AI and its Major Fields of Inquiry
By understanding the parameters and traditional applications that defines each type of AI and its subsequent specialties, one can develop a working framework in better understanding what each type of AI is, along with what it is not.
Presently, there is a strong shift away from historically utilized terminology toward a more encompassing, dynamic frame of reference as AI expands across multiple industries and works with increasingly complex algorithms and data sets with manifold outcomes: artificial narrow intelligence (ANI), artificial general intelligence (AGI), and artificial superintelligence (ASI). There are four major types of AI: reactive, limited memory, theory of mind, and self-aware. Generally speaking, most of today’s AI applications are in the ANI stage. Additionally, drilling down into concepts like “additive noise,” “optimization,” and “name binding” will validate one’s foundations.
From that vantage point, HIM experts position themselves to be more readily able to identify current limitations of AI, while recognizing where these growing knowledge domains might steadily advance. By cultivating this bandwidth of general frameworks on the types of AI, HIM experts can better ascertain the “how” and “where” their healthcare data sets can maximize its potential in scaling AI initiatives to strengthen their organization’s strategic pushes into the next stages in achieving their digital and computational transformations.
But how does one know which direction to go or where to concentrate their organization’s finite resources in using AI within the HIM sphere?
AI Through the Lens of HIM
One step forward is by returning to the roots of your health systems’ local communities’ philosophies. With a solid understanding in the fundamental mission for its future in the next ten years, you are equipped to realign realities with tangible possibilities in deciding how to overcome its present pain points in using AI.
To adopt this method of calibrating and aligning, start with designing a HIM workflow that is clear in articulating, adopting, and aspiring to those future goals embedded within your organization’s philosophies. The guiding principles that you define at each step will engineer the HIM environment in properly identifying which health data sets can truly benefit with AI initiatives.
For example, if your organization is more driven toward achieving economic prosperity, then an HIM expert should seriously consider incorporating back-end revenue cycle data during the training and validating phases of any algorithm(s).
If the hallmarks of boutique medicine emerge after reflecting on your organization’s future state goals, then time taken to examine how your patients initially and consistently interact with your clinical and administrative ecosystem will underpin the selection of which data sets you identify for utilizing an appropriate machine learning model to anticipate those patients’ (or alternatively, consumers’) demands and desires from chatbots to healthcare literacy.
From there, an HIM expert can meaningfully contribute to designing a seamless flow on a care continuum among providers, administrators, payers, and patients and their respective caretakers. On the other hand, if you discover that there is larger, fundamental shift occurring to prime your organization for profound changes to usher in a genuine digital transformation, then any AI applications and any automation processes should invest in multiple algorithms along with creating the cultural buy-in necessary to the restructuring of multiple workflows and user designs.
Within a digital transformation, HIM experts will need to join the conversations in scaling AI or any big technologies in multiple directions, as those tools hinge on understanding how the health information data is generated and for what purposes. That knowledge helps define and deploy quality data sets that are more applicable to the goals of their organization’s digital transformation initiatives.
The crossroads of the healthcare industry, information exchanges, and consumers’ expectations relies heavily on data that meet multiple demands, and it serves as a ripe opportunity for HIM experts to increase their relevancy. Some may refer to this activity as upskilling, but this term is problematic.
Admittedly, very few will dispute that learning new skills is useless or unlikely to prove advantageous to one’s career journey. Even still, the term “upskilling” carries a connotation that there is a need to replace prior skills as they are rendered defunct, antiquated, and utterly obsolete.
Therein lies the contention. Is it not theoretically true that prior skills can be utilized to appropriately inform the next steps on an AI roadmap for an organization as well as explain the rationale in advocating that not all problems necessitate high tech or AI to solve it (especially considering the large technology debt most organizations currently carry)? (A fuller explanation of what technology debt means within the larger healthcare landscape can be found on p. 116 of the work by Marx, Edward W. and Paddy Padmanabhan listed under References.)
Rather than thinking of understanding of AI merely as upskilling, it would prove more productive if HIM experts view the growing AI field as a dimension to incorporate into their HIM sphere.
In his new book, AI in Health, Tom Lawry writes that “… AI will benefit from the continuous learning nature of it. Just as humans do, health and medical AI systems will learn and adjust from past experiences based on patient and doctor responses and the outcomes seen from these systems over time.” All to say that a symbiotic context will prove immeasurable in expanding the HIM mindset, while watering the industry’s garden.
A dynamic HIM expert who can easily operate as liaison between EHR systems, the flow of data in conjunction with historical and future apparatuses to ascertain any perceived value in adding AI to help decide whether to “augment or automate” specific tasks (as the colloquial saying goes), or tap into predictive analytics will prove invaluable.
Who better knows precisely what is needed in a solution to solve your imminent data problems than you and your HIM teams, who work with these data sets each day? Most HIM experts are thoroughly enmeshed within an organization’s use of technologies and its workflows that may or may not be flawed in reaching the ideal state of a frictionless HIM network. By being a part of the cultural mind shift in the HIM industry in exploring AI with an open mind, you are positioning yourself to clear the biggest hurdles of most digital transformations: “as with most transformative technologies, the big challenge is not the technology; it is managing culture and changes to the workflow and processes.”2(Again, see the work by Marx, Edward W. and Paddy Padmanabhan for a fuller explanation of what technology debt means within the larger healthcare landscape.)
How AI Will Shape HIM
The HIM profession is bounded by regulations and jurisdictions, and yet HIM’s future is poised to take on greater responsibility and visibility within healthcare.
If you have not already, you will start to see more intensive demands for data integrity skills along with entirely new roles and departments existing as another node within the HIM network that are niche and hyper-focused on data science and technologies.
Within those newly formed HIM specialties, leaders can and will enhance the adoption and deployment of APIs as well as meet the new regulatory mandates of price transparency and interoperability.
Today, you can decide whether you will softly or aggressively adopt these new infrastructures to introduce the power of human ingenuity and harness cross-collaboration in maximizing the potential benefits of AI, all the while strategically minimizing risk to redesign your workflow(s) in yielding newer efficiencies. Somewhere on the spectrum of “task management” to “process change,” Paul Daugherty and H. James Wilson articulate in their seminal work, Human + Machine that (emphasis is the author’s):
“When trying to reimagine a process, it’s natural for people to become stuck on the old way of doing things, making it difficult for them to envision things that might be. To avoid that, they should always keep in mind the difference between traditional businesses processes versus the new, AI approach. Our research shows that outcomes are no longer linear but exponential. Change is no longer episodic and human-led; it’s self-adaptive, based on real-time input from humans as well as machines. Roles are not just limited to human-only and machine-only positions; they must also include collaborative work in the missing middle” (150, 155).
According to the Technical University of Munich’s Institute for Ethics in Artificial Intelligence, by reframing the scope and place of AI within the HIM sphere, one will discover the capacity for understanding the boundaries and crossovers between the human-centric and AI-centric models as a network of agents to target and capitalize on salient opportunities as “hybrid intelligence”
I would be remiss if I did not bring to the forefront of your mind that there comes responsibility in understanding and deploying AI. Leaders need a solid foundation on this responsibility, along with an active practice in formal ethics to define and handle AI’s potential for twists and turns that can manifest into consequential bias.
A note of caution: Not all “bias” is inherently a negative slur with regards to data integrity. Like the word “upskilling,” it is a problematic term that is culturally loaded at the time of this article. There are multiple applications and rationales for using specific types of bias within algorithms to help target predictions more effectively, especially common within supply chains meeting finicky customers’ preferences. While there is risk for bias to play out in the real world negatively, you can minimize that by actively and constantly validating your algorithms and building appropriate guardrails.
By outlining what type of formal ethical framework (e.g., Steve Torrance’s “Machine Ethics,” Robert Nozick’s “The Examined Life,” etc.) that an organization will adopt prior to deploying their algorithm(s) before it impacts their organization’s decisions will shine a light in providing accountability and explanations for the conclusions that leaders choose to act upon from their use of AI.
A legitimate recognition of the interconnections among AI, formal ethics and philosophy, and data proves the power of the “missing middle” that can and should be a part of the drumbeat in advancing healthcare.
With street chops in healthcare, Marx and Padmanabhan remind us that “healthcare remains more siloed. Combined with an ingrained aversion to any new technology with the potential to cause harm to patients, the implications for AI adoption in healthcare are clear: It will be slower than in other sectors” (p 132).
Only time will reveal whether their claim maintains veracity, as 2020 deftly demonstrated just how quickly humans and healthcare systems can adapt and scale to meet new demands when regulatory environments support those scaling efforts through a mutual, concerted shared aim.
On a more individualized organizational level, Marx and Padmanabhan are correct, especially in proportional relationship to an organization’s “technology debt” in surveying the reality of just how rooted those silos are and will likely remain.
Nevertheless, an HIM expert with some grasp of AI’s vernacular can contribute to those shaping conversations to break down silos by properly and efficiently identifying the needs and desires in using AI to address their most pressing problems (and also advocating whether AI is even an appropriate tool).
There is no one, true algorithm that works on all healthcare data sets, or only one definitive clear path to alleviate its real-world problems. In returning to the opening quote of this article, you will have to decide which is wiser in being the hare or the tortoise in pursuing AI.
When it comes to the deeply sensitive subjects of our lives—our health, finances, and personal convictions—we owe it to ourselves to acclimate to the next wave of technologies to understand its implications and harness its potential to place humanity in an advantageous position to benefit from its subliminal awe. While in the midst of our tension between wonder and hesitation, we need the wisdom to recognize when to act on AI in respecting our collective boundaries that defines and guides our moral compass of humanity.
Daugherty, Paul R. and H. James Wilson. Human + Machine: Reimagining Work in the Age of AI. Harvard Business Review P, 2018.
Lawry, Tom. AI in Health: A Leader’s Guide to Winning in the New Age of Intelligent Health Systems. Taylor & Francis, 2020.
Marx, Edward W. and Paddy Padmanabhan. Healthcare Digital Transformation: How Consumerism, Technology, and Pandemic are Accelerating the Future. CRC P, 2021.
Technical University of Munich’s Institute for Ethics in Artificial Intelligence (TUM IEAI). Reflections on AI Q&A with Andrea Martin. January 2021.
Adair Chesley (firstname.lastname@example.org) is an AI/ML advisor and data strategist to various health systems, tech companies, and digital health startups seeking to design and scale their AI roadmap with an ethical framework.