Artificial Intelligence Drives a New Medication Management Philosophy
By Erick Von Schweber
Erick Von Schweber is CEO of Surveyor Health of Foster City, CA.
There are times when advances in technology enable a radical re-envisioning of both what we do and how we do it. Medication management is at that stage, thanks to ongoing developments in artificial intelligence.
Metaphorically, will healthcare use these newfangled mechanical horses to pull its current wagon, or will it open up to radically new possibilities enabled by new technology? Several areas of AI inspire our imaginations. Let’s consider the philosophical inversions made possible.
Clinical pharmacists performing medication management interventions today spend most of their time poring over patient records, mentally integrating what they read. Then, with little time to ponder the patient’s situation, they go on to writing notes and elaborate documentation that few providers will read in entirety, if at all. Yet probabilistic AI reasoning engines coupled with semantic interoperability integrate multifaceted data without glossing over nuances, driving graphical user interfaces providing information visualizations that clinicians understand in seconds – the mental model is on-screen. The clinician can now understand the entire problem space and visually design a solution.
In this scenario it’s not a matter of man or machine, it’s the collaboration of man and machine, each doing what they do best. Some processes will automate the routine, such as production of documentation, freeing the clinician to spend time doing what only the trained, expert human mind can. Like a financial analyst, they can use that time running what-if simulations to inform their options.
This cooperative interplay between clinician and AI opens up a potential inversion of the customary workflow. Lacking AI, medication optimization today means the clinician attends to each medication in isolation, doing their best to address any issues specific to that therapy and its relational effects with other individual therapies, one at a time (such as duplications and interactions). This traditional workflow leads to Whac-A-Mole, where a considered solution to one issue creates more issues, frequently requiring back-tracking or outright starting over. By visually modeling the entire problem space and assisting the clinician in seeing how to address it fully, AI enables a more productive workflow.
For people outside the AI research community, it’s easy to believe that ML (machine learning) is AI, but the field is far broader. Where ML is about identifying patterns in existing data sets, other areas of AI, such as AI planners, Bayesian probabilistic reasoners, and combinatorial optimization engines, imagine numerous possible scenarios – therapeutic courses of action – then figure out which are viable, which present conflicts, and which make superior tradeoffs for both the patient and the healthcare system. Human cognition inextricably involves both learning and imagination, and in AI circles, imagination, creativity, and metaphor are the vanguard. Indeed, the next steps toward creating an AGI (artificial general intelligence) that operates at near human cognitive levels are focused more on imagination.
We urge those in medication management to free themselves from the bounds that prior generations of technology have restricted them to. It’s time to imagine the future of medication management.