Thomas Jefferson University’s College of Population Health is starting a new graduate degree program in Population Health Intelligence. As an adjunct faculty member in the program, I’m working on preparing one of the foundational courses, called Health Data Acquisition and Management.
Multiple connotations surround the word “Intelligence,” and I think it’s important to realize what this term means and what it’s trying to achieve. The meaning can vary depending on context and intention. Here are three examples of current uses of the word intelligence, and how I believe these connotations will apply to our Population Health Intelligence curriculum.
Most obviously, Population Health Intelligence is a riff on Business Intelligence, and Healthcare Intelligence. In our Pop Health context, Intelligence is very close to Analytics—but it aims to be more than that. It covers the comprehensive data process, starting with Data Collection, Aggregation and Management, continuing to Data Organization, Stratification and Analysis, and winding up with Data Visualization, Interpretation, Dissemination—and Action Planning. Population Health, by definition, is a data-driven activity, and Population Health Intelligence covers the entire data-cycle within Population Health.
How about Intelligence as in Artificial Intelligence? In this context, Intelligence means building computer systems that emulate human decision making. The AI revolution has been underway for quite some time now, beginning with limited-domain decisions in fields such as physical plant (predictive maintenance) and transportation (freight efficiencies, smart traffic lights). But now AI is rapidly evolving to take on more complex human tasks; self-driving cars, and drones delivering goods directly to homes are well-publicized examples.
What does Population Health Intelligence mean in the context of Artificial Intelligence? What decisions in the sphere of Pop Health do we see as suitable, amenable to an AI approach? I would argue for high-complexity, high-value decisions—and in Pop Health, that would include program design. In a Population Health setting—with risk sharing and a complex mix of incentives between payer and provider, in a large population of members, and a large set of providers—AI algorithms would be utilizing large sets of data, and AI learnings, to identify the best ways to improve the care for the largest set of patients possible, and optimize the value. There are so many groups to focus on—based on chronic illness, risk for admission or readmission, socio-economic determinants, and other factors, that it will require Population Health Intelligence to identify and track the best initiatives.
Finally, how about Intelligence as in CIA? In the CIA context, Intelligence is about connecting the dots and figuring out something that’s hidden underneath the surface but important to discover. It refers to pro-active fact seeking, pattern finding, and other investigative endeavors. This version of Intelligence also applies to Population Health. There are lots and lots and lots of dots to connect—and we need to help physicians, care givers, and ACO leaders figure out what trends are relevant and need their focus. All the while keeping data security, validity, and ethics in mind. Pop Health data includes a lot of noise. We can’t (and shouldn’t) fill all the Care Gaps. We can’t (and shouldn’t) chase every Quality Measure. We can’t (and shouldn’t) incentivize or reward every intervention. How do we know what matters most, and put our valuable time and money where it makes the most difference? That’s where Population Health Intelligence as in CIA applies.
I’m excited to be part of the faculty in the Pop Health Intelligence program. It’s about educating and training leaders who will bring Business Intelligence, Artificial Intelligence, and “Spidey Sense” into Population Health.
Jefferson Medical College