AI, the Workforce, and the Last Mile of Healthcare Innovation

Q:  Artificial intelligence is reshaping every field of inquiry. Research universities are describing AI as an accelerant for discovery. How should a healthcare and college think about AI when it is not a research institution?

TL:  By being honest about which end of the problem it owns. I have spent a long time inside research universities, and I have enormous respect for what they do. Discovery is genuinely accelerating there, and the convergence they describe — ideas from separate fields meeting and producing something neither could produce alone — is real. That is their work.

It is not ours. Cambridge does not run laboratories that invent clinical AI, and I would not want a prospective student to think otherwise. What we do is the part that comes after: preparing the people who have to operate these tools in buildings where nobody has an informatics department.

Those two jobs are not ranked. They are sequential. A discovery that never reaches a patient has not helped anyone yet. The gap between a validated tool and a tool in routine use is what health services researchers have called the last mile, and it is the most underestimated segment of the whole pipeline. AI is compressing the discovery half of the timeline dramatically. It is not automatically compressing the adoption half, because adoption is not a computational problem. It is a workforce problem.

That is the question I would put back to the research enterprise: you are asking how AI will reshape the workforce. We are the answer to that question. The reshaping happens in our graduates’ jobs.

Q:  You have described the last mile as Cambridge’s territory. What does that actually look like from where you sit?

TL:  It looks like a nurse manager at a community hospital in central Florida being told that the EHR now has a predictive deterioration score, with about forty minutes of vendor training and no one on staff whose job is to evaluate it.

It looks like an HIM director at a facility in Georgia discovering that the coding platform has begun auto-assigning codes, and having to decide what her team’s job now is.

It looks like a one-person IT department suddenly responsible for the security posture of three new AI-connected systems that all touch protected health information.

Our campuses in Florida and Georgia and our clinical site relationships put us inside those settings continuously. That is a genuine informational advantage, and I do not think we have made enough of it. Most published research on AI adoption is conducted at academic medical centers, because that is where the researchers are. But most American healthcare is not delivered at academic medical centers. It is delivered at exactly the kind of facility our students train in and go to work for. We watch AI arrive in those settings every week, in real time, in places that will never appear in a study.

The other thing that follows from our position is speed. A research finding travels through publication, replication, guideline revision, and eventually practice. That takes years. When we see something change in how coding actually works, or how documentation is actually being done, we can put it into a course within a term. We are not competing with universities on depth. We are competing with the clock.

“A discovery that never reaches a patient has not helped anyone yet. AI is an accelerant for the discovery half of that journey. The second half has a workforce — and the workforce is the bottleneck.”
— Dr. TERRY LAPIER, CEO, CAMBRIDGE HEALTH

Q:  Research institutions are rethinking their funding models as federal support becomes less predictable. Cambridge does not depend on that funding in the same way. What does sustainable innovation financing look like for an institution like yours?

TL:  It looks like being useful enough to employers that they fund it, which is a discipline I would recommend to anyone.

We are not going to win competitive federal research grants, and we are not organized to chase them. What we have instead is a direct relationship with the organizations that hire our graduates. When a health system tells us their coding department is being restructured around automation, or that they cannot fill privacy and security roles, that is not market research — it is a funding signal and a curriculum signal arriving at the same time.

Three mechanisms matter most to us. First, employer partnerships — direct relationships with health systems and facilities that hire from us and tell us what they need. Second, tuition reimbursement: a very large share of healthcare employers will pay for their people to advance, and that funding is dramatically underused by the employees eligible for it. Third, clinical site relationships, which cost us effort rather than dollars and return information we could not buy.

There is a version of this that sounds like a limitation. I would argue the opposite. An institution funded by grants can pursue an idea for years before anyone asks whether it worked. We find out within a placement cycle. That is a harsher feedback loop and a more honest one.

 Pell grant and Federal student loans should be added and listed first for H2-7. Workforce is called CareerSource now. 

Q:  What are the great challenges of our time that Cambridge is positioned to address?

TL:  Four, and they are all versions of the same one.

The adoption gap. Advanced tools are reaching well-resourced institutions years before they reach community hospitals, rural facilities, and long-term care. That gap is becoming a care-quality gap. It closes through people, not procurement.

The trust problem. A clinician who cannot evaluate an algorithm’s output has two bad options: defer to it completely or ignore it completely. Both are dangerous. The middle path — informed skepticism — has to be taught, and it has to be taught to the people at the point of care, not only to the people who buy the software.

The security surface. Every AI tool added to a health system network is another system holding protected health information. The expansion of that attack surface is outrunning the supply of people qualified to defend it, especially in smaller organizations. That is why we have built out healthcare cybersecurity and privacy as a distinct vertical rather than treating it as a footnote to general IT.

These are not narrow problems. They require clinicians, technologists, coders, administrators, and security professionals who can talk to one another. A college that trains all of those groups under one roof has an unusual opportunity, and an obligation to use it.

Q:  How can health systems, employers, and community partners engage with Cambridge right now?

TL:  Directly, and earlier than they usually do.

The most valuable thing an employer can give us is not a donation. It is specificity: tell us the role you cannot fill, the skill your new hires are missing, and the change coming to your department in eighteen months. That information reaches our curriculum faster than it reaches almost anywhere else, because we are structurally short and close to the ground. Employers can start at Cambridge for Employers or through Career Services.

For employers with tuition benefits, the second thing is simpler: tell your staff the benefit exists and point them at a program. Our online division is built for people who are already working — the RN to BSN, the Master of Health Informatics, Health Information Management, and Healthcare Administration exist precisely for that person.

And for clinical partners: the sites that host our students are giving us the clearest available picture of how healthcare technology is actually being adopted outside academic medicine. I consider that a form of collaboration, and I would like more of it.

Speed matters here in the same way it matters in research. The tools are arriving now. The people who have to run them are on shift now. The institutions that close that gap fastest will be the ones that decided not to wait for permission.

About Dr. Terry LaPier

Dr. Terry LaPier serves as CEO of CambridgeHealth.edu, which operates campuses in Atlanta, Miami, Orlando (Altamonte Springs), and Delray Beach, along with a growing online division. He holds a Ph.D. from the London School of Economics, Advanced Management Program at the University of Pennsylvania’s Wharton School, and earned his bachelor’s degree from the Maxwell School at Syracuse University. He spent seventeen years on the Wharton faculty. He is a Trustee at Johns Hopkins Medicine Sibley Hospital Foundation. He is on the Boards of Visitors at The Catholic University of America’s Busch School of Business and Conway School of Nursing.