Jonathan Shoemaker, the Chief Executive Officer of ABOUT, stands at the intersection of veteran healthcare leadership and cutting-edge information systems. With over 25 years of experience, including pivotal roles as the Chief Information Officer and Chief Improvement Officer at Allina Health, he has witnessed firsthand the friction that occurs when hospital operations fail to keep pace with patient needs. His career has been defined by a commitment to performance transformation, having navigated the complex landscapes of systems like NorthPoint Health and Wellness Center and Hennepin County Medical Center. Today, he is focused on a looming crisis in the American healthcare infrastructure—a capacity bottleneck that threatens to redefine the standards of patient safety and clinical efficiency over the next decade.
Our discussion centers on the urgent need for a paradigm shift in how health systems manage their daily operations. We explore the transition from “passive” data analysis, where administrators look at historical reports to see what went wrong, to “active” artificial intelligence that is woven directly into the clinical workflow. The conversation covers the financial and human toll of nurse turnover, the psychological weight of emergency department boarding, and the technical hurdles of post-acute transitions. By examining three specific use cases—predictive forecasting, patient-readiness signaling, and referral acceleration—we uncover how intelligence can be applied to existing data to unlock throughput without adding to the administrative burden of an already exhausted workforce.
With national hospital occupancy projected to reach a critical 85% threshold by 2032, what does this trajectory mean for the daily reality of patient safety and the functional integrity of our healthcare systems?
The shift from our pre-pandemic baseline of 64% to the current average of 75% has already created a palpable sense of friction in every hallway and waiting room across the country. When we talk about hitting that 85% mark in 2032, we aren’t just discussing a number on a spreadsheet; we are talking about a “functional bed shortage” that fundamentally changes how care is delivered. At that level of saturation, the system loses its elasticity, meaning any small surge—a bad flu season or a local accident—can send the entire facility into a tailspin where patients are stuck in hallways and surgeries are canceled. You can feel the tension in the air during those peak times, as the physical limits of the building begin to compromise the clinical team’s ability to provide timely, high-quality care. This 11-point increase we’ve seen is a warning shot, signaling that the traditional ways of managing capacity are no longer sufficient to keep patients safe.
Given that the average hospital faces a 16.4% RN turnover rate and significant vacancy levels, how is this workforce instability directly impacting the logistics of discharge planning and patient movement?
The workforce crisis is the hidden engine driving our capacity issues, because when you have a national vacancy rate of 9.6%, the remaining staff are forced to prioritize immediate clinical tasks over the “administrative” logistics of moving patients through the system. Each percentage point of nurse turnover costs a hospital approximately $289,000 annually, but the human cost is even higher as discharge planning and care coordination often fall by the wayside. Nurses are stretched so thin that they simply don’t have the bandwidth to stay ahead of the paperwork or the multi-step coordination required to clear a bed for the next person waiting in the emergency department. It creates a vicious cycle where the chaos of a full unit leads to more burnout, which leads to more turnover, further eroding the team’s ability to manage throughput effectively. We have to find ways to hand back the time these professionals lose to manual coordination so they can return to the bedside where they are most needed.
You have noted that many health systems are trapped in a cycle of “passive analytics.” Why have traditional data warehouses and retrospective dashboards failed to solve the throughput crisis?
For more than a decade, we have poured millions into interoperability and massive data warehouses, but the resulting reports often arrive a week after they could have actually made a difference. These retrospective dashboards are like looking in a rearview mirror while trying to drive through a thick fog; they tell you exactly where you hit a pothole, but they don’t help you avoid the one right in front of you. The problem is that the data is passive, living in a separate tab or a weekly email rather than sitting inside the workflow where a doctor or a bed controller is making a decision right now. We see beds sitting empty while patients wait in the ED because the right information didn’t surface at the right moment. To truly unlock throughput, we have to move away from explaining why we failed last Tuesday and start providing intelligence that drives action in the present second.
How does applying machine learning to historical patterns fundamentally change the “morning huddle” and the way hospital leaders manage daily capacity?
The traditional morning huddle is often a reactive exercise where everyone realizes the house is full, the ED is already boarding patients, and the surgical schedule is at risk before the day has even truly begun. By the time the huddle starts at 7 a.m., the bottlenecks have already formed, and leadership spends the rest of their shift frantically putting out fires. Machine learning changes that posture by analyzing variables like payer mix, seasonality, and unit classification to provide a dynamic, continuously updated forecast of who is likely to leave and when. Instead of a static snapshot, it offers a forward-looking view that allows staff to queue up the next patient before the current bed is even vacant. This proactive discipline means surge protocols can be activated based on what is coming tomorrow, rather than what arrived yesterday, allowing the team to breathe a little easier.
If AI can monitor clinical signals like vital signs and mobility status, how much of a difference does it make to identify discharge readiness just a few hours earlier in the day?
The difference between a patient being discharged at 11 a.m. versus 5 p.m. might seem small on an individual basis, but across a busy inpatient floor, that gap is the primary cause of emergency department boarding. When AI flags approaching readiness hours or even days early by reading lab values and oxygen requirements, it gives the care team a massive head start on the logistical nightmare of transport and family communication. We’ve seen that about 5% of patients admitted during peak times are forced to wait 24 hours or longer for a bed, which is a staggering statistic that speaks to the failure of our current timing. By shortening the length of stay through early identification, we don’t just open a bed; we reduce the dangerous congestion in the ED and ensure that case managers aren’t rushing through a transition at the end of a long shift. That head start is the margin of safety that keeps the entire hospital functioning smoothly.
The transition to post-acute care is often a major bottleneck. How can technology assist clinicians who are currently forced to comb through hundred-page referral packets?
When a patient is ready for a skilled nursing facility or rehab, their clinical history is sent over in a packet that can be dozens or even hundreds of pages long, creating an administrative wall that a human clinician has to climb. This review process is a significant drain on time, and every hour that a referral sits in a queue at a receiving site is another hour that an upstream hospital bed remains occupied. AI can act as an intelligent highlighter, surfacing the most critical information like active diagnoses, medication lists, and wound care needs so the intake coordinator can make an informed decision in minutes instead of hours. The technology isn’t making the call—it’s simply giving the clinician the essential context they need to say “yes” more quickly. This compression of the cycle time not only frees up capacity but also prevents the breakdown of referral relationships that occurs when communication is slow and cumbersome.
Clinician trust is often the biggest barrier to technology adoption. How do you ensure that these AI tools are viewed as helpful partners rather than intrusive interruptions?
The version of AI that earns clinician trust is the one that doesn’t ask them to learn a new system or interpret an abstract, confusing score that doesn’t match what they see at the bedside. It has to be embedded so deeply into the existing workflow that it feels like a natural extension of their own judgment, surfacing the information they would have wanted to know anyway at the exact moment they need it. It shouldn’t gate their decisions or pretend to “make the call”; instead, it should act as a silent partner that monitors the noise so the clinician can focus on the patient. When technology shows up at the right time with the right information inside the screen they are already using, it moves from being a burden to being a tool for empowerment. This is how we move the conversation from “more data” to “more action,” creating a system where the intelligence finally serves the people who are doing the hard work of healing.
What is your forecast for the evolution of hospital operations over the next decade as these technologies become more standard?
I believe the next ten years will see the death of the “dashboard” as the primary tool for hospital management and the rise of the “actionable clinical environment” where data is entirely invisible because it is so well integrated. We will reach a point where the predictive models are so accurate that bed management will feel less like a crisis and more like a synchronized air traffic control system, where every move is anticipated hours in advance. Hospitals that fail to make this transition from passive analytics to embedded intelligence will likely struggle with insurmountable safety risks and financial strain as occupancy continues to climb toward that 85% threshold. Ultimately, the successful health systems of 2032 will be those that used this decade to turn their data from a heavy burden of the past into a light that illuminates the path forward for every patient and provider.
