Emergency department patient volumes are hard to predict, a problem that makes meeting appropriate staffing demands tricky. Overcrowding creates an imbalance between supply (clinicians and staff) and demand (services). Leaner patient volumes result in cost overruns due to a disproportionate number of staff on hand.  

While the forecasting prowess of artificial intelligence (AI) isn’t exactly a crystal ball, it does enable hospitals to better determine staffing levels days, weeks, and even months in advance. We refer to this predictive analytic capability as Dynamic Staffing.  

What Is Dynamic Staffing?  

Dynamic staffing is the practice of adjusting provider staffing needs in real-time or near real-time to match patient volume. It results from a combination of anticipating needs before they happen and flexing coverage on the fly based on acuity changes or volume fluctuations. The goal is to predict volume influences 90 days out.  

Why EDs Need Dynamic Staffing 

The need for dynamic staffing in the ED became apparent during the height of the COVID-19 pandemic, resulting in wide patient volume fluctuations.  

“As a result of COVID, we’ve had some dramatic and crazy fluctuations in our volumes from 50 percent less to almost 50 percent more than we’re used to,” said Dr. Phil Parker, Senior Vice President, Group Medical Officer, SCP Health, in a presentation at SCP’s MLC21 conference held last September. “We want to find a way to anticipate those needs so that we’re not struggling  to either reduce or increase coverage at the last minute.” 

 Provider compensation through RVU plans is another reason dynamic staffing is so essential. With RVUs, the greater the productivity, the higher the compensation. Therefore, matching supply and demand creates ideal efficiency and productivity to guarantee the best pay possible.  

A third reason lies in dynamic staffing’s potential to improve provider satisfaction. “Knowing your schedule 90 days in advance, making sure that you can make your life plans, that gives you a good work-life balance,” Dr. Parker said.  

Creating shift parity is yet another benefit. Dynamic staffing resolves the supply and demand imbalance between providers who work on days when patient volume is greater than those who work when traffic is less.  

Artificial Intelligence: Key to Dynamic Staffing  

The key to unlocking dynamic staffing rests not in the hands of humans but in artificial intelligence, the use of computers and machines to mimic the problem-solving and decision-making capabilities of the human mind.  

Artificial intelligence consists of three constituents:  

• Machine learning, which relies on mathematic algorithms;  
• Neural networks, consisting of nodes, which model the neurons in a biological brain;  
• Time-series forecasting, scientific predictions based on historical time-stamped data.  

According to Dr. Parker, the element that makes AI so magical is its ability to learn from what it’s taught.  

“AI can learn from things we tell it and teach it—that’s called supervised learning—but it can also learn on its own,” he said. “That sounds very mystical, and a lot of people are afraid of that concept. But the truth is, artificial intelligence has an ability to look at numbers differently than we do as humans. It can find correlations that we don’t see on a spreadsheet.” 

Dr. Parker explained that historical volume data points are fed into the computer’s neural network. AI then comparatively reviews the entire database to make connections and correlations that humans would likely never reproduce. 

“One data point is going to get compared to every single data point of every kind of classification,” he said. “Whether it’s weather, acuity, or volume, it’s going to get measured against every single other data point in the whole data package … It’s going to run millions of times, and the more it runs, the smarter it gets.” 

AI’s relevance to dynamic staffing lies in its ability to output accurate patient volume data for every day of the year, more than enough time to allow medical directors to make equitable provider staffing schedules.  

Despite AI’s staffing forecast accuracy, there may be times when the need to flex up or down arises. The difference is in the amount of flexing necessary.  

“The beauty of getting very close to what the need is when you need to flex up, you’re not having to flex up very much, maybe one or two hours,” Dr. Parker said, citing the following example.  

“If you’re at a location and you see 50 more patients than you normally do on Mondays, you know ahead of time and we can staff you with two- or three-hours extra coverage. Then, if you do need to flex up, you only have to flex up maybe another hour to capture those patients. That’s the beauty of this model.” 

He also suggested there may be times when AI spots a need that we may not see exists.   

“It may say the third Saturday in February requires a couple of hours of extra coverage. And, intuitively, that might be confusing and be difficult to trust. But I can tell you, after having done this for a few years, the AI is pretty smart and rarely gets things wrong. And even though we may not understand the correlation, I can tell you that it’s very highly likely that it is correct.” 

Dynamic Staffing Benefits

The benefits dynamic staffing offers directly relate to the needs mentioned earlier. In summary, dynamic staffing enables hospitals to:    

• Set ED schedules up to 90 days in advance;
• Promise better work-life balance;  
• Increase shift parity; 
• Improve provider satisfaction;  
• Optimize RVU-related productivity to ensure higher pay;  
• Reduce flex up and down needs. 

Currently, SCP Health is piloting AI-driven dynamic scheduling at four client sites. Based on our results, we plan to expand the pilot to more facilities in 2022 and include hospital medicine in the mix. 

To learn more about dynamic scheduling, contact SCP Health.