Predictive Analytics Curb the Cost of Healthcare — An Hour at a Time

Posted by Sarah Anderson on Wed, Sep 04, 2013

Predictive analytics–based staffing solutions can save hospitals millions.

nurses-1.jpgWhile we wait to see how insuring more of the population affects healthcare costs for patients, hospital groups are also working hard to reduce the cost of healthcare. Some hospitals are reducing unnecessary department labor costs — in some cases staff hours have been reduced by up to 8% — by using solutions that analyze Big Data to predict patient inflows. With accurate patient forecasts, hospitals can optimize scheduling to better match supply (nurses) to demand (expected patients). Not only does this reduce costs, it makes their hospitals more efficient and even increases patient satisfaction.

By accurately predicting patient inflows, hospitals can reduce the number of nurses on shifts that are predicted to be slow and increase the number on shifts that are predicted to be busy. Many hospitals currently rely on personal experience or historical averages to achieve a patient forecast. While this is a start, the accuracy of these approaches is constrained by the narrow scope of data being evaluated. 

Getting the right data

To increase the accuracy of patient forecasts, the approach must incorporate data from additional internal systems as well as external data sources. For instance, internal systems can provide details on length of stay in each department, which affects how much time a nurse spends with a patient. Having this level of information can help hospitals adjust service time for hospitals that tend to treat patients for longer periods of time or flag potential issues that hospitals need to address. If one department has a much higher length of stay than others, for instance, you can find out why. The answer could be a shortage of beds, a shortage of techs to transport patients, or any number of other factors.

Outside of the hospital, a pop-up clinic opening nearby, major planned events, or a local flu outbreak can impact patient inflow. Incorporating data feeds that include such information is essential to developing accurate patient forecasts.

Seeing historical data in a new way

Historical data is typically broken down by hour, day, week, month, season, and year. But it needs to be broken down by other factors as well. Holidays, for instance — especially those with floating dates — can provide a level of insight not readily available when making future predictions based on historical data. So while a hospital can easily predict the inflow for Christmas because it’s consistent from one December 25th to the next, historical data won’t provide any inherent insights about Rosh Hashanah or Easter, both of which occur on a different date each year.

Taking this example one step further, making the correlation between holiday, date, and day of week also provides additional insight. For holidays with static dates, the impact can change depending on what day the holiday falls. When a system breaks down data by holiday, it can quickly ascertain the impact of the 4th of July when it falls on a Saturday versus a Wednesday.

Measuring the impact on hospitals, staff, and patients

We’ve found that hospitals that use regression models that incorporate this level of data produce forecasts that are 18% more accurate than solutions using historical averages alone. The impact of accurate patient forecasts — and subsequently more accurate scheduling — can be felt throughout the entire hospital. With optimized scheduling, both patients and nurses experience increased satisfaction.

Patients have reported higher satisfaction rates thanks to shorter wait times and happier nurses. Nurses are happier because they have fewer schedule disruptions (call-ons and call-offs) and they’re less likely to be overwhelmed — or even bored — by the patient load. This makes nurses less likely to quit, which means patients get nurses who are experienced and familiar with the facility and processes, making for a generally smoother patient experience.

Thanks to a comprehensive solution that incorporates all the necessary data to generate accurate forecasts, hospitals are seeing benefits in all areas: financial, operational, and clinical. One hospital group, which tried an advanced staffing solution in seven of its hospitals, reduced its personnel hours by 3–8 percent. In an industry that pays overtime, this adds up to millions of dollars annually. We predict hospital groups can see a 400% ROI over three years with such a solution. 

A real solution, available now

Opera Solutions offers a staffing solution that does all this — and much more. In fact, forecasting patient census is really just the beginning. It uses advanced optimization techniques to help hospitals meet patient and financial targets; produces a nearly complete schedule, drastically improving productivity for department directors and allowing them more time to spend with staff and patients; and incorporates an intuitive cloud-based user interface that gives nurses and techs real-time access to their schedules.


Read more about our Hospital Staffing Optimizer by downloading our newest white paper.

  Download Now



Topics: Healthcare, Big Data