The Power of Predictive Analytics

Puneet Pandit

The beginning of 2022 brings more of the same:  acute staff shortages and increasing pressure from physicians and insurers to move the patient through the hospital in the shortest amount of time. 

Staffing continues to be a challenge for small regional hospitals, large medical centers, and hospital corporations. From med techs to top of license clinicians, the stress of performing critical tasks can be overwhelming when the imaging and diagnostic equipment suddenly stops working. The issues cascade across the organization: patients are inconvenienced with long wait times and call backs, physicians cope with delaying treatment and rescheduling operating rooms, the length of the workday increases for everyone as they adapt to a fewer number of available hours to complete the tests scheduled for the day.

The state of diagnostic machines has been improving. Manufacturers have embraced IoT (internet of things) in order to digitize machines. Machines generate logs and the logs are monitored to quickly identify failures. This has improved on automating response. Response could be calling a technician, replace a bulb, adjust room temperature. In combination with keeping an inventory of frequently failing parts and service contracts that dispatch the manufacturers’ service personnel based on monitored events, the availability has improved.

But what if you could know in advance of a pending failure, like predicting that a critical part is about to go bad? Today the capabilities of machine data analytics, supported by advanced algorithms, provide greater visibility. Visibility gives you more runway, gives you the ability to get out ahead of degrading service. Your team can schedule parts replacement and diagnostics in the off hours rather than wait for a failure to disrupt the day. Imaging teams stay to their schedule, patients are seen at their appointment times, clinicians are utilized for 100% of their available hours. Predictive analytics is at the core of optimizing operations, lowering costs, and improving the patient experience.

First, lets look at some numbers. Here are two examples provided by Glassbeam users of what predictive analytics did for them:

> Over 40% reduction in meantime to resolution (MTTR) for service tickets resulted in revenue increases by 30%

> Improved uptime: Improving the uptime for MRI and CT scanners by just 2 points – from 97% to 99.9% – increased revenues for one hospital by $4 million dollars

Glassbeam Clinsights provides data driven insights that will change your dialogue from ‘what happened yesterday’ to ‘what can happen today’?  Predictive analytics enables you to take actions during downtime to prevent interruptions that negatively impact patients and clinicians.

Health system executives have been forced by pandemic working conditions to change their operational approaches. Executives are taking these changes and applying them as part of the ‘new normal’ – the ability to be proactive and use predictive analytics, adding value over and above automating response. Proactive and predictive are the two ways to describe operational changes learned in the pandemic and sustaining the impact of operational optimization is the key for the future.

The use of predictive analytics can prevent problems that impact machine availability and utilization. So, what is predictive analytics? Predictive analytics uses historical data to predict future events. Typically, machine learning relies on historical data and applies algorithms or models that capture critical trends. is used to build a mathematical model that captures important trends. That predictive model is then used on current data to predict what will happen next.

By extension, the predictions can suggest actions you can take to deliver an optimal outcome. Using predictive analytics for equipment maintenance, or predictive maintenance, enables you to anticipate equipment failures and reduce operating costs. For example, sensors that measure cryogen levels MRIs can predict service problems and lead to higher machine uptime by avoiding related problems. Another example is predicting the need to replace tubes in CT scanners. Fewer service tickets lead to high machine uptime. Machine learning techniques are used to find patterns in data and to build models that predict future outcomes.

The most valuable outcome of using predictive analytics is prevention. While machine data analytics can be used to effectively automate response to a problem,  predictive analytics prevents the problem from occurring in the first place.