At LeanTaaS, our focus is to use predictive analytics, optimization algorithms, machine learning and simulation methods to unlock the capacity of scarce assets in a health system – a challenging problem due to the high variability inherent in health care.
The solution must be able to generate recommendations that are specific enough in order for the front line to make hundreds of tangible decisions each day. The staff must have the confidence that the machine arrived at those recommendations having processed vast amounts of data in addition to having learned from all of the changes in the patient volume, mix, treatments, capacity, staffing, equipment, etc., that will inevitably occur over time.
Consider a solution that provides intelligent guidance to schedulers on the right time slot in which a specific appointment should be scheduled. Machine learning algorithms can compare the patterns for the appointments that were actually booked versus the recommended pattern of appointments. Discrepancies can be analyzed automatically and at scale to classify the “misses” as either unique events, scheduler errors or an indicator that the optimized templates are drifting out of alignment and therefore warrant a refresh.
As another example, there are dozens of reasons why patients may arrive early, on time or late to their scheduled appointments. By mining the pattern of arrival times, algorithms can continuously “learn” the degree of punctuality (or lack of) based on the time of the day and the specific weekday. These can be incorporated into making specific tweaks on the optimal appointment template so that they are resilient to the inevitable shocks and delays that occur in any real-world system involving patient appointments.
