Existing systems did not account for subspecialty information, which was determined to be a critical variable in scheduling, along with a variety of factors that could impact optimal use of appointment slots: Time (Daily, Weekly, Yearly), Appointment Type, Appointment Duration, Patient Geographic Location.
In order to understand the supply of appointments needed, a demand forecast by each subspecialty was evaluated to understand trending and seasonality. Machine learning techniques were used to minimize forecast error. Optimization algorithms minimized the number of staff needed while matching existing calendar templates with forecasted demand.
Subspecialty forecasts, combined with an understanding of the patient’s preference in terms of wait time for an appointment, were used to set the correct level and mix of slots on Neurology’s calendars.