Summary

This dissertation investigates dynamic scheduling in healthcare and delivery services, focusing on real-time updates to enhance efficiency. The research explores appointment scheduling in healthcare, where schedules are adjusted based on real-time data, such as patient arrival times or congestion in waiting rooms. With modern technologies enabling real-time updates, this thesis investigates dynamic rescheduling, showing that updating schedules based on current information can significantly reduce costs.

Through dynamic programming and heuristic methods, the research presents various rescheduling paradigms that optimize the timing and frequency of updates, ensuring minimal disruption to clients while maximizing service efficiency. The methods show that a limited number of well-timed updates can yield substantial benefits. To enhance real-world applicability, the research develops a heuristic that provides real-time schedule updates while accounting for practical constraints.

In the domain of last-mile delivery, the dissertation focuses on dynamically updating delivery windows based on real-time driver progress. A method tailored for PostNL was implemented to improve customer experience by narrowing delivery windows throughout the day, allowing for accurate live tracking of parcel arrivals. The research also analyzes driver behavior, revealing deviations from planned routes as a key challenge in accurate scheduling. Using data from Amazon's Last Mile Routing Research Challenge, a method is developed to predict driver routes by learning from past delivery patterns. The integrated approach ranks among the top submissions.
The findings have practical applications in healthcare, logistics, and other service industries, contributing to the broader field of operations management by offering innovative solutions for dynamic, real-time scheduling.