AI-powered dispatching for non-emergency medical transport

Non-emergency medical transport (NEMT) covers planned hospital visits, admissions and transfers for patients with special needs. Transport providers assign each day’s trips to vehicles, drivers and assistants by hand, because general-purpose routing tools on the market are not built for the constraints of healthcare transport: unpredictable return times, waiting and delay, medical complications, equipment requirements, infection risk, and the cleaning required between trips.

This project, delivered for the Belgian software platform Triptomatic and co-funded by the Flemish innovation agency VLAIO, developed an automatic dispatching module that autonomously drafts the day’s schedule and keeps re-optimising it as new information arrives. Dispatchers receive an explainable proposal that they can accept, adjust or override, rather than building the schedule from scratch.

On the operations research side the problem is a large-scale, dynamic combinatorial scheduling problem: the schedule has to be re-planned many times per day as requests, cancellations and delays come in, while respecting the medical, legal, contractual and cost priorities that each provider configures for itself.

I led the research, development and implementation of the combinatorial optimization methods behind the module. The approach was validated in side-by-side pilot runs with provider dispatchers, targeting large reductions in dispatcher planning time, fewer kilometres and overtime hours, better punctuality, and more consistent, less bias-prone schedules.

The dispatching technology and its underlying algorithms are owned by Triptomatic.