Improving the efficiency of patient diagnostic specimen collection with the aid of a multi-modal routing algorithm

The Sustainable Specimen Collection Problem (SSCP), in which diagnostic specimens are collected from GP surgeries (doctor's office/clinics) and subsequently transported to a hospital laboratory for analysis using more sustainable transport modes, is introduced in this paper. Using a weighted objective function, we solve a new multi-objective problem using cycle consolidation to limit driving time and the numbers of vans used whilst improving overall service quality, reducing costs and emissions. This particular heterogeneous vehicle routing problem is explored and applied to two real-world case studies in the UK, where 97 and 22 sites (respectively) are currently served, using a column generation based heuristic algorithm with some additional improvement heuristics. The results demonstrated a potential improvement in the system's maximum delivery time between 41% and 74% compared to business-as-usual activity using solely road vehicles. Road vehicle (van) fleets could be reduced by up to 40%, and the total driving time across the fleet by between 41% and 65%. Operational costs were estimated to increase by up to 38%, though additional workloads for gig-economy cycle couriers and improvement in specimen quality and service reliability may make this trade-off worthwhile. Tailpipe CO2 emissions were also reduced by up to 43%. The proposed algorithm was effective, reducing computational time by up to 99% whilst achieving an average of 5% deviation from optimality.

Automated summary

This paper introduces the Sustainable Specimen Collection Problem (SSCP), where diagnostic specimens are transported from GP surgeries to a hospital laboratory using more sustainable transport modes. By using a weighted objective function and cycle consolidation, this study solves a multi-objective problem to improve service quality, reduce costs and emissions. Applied to two real-world case studies in the UK, the proposed algorithm significantly improved delivery time, reduced vehicle fleets and driving time, and decreased CO2 emissions. The computational time was also reduced by 99% while achieving a 5% average deviation from optimality.