An integrated routing and scheduling problem for home healthcare delivery with limited person-to-person contact

This work presents a generalized model for the integrated problem of routing and scheduling of the home healthcare delivery staff (caregivers) to maximize the revenue generated. The model considers most of the commonly used constraints from the extant literature and specific concerns such as patient's preferences for the gender and language of the caregiver, inconveniences time window, and multiple visits for certain procedures that are motivated from the case of an Indian home healthcare service provider. The model is capable of handling multiple visits as well as multiple staff requirements for procedures in order to generate proper assignments, schedules, breaks and routes. A unique prospect of the model is the consideration of limited person-to-person contact to minimize the risk of exposure. The model tries to serve maximum number of patients fully without violating a predetermined maximum allowed contact limit for every patient as well as healthcare staff. Additional policy decision of allowing partial accommodation of patient's request to maximize resource utilization along with the financial viability of hiring additional capacity to fulfill all the demand is also tested. To solve the model efficiently, two different heuristic procedures based on mixed-integer programming decomposition are proposed. Further, a modified Genetic Algorithm, called p-GA, that utilizes inherent parallelism in the evolutionary process is also developed. Application of the proposed model and solution approaches to the practice is tested through extensive numerical experiments on the hybrid problem instances and some benchmark instances from the literature. (C) 2022 Elsevier B.V. All rights reserved.

Automated summary

This work presents a generalized model for the integrated problem of routing and scheduling of home healthcare delivery staff to maximize revenue. The model considers various constraints and specific concerns, such as patient preferences and limited person-to-person contact. Two heuristic procedures and a modified Genetic Algorithm are proposed to efficiently solve the model.