J C Boyd1, J Savory. 1. Department of Pathology, University of Virginia Health System, PO Box 800214, Charlottesville, VA 22908, USA. jboyd@virginia.edu
Abstract
BACKGROUND: Staffing core laboratories with appropriate skilled workers requires a process to schedule these individuals so that all workstations are appropriately filled and all the skills of each worker are exercised periodically to maintain competence. METHODS: We applied a genetic algorithm to scheduling laboratory personnel. Our program, developed in Visual Basic 4.0, maximizes the value of a fitness function that measures how well a given scheduling of individuals and their skills matches a set of work tasks for a given work shift. The user provides in an Excel spreadsheet the work tasks, individuals available to work on any given date, and skills each individual possesses. The user also specifies the work shift to be scheduled, the range of dates to be scheduled, the number of days that an individual stays on a given workstation before rotating, and various parameters for the genetic algorithm if they differ from the default values. RESULTS: For >22 months, the program matched individuals to those tasks for which they were qualified and maintained personnel skills by rotating job duties. The schedules generated by the program allowed supervisory personnel to anticipate dates far in advance of when worker availability would be limited, so staffing could be adjusted. In addition, the program helped to identify skills for which too few individuals had been trained. This program has been well accepted by the staff in the clinical laboratories of a 670-bed university medical center, saving 37 h of labor per month, or approximately $11,000 per year, in time that supervisory personnel have spent developing work schedules. CONCLUSIONS: The genetic algorithm approach appears to be useful for scheduling in highly technical work environments that employ multiskilled workers.
BACKGROUND: Staffing core laboratories with appropriate skilled workers requires a process to schedule these individuals so that all workstations are appropriately filled and all the skills of each worker are exercised periodically to maintain competence. METHODS: We applied a genetic algorithm to scheduling laboratory personnel. Our program, developed in Visual Basic 4.0, maximizes the value of a fitness function that measures how well a given scheduling of individuals and their skills matches a set of work tasks for a given work shift. The user provides in an Excel spreadsheet the work tasks, individuals available to work on any given date, and skills each individual possesses. The user also specifies the work shift to be scheduled, the range of dates to be scheduled, the number of days that an individual stays on a given workstation before rotating, and various parameters for the genetic algorithm if they differ from the default values. RESULTS: For >22 months, the program matched individuals to those tasks for which they were qualified and maintained personnel skills by rotating job duties. The schedules generated by the program allowed supervisory personnel to anticipate dates far in advance of when worker availability would be limited, so staffing could be adjusted. In addition, the program helped to identify skills for which too few individuals had been trained. This program has been well accepted by the staff in the clinical laboratories of a 670-bed university medical center, saving 37 h of labor per month, or approximately $11,000 per year, in time that supervisory personnel have spent developing work schedules. CONCLUSIONS: The genetic algorithm approach appears to be useful for scheduling in highly technical work environments that employ multiskilled workers.