A data driven solution approach for the home health care problem

Development in technology and healthcare has extended the average life expectancy globally and has led to an increase in the elderly population. Due to the change in population and increase in health awareness, the demand and need of individuals for healthcare services that prevent and treat diseases have increased as well. Especially during the COVID-19 pandemic, solutions have been developed for patients and elderly people who should not leave their houses to receive treatment at home. These services are called Home Health Care (HHC) systems. In the HHC system, there are patients who need to receive treatment at home and health care personnel who provide various health services to patients at their homes. In this context, human resource planning in the HHC system is crucial. The decisions that need to be made in human resources planning for HHC system consist of assigning the health care personnel to patients, scheduling the visits of health care personnel to patients, and constructing routes for the visits. The problem that integrates these three decisions is called the Home Health Care Problem (HHCP). Since HHCP is of NP-Hard complexity, solving it in an integrated manner within a reasonable time is not always possible. In this thesis, a HHCP variant that takes continuity of care, time windows of patients, and workload of operators into consideration is solved using a two-phase solution approach where the assignment and scheduling problems are handled in the first stage and the routing problem is handled in the second stage. In order to deal with the possible infeasibility in the routing stage and achieve high-quality results, the solution obtained from the two-phase approach is improved using a matheuristic framework that employs a Variable Neighborhood Search metaheuristic. Within the framework, the resulting routes are evaluated by a data driven method that estimates the route durations using the geographical locations of the patients from the historical routing information. In this context, the main topic of this thesis is to solve the HHCP using a matheuristic framework and to develop a data driven method to improve the solution obtained by the two-phase approach.