As people get older mobility tends to decrease and there is more need for long-term support in daily activities. There is a higher preference to live at home as compared to staying in a hospital or a nursing home, and home care was found to be more cost-effective (OECD, 2017). This is happening worldwide as a consequence of the continuous increase in longevity (see figure 1). Thus, demand for home health and social care is on the rise. 


Home care planning means scheduling the care services that a set of caregivers give to different users during a period of time. This task may seem simple, but it is extremely complex and time-consuming for actual planners due to the different goals pursued (cost minimization and good quality of service) and the large number of working regulations and user-related restrictions. 

Nowadays, planners do these tasks manually. They typically split the pool of caregivers or users into small groups and, based on their experience, assign services to caregivers while trying to account for all the regulations and logistic restrictions. Social and health care providers are interested in a system that can automate and accelerate the laborious planning process, and they are also keen to find answers to some logistical questions. Operations research methods can provide solutions for home care that reduce costs and improve the quality of service for caregivers and users. The experience of care providers in Barcelona inspired the design of a tool based on two intelligent algorithms.

The importance of caregivers’ satisfaction with their work is crucial for good quality of service. Thus, organizations in Barcelona are considering the scenario of transitioning some of their part-time workers into full-time workers. This would increase their satisfaction and they would not have to complement their salary with additional jobs. 

Assessing the outcomes of certain scenarios reflecting questions of interest is of utmost importance for home care organizations. Thanks to the decision support system the planner can compare costs and quality of service and decide. For instance, a shift from part-time to full-time work involves a reduction in the number of unscheduled services, a slight improvement in continuity of service, and an increase in non-effective working time and cost.

Improving the workload balance between caregivers with the same type of contract is also important. Figure 3 compares the workload using the manual current solution, and one solution provided by the tool, showing that the latter involves a better balance, i.e., there is less variation in working time across caregivers, and this in turn should improve their job satisfaction.

A hospital nurse may need to collect a blood sample from a fasted user before a social caregiver gives him/her the first meal of the day. In these situations, users receive health and social services, raising the need for coordination between both. Planning combined home care services is a very difficult task for current home care organizations, since it involves extra restrictions to synchronize both types of services.  

The system makes it possible to specify dependencies between jobs, which should increase the quality of service for integrated health and social home care. Although this may initially jeopardize the obtaining of feasible solutions and/or worsen some indicators, there is also a positive effect: in the long run it avoids repetition or loss of services (e.g. when caregivers arrive on time, but finally it is not possible to serve the user due to lack of coordination with the other service), and users perceive better quality when both services are planned jointly.

The experience of care providers in Barcelona inspired the design of a decision support system, a tool based on two intelligent algorithms. One operates from the users’ perspective (CxU), which builds solution planning services user by user; and the other operates from the caregivers’ perspective (CxC), which builds solution planning services caregiver by caregiver. Both of them can yield high-quality solutions when planning home care services while optimizing for different objectives, such as minimizing non-effective working time and maximizing continuity of service. They also help to optimize secondary objectives such as minimizing the number of unscheduled services, workload differences between caregivers, and starting time differences between services for the same user over several days. Additionally, they can be adapted to plan other areas with different requirements. 

Thanks to care providers’ experience, these algorithms match reality and it is possible to compare their solutions with those offered by actual practitioners. The results show that both algorithms can obtain efficient solutions regarding the comprehensive set of objectives considered, with low computational times. The proposed automated planning system is available via a web-based decision support system which planners can use to obtain an efficient solution in an intuitive manner. Figure 4 shows an example of what a planner can see on the platform: the daily scheduling for each caregiver and a summary of the main indicators that define the quality of the solution.

The system can provide various solutions through changes to the different parameters. There is a guide to select the right combination of parameters that the home care organizations can use. Or they can try different parameters and represent solutions such as in figure 5 to trade off alternatives and choose the one that best fits their preferences.

Figure 6 compares a manual solution in an area of Barcelona (current solution) with other solutions using the built system. The results show that the latter works better than the current one in most of the measures, and offers a lower potential continuity of service than the values required by care organizations (two caregivers per user, preferably one). The scheduling of a larger number of services and the reduction of non-effective time may justify accepting higher values of continuity of service. 

This on-line decision support system provides an easy-to-use tool available anywhere and at any time to improve the quality and costs of home care. The intelligent algorithms designed to construct solutions have proven efficient in comparison with current practice, where planners take several hours or days to schedule manually. Moreover, the system makes it possible to quickly attempt different possible combinations of parameters and methods, and generate multiple schedules in a very short time. Planners can adjust parameters according to goals and preferences regarding non-effective time and continuity of service. If caregivers and/or users have specific needs, the tool also makes it possible to obtain schedules that can be manually adjusted according to the organization’s preferences. 

To summarize, this tool provides final, implementable solutions, and also enables testing different options and evaluating the impact of some interventions/decisions. Although it has been designed to address the home care situation in Barcelona, it can be easily adapted to needs or peculiarities of other cities. The benefits for society are clear and direct, since the optimization of these services makes it possible to better serve more users.

The tool has been accepted and validated by the collaborating organizations, but it encountered barriers to implementation because of the lack of coordination between the social and health agents. The main barrier is the absence of a centralized entity in charge that would make this tool officially available to both types of services. Nevertheless, actions are being undertaken with all the entities involved to effectively establish such an organization in the near future and/or to share data.

OECD, 2017. Caring for Quality in Health

Original article available: Optimizing an integrated home care problem: A heuristic-based decision-support system