Machine Learning in social services

Purpose and goal:

The project had three objectives: Organizational learning, creating a basis for forecasting and creating knowledge about patterns that precede placements

Organizational learning is achieved with four themes: Prerequisites for AI, Uses for AI, Information management and knowledge of our own Data.

Within the framework of the project, we have not been able to create an active forecasting tool, but we have a model with roughly 87% accuracy that could be developed. We have identified patterns in the data which, after further investigation, could lead to a different way of work. Expected results and effects:

The increased knowledge about AI contributes to better conditions for our municipality to work with AI in the future.

The model for forecasting tools developed will not have any effect on social services in Linköping, as we do not intend to further develop it. Increased understanding of our data will impact the datahandlingprocesses throughout the datalifecycle.

The patterns we have seen may lead to changed working methods and thus contribute to a social service with the right efforts at the right time. Approach and implementation:

Datastructuring took longer than expected. We prioritized completing the project on time and focused on lessons learned more than the forecasting tool, we see the need of strengthening the infrastructure before developing further. Making learning a goal at start was good. Dissemination activities started earlier then planned,

Frequent meetings with data scientists were valuable. Participating in networks with the other projects contributed to both learning and the dissemination of learnings. Collaboration with DPO for secure information management was beneficent.