Mental health treatment paths in occupational health care (MielenPolut)
The overall goal of the project is to assess mental health treatment paths in occupational health care by utilising machine learning methods and predict future ability to work based on treatment paths, individual factors and workplace features.
Our aim is to identify paths, events and service themes that predict a future mental health diagnosis or ability to work. A treatment path is widely understood to include everything from the basic processes of occupational health care cooperation and events preceding a mental health diagnosis to various support measures that enhance the ability to work.
The project findings will help identify paths that lead to a positive outcome, which will allow treatment to be planned efficiently and promote the preventative work done in occupational health care.
The sub-objectives of the study include:
- identifying and categorising occupational health care treatment paths pertaining to mental health, so that they can be utilised in creating prediction models and determining practical guidelines
- generating new information on the ways in which business texts connected to the occupational health care cooperation’s core processes (health check-up, workplace review and action plan) steer the treatment paths linked to mental health
- creating a categorisation model that is able to predict the risk of a mental health diagnosis at the individual level, based on business texts, occupational health care surveys and events preceding a diagnosis, and identifying the key predictive factors
- predicting the future ability to work and identifying those occupational health care treatment paths that will have a positive impact on the future ability to work
- generating new information on the seasonal fluctuation of mental health related service themes and treatment paths, as well as on their temporal changes in connection with the coronavirus pandemic.
Materials and techniques
As our research material, we will utilise service event data, medical records, occupational health care surveys, workplace reviews and action plans from occupational health care. The material consists of a combination of various types of texts, survey data and records. It includes the service event data of over 100,000 occupational health care clients. All material will be machine processed without direct identifiers. The research permit for using the material is granted by Findata.
The methods used in the project involve various types of machine learning models. The machine learning methods used include clustering, process mining, topic models and classifying predictive modelling. The first three methods will allow the material to be processed into a form that will enable any part of the material to be utilised as prediction model variables.
Results and effectiveness
The research findings will be published in scientific articles, conference presentations and in a final report, which will be made openly available. In addition to that, the findings will be presented at events and in publications intended for occupational health care professionals, as well as in the Work-Life Knowledge Service.
Based on the project findings, we will develop knowledge-based operating models for occupational health care and occupational health care cooperation together with professionals of these spheres.
The operating models are divided into sub-categories based on the following themes:
- the significance of occupational health care cooperation in promoting mental health and the ability to work
- early identification of mental health problems
- planning treatment paths to achieve positive effects on the ability to work.
Inquiries regarding the project
Project Manager, Research Manager
The project is carried out in collaboration with Terveystalo.
The Finnish Work Environment Fund