Arrow Left Health care

Use case: Using artificial intelligence to support treatment decisions in mental health

Developing an online tool that supports therapists when making treatment decisions for patients with depression and/or anxiety.

Project aim

Developing an online tool that supports therapists when making treatment decisions for patients with depression and/or anxiety. The tool should make a prediction of the chance of success halfway a specific treatment, based on patient symptoms combined with the original diagnosis and information like age and gender. 

Medical relevance

The tool could improve quality of life of patients and avoid costs for unnecessary treatments by helping therapists in deciding about whether switching to another treatment could benefit the patient.

Summary of the project

The Personal Health Train provides a good approach for the researchers in this project to preserve privacy of patients, thus conducting their research in line with the latest General Data Protection Regulation, while still making use of information from multiple institutes. In this project four mental health institutions will jointly work towards an optimal prediction model without data ever leaving the institute.

Each institute uses the same variables with standardised rating scales and standardised names. The participating institutes, however, still use their own questionnaires to measure interim outcomes.  Standardising these turned out a bridge too far, but with standardised rating scales this isn’t a big problem.

Moreover, each of the four institutes will also do their own analyses, in order to find out whether there is real added value in combining information of different institutions.

The researchers will use the open source software of IKNL, the Dutch quality institute for oncological and palliative research and practice, and Maastro Clinic to implement the concept of the Personal Health Train in their study.

Main results

The project started in September 2019 and first results are expected at the end of 2020.

Lessons learned

It proved to be a good idea to appoint one specific person at each cooperating mental health organisation who has an overview of the available patient characteristics and available routine outcome monitoring measurements, and who has time to standardise the data.

Follow up

Implementation of the tool that should give therapists a warning halfway a treatment when there is a serious chance the treatment doesn’t benefit the patient. Next step is a clinical evaluation of how the tool is used, whether the results are valid in clinical practice, and where it could be further improved.

Project details

Project leader

Joran Lokkerbol
Funders SIDN fonds
Collaboration partners

Trimbos Institute, Radboud University, Leiden University, GGZ Noord-Holland Noord, GGZ Rivierduinen, Indigo, Dimence and Maastro Clinic

Share this page…