eScience Engineer Dr Erik Tjong Kim Sang on 'What Works When for Whom', a joint interdisciplinary project on improving the effectiveness of e-mental health (EMH) interventions.
Mental health issues like depression and anxiety are among the leading causes of the global burden of disease. To improve access to psychotherapy for a wider audience, health care professionals are increasingly using e-mental health interventions (EMH) such as web-based psychotherapy treatments. Whereas these treatments tend to be equally effective, the responsiveness to a specific treatment shows large differences. The personalization of treatments is seen as a major road for improvement
In the interdisciplinary project ‘What Works When for Whom’, the Netherlands eScience Center has teamed up with Dr Anneke Sools from the University of Twente (UT) to use eScience methods and tools to analyse patterns in therapy-related textual features in email correspondence between counselor and client. By connecting patterns of known change indicators, the aim is to identify which therapy works best for whom and, ultimately, improve the effectiveness of EMH.
In the following interview, Dr Erik Tjong Kim Sang briefly talks about the project and the progress that’s been made.
How far along is the project?
In September 2019 we entered the final year of this 4-year project.
What is the eScience Center’s exact contribution?
The eScience engineers are working on two topics: text analysis and data visualization. The project data consists of a collection of emails written by therapists and their patients. By performing an automatic text analysis, we hope to reveal patterns in the mails that can help the domain researchers to better understand the effects of the therapeutic process. We use data visualizations to find the patterns discovered by the text analysis.
What are the major technical challenges you’ve run up against?
Like many of our other projects involving textual data, the major challenge is how to deal with the available data. There is never enough data available to adequately deal with the plans of the domain researchers. Furthermore, the data weren’t immediately ready to use for the purposes of the project. Cleaning the data is a process that has taken years in this project and which still remains to be finished.
Two weeks ago, you held a hackathon with students at UT. What was the aim and outcome?
The hackathon had three aims:
1. Integration: collect four different text analysis tools in one environment.
2. Modularization: separate the text preprocessing and data visualization tasks from these analysis tools so that in future they can be used with other preprocessing and visualization software.
3. Training: learn to work with and to develop for Orange3, the platform system on which we will offer the tools.
Obviously, privacy is a major issue when dealing with personal data, especially in health care. How are you ensuring the data you work with remains secure and anonymous?
All personal information has been removed from the emails. We have agreed with the data provider on a protocol for dealing with possible misses of the automatic anonymization process. We have received the data on encrypted password-protected USB sticks. Making copies of the data is discouraged and may only be done to hardware which is also password-protected and encrypted. Every researcher and student who receives access to the data has to sign a form in which they pledge to deal with the data in a responsible way.
What will you do with the tools and methods you develop for this project?
Like our other projects, the tools we develop will be made available to the research community on Github. For this particular project, the management of the tools will be transferred to the BMS Lab, the software support department of the UT’s Faculty of Psychology. They will maintain the tools after the project has finished.
When will the project be completed?
The project will be completed on 31 August 2020. By that time, we want the domain researchers to have access to flexible text analysis tools that will enable them to study larger volumes of text than before. The researchers will definitely develop new ideas for automatic analysis once the project ends. We hope that in future the BMS Lab and their computer science students will be able to help the researchers to improve and expand the tools we’ve developed.
Read more about the project
Read more about Anneke Sools