Monitoring emotional well-being through journaling
NLP-supported analysis

Journaling can be understood as classic diary writing and promises to be an effective tool for understanding yourself better, recognizing your thought and behaviour patterns and documenting your development. It is not only very popular in social media, but also in cognitive behavioral therapy. There are a wide variety of methods: Structuring your thoughts with a bullet journal, starting the day with a 5-minute journal, reflecting on what you are grateful for with gratitude journaling or writing down feelings and thoughts about stressful situations unfiltered with expressive writing. The positive effects of journaling include increased self-efficacy, greater resilience and more life satisfaction. Even an improvement in the symptoms of mental illnesses such as depression could be shown. 

AI encompasses technologies and methods that enable machines to perform tasks that typically require human intelligence. These include learning, pattern recognition, problem solving and decision making.²

Natural Language Processing (NLP) is a branch of artificial intelligence that focuses on processing and analyzing human language. By using NLP, information such as emotions, moods and topics can be derived from texts.

AI-supported diary apps not only make it possible to improve emotional well-being, but also to monitor it at the same time. This can be used to identify patterns and react to negative trends. This monitoring can also hold a mirror up to the user and provide insightful feedback, which is thought to reinforce the positive effects of journaling by encouraging self-reflection. To make journaling even more flexible, many diary apps allow different input modalities. Students can therefore now not only write their diary entries in the traditional way with pen on paper, but also type them into their smartphone or speak them.

The different forms of input are relevant, as spoken language could reflect emotional states more spontaneously and authentically, while typed language could be formulated in a more considered and targeted manner. Due to these differences, it is possible that information about the emotional state is transmitted differently and the input form therefore has an influence on the implementation of the monitoring.

After the participants have agreed to the declaration of consent on SoSci Survey, we ask them to download the Famedly app from the Apple App Store or the Google Playstore. To register, the participants receive an individual QR code from us. No telephone number or e-mail address is required in the app.

We then ask the participants to complete the baseline survey questionnaires on SoSci Survey. These cover the areas of demographics, mental health (anxiety, stress and depression), general well-being and self-control. It takes about 10-15 minutes to complete the questionnaires.

The baseline survey is followed by 14 daily surveys. A journaling entry is created every evening in the Famedly app ("Tell us about your high and lowlights of the day. How did you feel and why?"). Participants are randomly assigned to one of two groups at the start of the study: One group types their journaling entries and the other group records voice messages. Subsequently, questionnaires on emotional state, stress level and self-control are filled out on SoSci Survey, and the topics of the day (e.g. friends, work) are selected from a list. The daily effort is between 5 and 10 minutes. This depends on the length of the journaling entry.

At the end of the study, participants receive individual feedback (see figure). This is created by the study team based on the questionnaires collected and is not AI-generated. It shows the progression of their mood values and the mentioned topics over the study period and is also intended to show the correlations between them. The more extensive the entries are, the more can be reflected to the participants in the feedback. Participants can contact the study team at any time if they have questions about the feedback.

Famedly is a messenger app that was specially developed for the healthcare sector. Data protection is very important here, which is why the app is also GDPR-compliant. It can be downloaded from the Google Playstore or the Apple App Store. Participants receive a personal QR code to register in the app.

SoSci Survey is a GDPR-compliant platform for online surveys. It is not an app. Instead, the questionnaires are filled out in the browser on a cell phone or laptop. Participants receive the relevant links from us via the Messenger Famedly.

The study can be conducted entirely online. Nevertheless, we have contact persons on site who are also available online.

The personal data is anonymized or pseudonymized, used only for research purposes, treated confidentially and not passed on to third parties.

The feedback will consist of several visual representations of the data collected. The mood and stress values over the study period will be displayed. It will also reflect which topics were talked about a lot. In addition, there should be visualizations that link the topics and mood values. See the illustration in the "Individual feedback" section.

At the end of the study, we would be delighted to hear from participants who are willing to give us their opinion on the various feedback options in an interview. We will make a separate call for this. The insights gained from this will contribute to the user-centered development of journaling apps.

There are no risks for the participants as part of the study.

Participation in the study is voluntary. Participants can stop or terminate the study at any time without giving reasons and without any disadvantages for the test subjects.

Study participants can decide for themselves how many surveys they would like to take part in. Nevertheless, a high willingness to participate is essential for the success of this study and we ask the test subjects to take part in each of the surveys if possible.

In the study, we track the emotional well-being and stress levels of the participants over a period of two weeks. If some of the surveys are missed, it will be difficult to make statements about the progress of the participants' well-being and this will also affect the individual feedback to the participants.

Thanks to our digital application, the survey can take place anywhere and at any time. However, as we ask participants to reflect on their day in their journaling entry, it would be very good for the comparability and high quality of the data if the survey takes place in the evening after 6 pm. You will receive a push notification as a reminder.

No journaling experience is necessary. We explain exactly how it works.

We want to investigate whether the type of input influences the ability to monitor the emotional state. We therefore compare the results of the two groups "speaking" and "typing".

As a free choice would impair the validity of our study, this is unfortunately not possible. The allocation to the groups is random.

Journaling is also suitable for people with a pre-existing mental illness. These participants are therefore also very welcome. However, participation is not possible if there is an acute mental health crisis that requires action, such as acute suicidal tendencies. In this case, inpatient treatment would be indicated and the support provided by JAM-STEP would not be sufficient.

If you notice a significant deterioration in your mood, please contact our study team (jam-step@med.uni-tuebingen.de) or your nearest psychiatric clinic.

Study is currently being conducted; study protocol is in preparation:

• Schmitt, Schlicher (in prep) Monitoring Students' Well-Being through Journal Analysis - Study Protocol of an Explorative Approach Using NLP on Typed and Transcribed Entries to Monitor and Generate Personalized Feedback.