Mobile Data Collection for Depression Analysis: An App Framework for

Monitoring Mood and Depression Using Smartphone and Wearable Data

Eliane Schr

oter

1 a

, Franziska Klein

1 b

, Patrick Elfert

1 c

, Fynn Bredehorn

, Julien R

aker

1 d

Frerk M

uller-von Aschwege

1 e

and Andreas Hein

1,2 f

R&D Division Health, OFFIS - Institute for Information Technology, Germany

Assistance Systems and Medical Technology, University of Oldenburg, Germany

Keywords:

Depression, Mood Tracking, Digital Health, Psychotherapy Accessibility.

Abstract:

Depression is a leading cause of disability worldwide, affecting around 5% of the global adult population. To

address this problem, researchers are exploring methods for early detection of relapses, mood swings and their

relation with health data and external inﬂuences. The aim of the present study was to evaluate the usability

and feasibility of a mobile application designed for active and passive data collection, with potential future

applications for improving mental healthcare through a virtual therapy assistant. The application allows users

to self-report their mood, complete PHQ-9 questionnaires, and track measures such as sleep, physical activity,

location, smartphone usage, and social media engagement.

A six-week pilot study was conducted with 22 healthy participants (68% male, 32% female). Participants

recorded their mood three times a day and completed weekly mental health assessments. Results showed

that the application effectively collected relevant data and was user friendly. However, limitations included

reliance on self-reported data, short study duration, and occasional technical issues with data collection. De-

spite these limitations, the study showed that it is possible to use smartphones and wearable technologies to

monitor mental health, laying the foundation for future developments in digital therapeutic interventions and

personalized healthcare through app-based virtual therapy assistants.

1 INTRODUCTION

According to the World Health Organization, depres-

sion affects approximately 5% of the global adult pop-

ulation, making it one of the most prevalent mental

health disorders worldwide (WHO, 2023).

Depression can affect anyone. Risk factors in-

clude, for instance, traumatic experiences, genetic

predisposition, age, pre-existing illness, stress or so-

cial isolation (WHO, 2023). The core symptoms in-

clude persistent depressed mood, loss of interest and

lack of motivation (Nationale Versorgungsleitlinien,

2024). Depression is typically diagnosed when symp-

toms persist for at least two weeks. The severity of the

disorder is determined by the number and intensity of

these main symptoms as well as secondary symptoms

such as fatigue and sleep disturbances (Nationale Ver-

https://orcid.org/0009-0001-4637-0942

https://orcid.org/0000-0002-9752-5203

https://orcid.org/0000-0002-9834-0702

https://orcid.org/0009-0005-5153-906X

https://orcid.org/0009-0001-4960-4097

https://orcid.org/0000-0001-8846-2282

sorgungsleitlinien, 2024). In severe or chronic cases,

depression can lead to signiﬁcant health risks, includ-

ing inability to work, and, in extreme cases, suicide

(WHO, 2023).

Treatment for depression commonly includes cog-

nitive behavioral therapy (CBT) (Kazantzis et al.,

2018), which is often supplemented with medica-

tion in severe cases (WHO, 2023). Despite increas-

ing awareness, signiﬁcant challenges remain, such

as limited access to treatment (Deutscher Bundestag,

2022), misdiagnosis, and reluctance to seek treat-

ment (Kessler et al., 2002). In addition, current

healthcare systems often lack the capacity to provide

continuous monitoring, although the relapse rate for

depression ranges between 40% to 60% (Nationale

Versorgungsleitlinien, 2024). Continuous monitor-

ing of mood, sleep patterns, physical activity, and so-

cial interactions could enable more personalized and

timely interventions, thus improving long-term out-

comes (Mohr et al., 2017). Integrating innovative dig-

ital tools such as mobile applications and wearable

technology offers a scalable and accessible approach

to complement current treatment methods. Thanks

to unobtrusive sensors, these devices can be used to

Schröter, E., Klein, F., Elfert, P., Bredehorn, F., Räker, J., Aschwege, F. M. and Hein, A.

Mobile Data Collection for Depression Analysis: An App Framework for Monitoring Mood and Depression Using Smartphone and Wearable Data.

DOI: 10.5220/0013298700003911

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 18th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2025) - Volume 2: HEALTHINF, pages 781-788

ISBN: 978-989-758-731-3; ISSN: 2184-4305

781

monitor mental health, detect mood swings, and sup-

port early interventions (Torous et al., 2018).

Research has linked several factors to mood

swings in depression, including demographic vari-

ables (e.g., gender, age, relationship status) (WHO,

2023; Grundstr

om et al., 2021), physical activity

(Kvam et al., 2016; Stanton and Reaburn, 2014), vi-

tal signs, sleep quality (Irwin, 2015; Short and Louca,

2015; Rykov et al., 2021), weather conditions (Denis-

sen et al., 2008; Klimstra et al., 2011; Taniguchi et al.,

2022), and location (Masud et al., 2020). Moreover,

research showed that individuals with depression of-

ten have increased phone use, and excessive screen

time was shown to be associated with poorer men-

tal health outcomes (Razavi et al., 2020; Asare et al.,

2021; Javaid et al., 2022). At the same time excessive

social media use, particularly among young women,

is correlated with greater depressive symptoms as

well (Fardouly et al., 2015; Thorisdottir et al., 2019;

Bengtsson and Johansson, 2022; Cunningham et al.,

2021; Drouin and Abbasi, 2019). Moreover, Hong et

al. applied machine learning based to Global Posi-

tioning Service (GPS) and phone usage to detect de-

pressive symptoms (Hong et al., 2022), while Ghosh

et al.’s achieved 78% accuracy in emotion recognition

from typing patterns (Ghosh et al., 2019). Similarly,

Masud et al. integrated several smartphone metrics

such as physical activity and time spent at home to

predict depression severity using PHQ-9 scores (Ma-

sud et al., 2020).

Building on these ﬁndings, we developed a mobile

application (app) that collects both active data such

as mood ratings and questionnaire responses, as well

as passive data from smartphones and wearable de-

vices. This approach aimed to complement existing

mental health apps by incorporating a wider range

of behavioral and lifestyle factors. Although exist-

ing apps have been shown to be effective (Bakker

and Rickard, 2017), they often have limitations. For

instance, MindDoc (MindDoc, 2023) tracks mood

symptom management, but does not integrate social

media data. Woebot (Fitzpatrick et al., 2017) provides

AI-driven CBT and personalized mood feedback, but

does not collect passive behavioral data from wear-

able devices. In addition, many mental health apps

lack formal regulations and rely heavily on user in-

put, often overlooking individual differences.

Our goal was to embed established processes into

an app and evaluate that app, which was iteratively

adapted to meet the speciﬁc needs of the target group

while adding factors such as social media consump-

tion and smartphone usage. By linking mood data

with passively and actively recorded data, the app al-

lows for monitoring a comprehensive range of factors

inﬂuencing depression. Developed within the DAIsy

project (Developing AI ecosystems improving diag-

nosis and care of mental diseases), the DAIsy data

collector (DC) app will be integrated into a larger

ecosystem that aims to address limited mental health

resources, support continuous monitoring of condi-

tions with high relapse risk, and personalizing treat-

ment. Wearables integrated into the ecosystem cap-

ture health and behavioral data, like heart rate and

physical activity, to improve mood monitoring (Klein

et al., 2023). To assess the functionality and usability

of the DAIsy DC app, we conducted a six-week pilot

study with 22 healthy individuals, focusing on quali-

tative assessment of user experience and app perfor-

mance.

2 DAIsy DATA COLLECTOR (DC)

APP

DAIsy DC was developed for Android and collects

data both actively (user input required) and passively

(no user input required). The overall architecture of

the app is shown in Figure 1.

2.1 Passive Data Acquisition

Table 1 summarizes the passive data collected via

smartphones and wearable devices. Wearable devices

such as smartwatches can be integrated via the Google

Health Connect interface. Currently, the app focuses

on three key categories of passive wearable data: ac-

tivity, sleep, and vital signs are included, as these

metrics provide more accurate and reliable measure-

ments compared to manual user input. If wearables

are unavailable, apps like Google Fit can track phys-

ical activity. Heart rate and heart rate variability are

primarily collected through smartwatches or compati-

ble apps such as Google Fit. Similarly, sleep duration

is measured using these sources.

Weather and bio-weather data are obtained from

OpenWeather and the German Weather Service

(Deutscher Wetterdienst, 2024) and capture param-

eters such as minimum and maximum temperatures,

perceived temperature, air pressure, humidity, and the

corresponding date and time. For bio-weather, the

date is logged along with its potential impact on well-

being, pain sensitivity, headaches, sleep quality, con-

centration, irritability, and depression. The location is

also stored in the database in encrypted form.

Phone usage, including screen time and time spent

on speciﬁc apps, is passively recorded using An-

droid’s internal functionality through the Android Us-

age Stats Manager.

Using GPS technology, the user’s location is

recorded and categorized into predeﬁned categories

such as home, work, or leisure areas. Users must

specify these locations once in the app. Any location

not speciﬁed is classiﬁed as ’location’ without further

HEALTHINF 2025 - 18th International Conference on Health Informatics

782

Figure 1: Illustration of the app architecture and data sources.

Table 1: Summary of passively collected data.

Data Type Sampling Rate Source

Physical Activity once per day Smartwatches, Withings App, Google Fit

Sleep once per day Withings Sleep Analyzer, Google Fit

Weather every eight hours German Weather Service, Open Weather API

Location every 15 minutes GPS

Smartphone Usage once per day Android Usage Stats Manager

Social Media Consumption once per day Android Usage Stats Manager

categorization. To protect privacy, all location data

are encrypted.

Compared to existing apps like mindDoc (Mind-

Doc, 2023), DAIsy DC differs in that it captures a

broader range of passively collected data, including

social media consumption, smartphone usage, and

bio-weather inﬂuences. These factors have previ-

ously been underrepresented in other platforms fo-

cusing on mood and well-being, making DAIsy DC

a more comprehensive mental health monitoring tool.

2.2 Active Data Acquisition

Active data acquisition includes user-initiated inter-

actions with the app, including submitting current

moods, providing reasons for that speciﬁc mood and

completing surveys.

2.2.1 Questionnaires

Demographic data is collected via a short question-

naire. The questionnaire includes six questions cov-

ering age, gender, marital status, educational back-

ground, number of children, and smoking habits. In

addition, the app integrates the PHQ-9 questionnaire,

a widely used screening tool for assessing the sever-

ity of depression (Kroenke et al., 2001). It assesses

key indicators of depression such as sleep quality,

physical condition, mood, eating behavior, and suici-

dal ideation. Results are presented graphically, along

with explanations of the scores and links to resources

for further support if needed.

2.2.2 Mood Tracking

The mood tracking feature requires active user input

as it cannot be passively captured. The app uses a

simpliﬁed version of the Circumplex Model of Af-

fect (Posner et al., 2005), which organizes emotions

along two dimensions: valence (positive vs. neg-

ative) and activation (energizing vs. deactivating).

Users choose from a range of emotions (e.g., ”happy,”

”sad,” ”angry,” and ”nervous”) using sliders to adjust

the intensity of their mood. Each emotion is visu-

ally represented by a smiley and ranges from unex-

pressed to very expressed. The values are stored in

the database with values from 0 (unexpressed) to 3

(very expressed) (cf. Figure 2a). In addition users can

adjust sliders for energy levels, lethargy, and general

discomfort. As shown in Figure 2b, mood changes

are visualized on a separate screen, allowing users to

track their emotional states.

2.2.3 Mood Reasons

After submitting their mood, users are asked to indi-

cate the factors that inﬂuence their emotional state (cf.

Figure 2c). Users can choose from a variety of prede-

ﬁned categories such as news, weather, stress, fam-

ily, and more. For more ﬂexibility, an open text ﬁeld

is available for additional input if none of the listed

reasons apply. These reasons were selected based

on aforementioned known inﬂuences on mental well-

being, such as physical health, environmental condi-

tions, and social relationships (Umberson and Mon-

tez, 2010). This feature allows users to track both

Mobile Data Collection for Depression Analysis: An App Framework for Monitoring Mood and Depression Using Smartphone and

Wearable Data

783

(a) (b) (c)

Figure 2: Image of the selectable moods and according sliders (a), according overview of chosen moods and energy level (b)

and the list of selectable inﬂuences on the current mood (c).

general and personal mood inﬂuences that may not be

captured through passive data collection. Users can

also indicate whether the selected reason had a posi-

tive (green highlighted) or negative impact (red high-

lighted) on their mood (cf. Figure 2c).

2.3 Exemplary Data Collection

2.3.1 Participants

Table 2: Overview of participants’ demographics.

User m f 18-23 24-30 31-40 51-60

22 15 7 4 10 2 6

A total of 22 individuals (15 male, 7 female) partici-

pated in the study to assess the functionality and us-

ability of the DAIsy DC app. Participants ranged in

age from 18 to 60 years, with a mean age of 34 ±

12 years. The demographic data of the participants

are listed in Table 2. The study was approved by the

OFFIS Ethics Board and participants gave their writ-

ten informed consent.

To participate, individuals were required to own

an Android-compatible smartphone with an operating

system version of at least Android 10. Individuals di-

agnosed with depression were excluded, as the study

focused on evaluating the app in a healthy population

to enable a comprehensive assessment of the app’s us-

ability and functionality prior to its use among users

with speciﬁc mental health needs.

During the six-week pilot study period, partic-

ipants were provided with a smartwatch (Withings

Steel HR) and/or a Withings sleep analyzer. Of the 22

participants, ten used the Withings sleep analyzer, six

received a watch and four preferred to use their own

smartwatch (Fitbit Charge 4, Fitbit Versa 4, Withings

Scan Watch). In six participants, there was an overlap

between the sleep analyzer and the smartwatch.

2.3.2 Data Collection

Active Data. At the start of the study, participants

completed a proﬁle to collect demographic informa-

tion, including their place of residence and, if appli-

cable, other frequently visited locations.

The participants’ primary task was to record their

current mood via the app at least three times a day.

In addition, individuals were asked to provide context

or reasons that inﬂuenced their mood for each entry.

Reminders to report their mood were sent twice a day.

In cases where devices, such as a smartwatch or Sleep

Analyzer, were not available, participants manually

logged their sleep duration and activity in Google Fit.

Additionally, participants were asked to complete the

PHQ-9 questionnaire weekly.

Passive Data. Data were continuously collected,

with most data passively gathered in the background,

requiring no active input from participants (cf. Ta-

ble 1).

Questionnaires. At the end of the study, partici-

pants were asked to provide feedback on the app’s

usability and their own technological readiness. Two

questionnaires were used for this purpose: the System

Usability Scale (SUS) (Brooke, 1995) measured the

app’s usability, while the Technology Commitment

(TC) short scale (Neyer et al., 2016) assessed the par-

ticipants’ willingness to use technology.

HEALTHINF 2025 - 18th International Conference on Health Informatics

784

3 DATA EXPLORATION

3.1 Active Data

3.1.1 PHQ-9

The PHQ-9 questionnaire was completed an average

of 4.45 ± 1.68 times. The minimum number of com-

pleted questionnaires was 1, the maximum 6. The av-

erage score was 5.9 for women and 7.3 for men, both

of which were in the non-depressive ranges (healthy

to inconspicuous 0 - 9). Most participants scored

within a range indicating a generally healthy mental

state. However, three participants recorded scores in-

dicating mild or moderate depression (mild depres-

sion: 10 - 14, moderate depression: 15-19).

3.1.2 Mood Tracking and Subjective Mood

Reasons

Table 3: Submission of the moods for each participant and

across all participants per day, week and the entire study

period.

User Day Week Study Period

1 3.88±1.64 26.57±8.9 186

2 1.88±0.78 10.33±5.75 62

3 2.35±1.58 10±6.48 40

4 2.4±1.2 16.43±6.72 115

5 1.93±0.88 11.86±3.48 83

6 1.38±0.5 4.71±2.63 33

7 1.22±0.74 3.5±2 28

8 2.71±0.97 18.57±2.99 130

9 2±0.8 12.86±2.91 90

10 2.93±0.65 19.29±6.42 135

11 3.55±1.1 13.5±7.18 156

12 2.57±1.33 12±8.1 108

13 1.95±0.43 18.71±4.47 84

14 3.12±1.25 14.14±8.9 131

15 2.36±1.05 5.6±6.44 99

16 1.27±0.55 11±2.4 28

17 2±0.79 22.29±2.97 66

18 2.03±0.82 10.83±4.92 65

19 2.59±0.76 13.71±8.26 96

All 2.32±0.72 13.47±5.89 91.34±44.36

Total 34.7±7.86 216.88±91.52 1735

Table 3 displays mood data across daily, weekly

and entire study period. The total number of submis-

sions was 1735 across all participants. Participants re-

ported their mood averagely 2.32 ± 0.72 times a day.

Although six participants completed the study ear-

lier than planned, three of them provided continuous

data over a period of at least three weeks. After each

mood submission, participants were asked to indicate

factors that positively and negatively inﬂuenced their

emotional state. Figure 3 illustrates the frequency of

each factor. Although optional, participants provided

reasons for 97% of the mood entries. Stress and sleep

Figure 3: Presentation of the frequencies with which a spe-

ciﬁc reason was selected either positively or negatively.

were the most frequently cited negative inﬂuences. In

contrast, family, friends, partners, and relaxation were

the most commonly recorded positive inﬂuences.

3.2 Passive Data

The results of passive data collection can be summa-

rized as follows:

• Step Count: The participants took an average of

4,450 ± 2,504 steps per day, with individual steps

ranging from 2,009 to 11,938 daily steps. The im-

pact of physical activity on mood was individual,

with some participants reporting mood improve-

ments on days with higher step counts. No partici-

pant showed a negative association between phys-

ical activity and mood.

• Sleep Data: Average sleep duration across all

subjects was 7.6 ± 0.8 hours of sleep per day,

which meets recommended guidelines for healthy

adults (Irwin, 2015). Although sleep disturbances

were rarely reported, some technical issues, such

as loss of connection from the sleep analyzer or

sleeping in another bed, resulted in missing data.

• Number of Visited Unspeciﬁed Locations: On

average, participants spent 14.5 ± 2.9 hours at

home daily, including sleeping time. During the

study period, participants visited an averaged to

12 ± 6 different locations, apart from the time

spent at home or at work.

• Weather Conditions: The weather’s inﬂuence on

varied individually. Most experienced mood im-

provements on clear and snowy days, while rainy

or cloudy conditions tended to have negative ef-

fects. However, some participants reported better

moods on rainy days. Additionally, weather was

included in the list of selectable reasons for the

current mood. Only three participants did not cite

that weather had an impact on their mood. Due to

their variability and the short study period, the bi-

ological weather data did not provide much infor-

Mobile Data Collection for Depression Analysis: An App Framework for Monitoring Mood and Depression Using Smartphone and

Wearable Data

785

mation and needs to be reevaluated after a longer

time period.

• Screen Time and Social Media Use: Daily

smartphone use averaged approximately 3 hours,

with a range of 0.3 hours to 6.5 hours. Time spent

on social media varied from 0.25 to 1.75 hours

per day. The effects of social media use on mood

were highly individual.

3.3 Usability and Technology

Commitment Questionnaire

Following the study, participants provided feedback

on the design and usability of the DAIsy DC app.

The evaluation of the SUS resulted in an average

score of 89.4 ± 3.5 out of 100, which corresponds to a

“very good” usability rating. The average technology

commitment score was 46.4 ± 5.9 out of 60, indicat-

ing that the sample consisted of tech-savvy partici-

pants.

A common point of feedback was the desire to ex-

pand the list of selectable mood inﬂuences. Five par-

ticipants suggested adding ”Trafﬁc” or ”Public Trans-

port” as these were common causes of frustration or

anger. Participants also requested an ”Outdoors” or

”Walking” category. Additionally, some participants

suggested adding ”Myself” to reﬂect self-centered

negative moods that often stem from personal dissat-

isfaction. Other desired additions included ”Procras-

tination” or ”Unproductivity”, with one participant

recommending ”Sexuality” as an option. One par-

ticipant recommended distinguishing speciﬁc work-

related inﬂuences (colleagues, bureaucracy, home of-

ﬁce).

Participants also suggested extending the list of

predeﬁned locations. Suggestions included adding

frequently visited locations such as family or friends’

homes, to the existing options.

3.4 Discussion

We have developed a ﬁrst version of a mental health

monitoring app that could, in the future, help with

both early detection of depression and monitoring re-

lapses. We evaluated the app’s usability and function-

ality in a six-week study with N = 22 healthy par-

ticipants. The aim of the app is to collect active in-

puts such as mood and passive data that could later

be used to identify connections between mood and

external inﬂuences. The results showed that the app

is suitable for a long-term study and users generally

found it easy to navigate. However, user feedback

also highlighted areas for improvement. Speciﬁcally,

participants suggested adding more options for mood

reasons and additional location options. Key areas for

further improvement include:

• Differentiation of Time Spent at Home: The

current GPS-based tracking system does not dif-

ferentiate between work and leisure activities at

home. Introducing a calendar or user input could

be helpful, but may increase user effort.

• Addition of Opposing Moods: The mood track-

ing feature could be reﬁned by allowing opposing

mood categories (e.g., happy - sad, nervous - re-

laxed) to provide deeper insights. In addition, a

stress level query should be incorporated.

• Gamiﬁcation Methods: Taylor et al. found that

incorporating gamiﬁcation signiﬁcantly improved

participation in a study (Taylor et al., 2018). How-

ever, these features should be carefully tailored

to the audience to ensure they increase engage-

ment without overwhelming users or imposing ad-

ditional requirements.

• User Interface (UI) and User Experience (UX):

Further reﬁning the UI and UX will likely im-

prove navigation and create a more cohesive and

visually appealing experience. Improving data

visualization would provide clearer insights into

mood trends over time. Additionally, all data en-

tries, including mood reasons, could be organized

and displayed in a calendar format.

• Personalized Reminders: The timing of the re-

minders to report on the current mood could be

adjusted to suit users’ individual routines (e.g.,

wake-up and sleep times) or triggered when their

smartphone is in use (Nahum-Shani et al., 2016).

• Additional Features: Future app iterations could

include additional features known to support men-

tal health such as mood and depression detection

through voice recognition (Schuller, 2018), jour-

naling (Baikie and Wilhelm, 2005) and to-do lists

to help establish routines. Personalized recom-

mendations triggered by signiﬁcant mood swings

as well as AI-driven mood trend predictions could

further tailor user support and intervention timing

(Klein et al., 2023).

Limitations of the study include a relatively small

sample size (N = 22), which limits the generalizabil-

ity of the results, although it provided valuable in-

sights at the individual-level. Future research with

larger, more diverse samples, including participants

diagnosed with depression, would allow for a more

comprehensive assessment of the app’s effectiveness.

A broader dataset could also enable users segmenta-

tion based on behavioral patterns such as activity lev-

els or weather sensitivity. Reliance on self-reported

data for mood and behavior metrics may under- or

over-report their experiences. Additionally, the six-

week study duration limits the ability to observe long-

term trends and effects that may develop over longer

periods of time. It would also be useful to link the

HEALTHINF 2025 - 18th International Conference on Health Informatics

786

stated reasons for the current mood with the passively

collected data. Furthermore, technical challenges as-

sociated with data collection by wearables and smart-

phones may have affected data reliability. Addressing

these limitations in future studies will be critical to

reﬁning and improving the functionality and general-

izability of the app.

The integration of DAIsy DC into the DAIsy

project’s virtual therapy assistant (VTA) offers the po-

tential to signiﬁcantly expand its capabilities in the

future. As described in Klein et al. (Klein et al.,

2023), DAIsy’s framework will enable personalized,

AI-driven therapeutic support with real-time recom-

mendations. This integration would enhance the func-

tionality of the VTA, by providing users with tai-

lored therapeutic suggestions and feedback and en-

abling real-time interventions based on passive data

(e.g., activity levels, location) and active inputs (e.g.,

mood reporting). Additionally, DAIsy VTA includ-

ing DAIsy DC could facilitate tracking user progress,

offer CBT exercises, and provide to support mental

well-being, creating a more personalized, dynamic

and responsive tool for treating depression.

4 CONCLUSION

In this study, we developed and tested DAIsy DC, a

mental health monitoring app that collects both ac-

tive and passive data to support early detection of de-

pression in the future. Our evaluation with healthy

individuals showed that the app is suitable for long-

term use and that overall navigation is easy for users.

However, the study identiﬁed key areas for improve-

ment, such as expanding the selection of mood rea-

sons, modifying the mood submission process, and

potential for personalized reminders and gamiﬁcation

features to increase engagement.

Future iterations of the app should focus on UI

and UX to improve data visualization and usability.

Additionally, expanding functionality to include per-

sonalized interventions, voice recognition, and mood

predictions could further support mental health man-

agement. Integration into the broader DAIsy project

framework offers the potential for tailored, real-time

therapeutic support, leveraging AI-driven recommen-

dations and user insights to create a responsive and

preventive mental health tool.

ACKNOWLEDGEMENTS

The project DAIsy - Developing AI ecosystems im-

proving diagnosis and care of mental diseases is

funded by the German Federal Ministry of Education

and Research (reference: 01IS22085E). Furthermore,

we would like to thank Niclas Braun, Annika Wiebe,

Benjamin Selaskowski and Matthias Guth from the

University Hospital Bonn for their valuable support,

insightful input, and guidance in navigating challeng-

ing situations throughout the study. Additionally, we

would like to express our gratitude to all participants

of this study for their time, commitment, and valuable

feedback.

REFERENCES

Asare, K. O., Terhorst, Y., Vega, J., Peltonen, E., Lager-

spetz, E., and Ferreira, D. (2021). Predicting depres-

sion from smartphone behavioral markers using ma-

chine learning methods, hyperparameter optimization,

and feature importance analysis: Exploratory study.

JMIR mhealth and uhealth, 9:e26540.

Baikie, K. and Wilhelm, K. (2005). Emotional and physical

beneﬁts of expressive writing. Advances in Psychi-

atric Treatment, 11:338–346.

Bakker, D. and Rickard, N. (2017). Engagement in mobile

phone app for self-monitoring of emotional wellbeing

predicts changes in mental health: Moodprism. Jour-

nal of Affective Disorders, 227:65–72.

Bengtsson, S. and Johansson, S. (2022). The meanings

of social media use in everyday life: Filling empty

slots, everyday transformations, and mood manage-

ment. Social Media + Society, 8.

Brooke, J. (1995). Sus: A quick and dirty usability scale.

In Usability Evaluation in Industry, page 189.

Cunningham, S., Hudson, C. C., and Harkness, K. L.

(2021). Social media and depression symptoms: a

meta-analysis. Research on Child and Adolescent Psy-

chopathology, 49:241–253.

Denissen, J., Butalid, L., Penke, L., and Aken, M. (2008).

The effects of weather on daily mood: A multilevel

approach. Emotion (Washington, D.C.), 8:662–667.

Deutscher Bundestag (2022). Wartezeiten auf eine

psychotherapie: Studien und umfragen. Wis-

senschaftliche Dienste, 36:987–1007.

Deutscher Wetterdienst (2024). Deutscher Wetterdienst:

National Meteorological Service of Germany. https:

//www.dwd.de. Accessed: 2024-10-08.

Drouin, M. and Abbasi, I. (2019). Neuroticism and face-

book addiction: How social media can affect mood?

The American Journal of Family Therapy, 47:1–17.

Fardouly, J., Diedrichs, P. C., Vartanian, L. R., and Halli-

well, E. (2015). Social comparisons on social media:

The impact of facebook on young women’s body im-

age concerns and mood. Body Image, 13:38–45.

Fitzpatrick, K. K., Darcy, A., and Vierhile, M. (2017). De-

livering cognitive behavior therapy to young adults

with symptoms of depression and anxiety using a fully

automated conversational agent (woebot): A random-

ized controlled trial. JMIR Mental Health, 4:e19.

Ghosh, S., Sahu, S., Ganguly, N., Mitra, B., and De, P.

(2019). Emokey: An emotion-aware smartphone key-

board for mental health monitoring. In 2019 Inter-

national Conference on COMmunication Systems &

NETworkS (COMSNETS), pages 496–499.

Mobile Data Collection for Depression Analysis: An App Framework for Monitoring Mood and Depression Using Smartphone and

Wearable Data

787

Grundstr

om, J., Konttinen, H., Berg, N., and Kiviruusu, O.

(2021). Associations between relationship status and

mental well-being in different life phases from young

to middle adulthood. SSM - Population Health, 14.

Hong, J., Kim, J., Kim, S., Oh, J., Lee, D., Lee, S., Uh,

J., Yoon, J., and Choi, Y. (2022). Depressive symp-

toms feature-based machine learning approach to pre-

dicting depression using smartphone. Healthcare,

10(7):1189.

Irwin, M. R. (2015). Why sleep is important for health: A

psychoneuroimmunology perspective. Annual Review

of Psychology, 66(1):143–172.

Javaid, S., Khan, M., and Daniyal, M. (2022). The re-

lationship between cellphone usage on the physical

and mental wellbeing of university students: A cross-

sectional study. International Journal of Environmen-

tal Research and Public Health, 19.

Kazantzis, N., Luong, H., Usatoff, A., Impala, T., Yew, R.,

and Hofmann, S. (2018). The processes of cognitive

behavioral therapy: A review of meta-analyses. Cog-

nitive Therapy and Research, 42.

Kessler, R., Berglund, P., Bruce, M., Koch, J., Laska, E.,

Leaf, P., Manderscheid, R., Rosenheck, R., Walters,

E., and Wang, P. (2002). The prevalence and correlates

of untreated serious mental illness. Health services

research, 36:987–1007.

Klein, F., Aschwege, F. M.-V., Elfert, P., R

aker, J.,

Philipsen, A., Braun, N., Selaskowski, B., Wiebe, A.,

Guth, M., Spallek, J., Seuss, S., Storey, B., Geppert,

L. N., L

uck, I., and Hein, A. (2023). Developing ad-

vanced ai ecosystems to enhance diagnosis and care

for patients with depression. Studies in Health Tech-

nology and Informatics, 309:18–22.

Klimstra, T. A., Frijns, T., Keijsers, L., Denissen, J. J. A.,

Raaijmakers, Q. A. W., van Aken, M. A. G., Koot,

H. M., van Lier, P. A. C., and Meeus, W. H. J. (2011).

Come rain or come shine: individual differences in

how weather affects mood. Emotion, 11(6):1495–

1499.

Kroenke, K., Spitzer, R. L., and Williams, J. B. (2001). The

phq-9. Journal of General Internal Medicine, 16:606–

613.

Kvam, S., Kleppe, C. L., Nordhus, I. H., and Hovland,

A. (2016). Exercise as a treatment for depression:

A meta-analysis. Journal of Affective Disorders,

202:67–86.

Masud, M., Mamun, M., Thappa, K., Lee, D., Grifﬁths,

M., and Yang, S.-H. (2020). Unobtrusive monitoring

of behavior and movement patterns to detect clinical

depression severity level via smartphone. Journal of

Biomedical Informatics, 103:103371.

MindDoc (2023). Mental health support and therapy online.

https://minddoc.com/de/de/. Accessed: 2024-10-08.

Mohr, D., Zhang, M., and Schueller, S. (2017). Personal

sensing: Understanding mental health using ubiqui-

tous sensors and machine learning. Annual review of

clinical psychology, 13.

Nahum-Shani, I., Smith, S., Spring, B., Collins, L.,

Witkiewitz, K., Tewari, A., and Murphy, S. (2016).

Just-in-time adaptive interventions (jitais) in mobile

health: Key components and design principles for on-

going health behavior support. Annals of Behavioral

Medicine, 52.

Nationale Versorgungsleitlinien (2024). Nationale ver-

sorgungsleitlinie unipolare depression - langfas-

sung. https://www.leitlinien.de/themen/depression/

version-3. Accessed: 2024-03-05.

Neyer, F. J., Felber, J., and Gebhardt, C. (2016). Kurzskala

technikbereitschaft (tb, technology commitment).

Zusammenstellung sozialwissenschaftlicher Items

und Skalen (ZIS). DOI: 10.6102/zis244.

Posner, J., Russell, J., and Peterson, B. (2005). The circum-

plex model of affect: An integrative approach to af-

fective neuroscience, cognitive development, and psy-

chopathology. Development and Psychopathology,

17:715–734.

Razavi, R., Gharipour, A., and Gharipour, M. (2020). De-

pression screening using mobile phone usage meta-

data: a machine learning approach. Journal of the

American Medical Informatics Association: JAMIA,

27.

Rykov, Y., Thach, T.-Q., Bojic, I., Christopoulos, G.,

and Car, J. (2021). Digital biomarkers for depres-

sion screening with wearable devices: Cross-sectional

study with machine learning modeling. JMIR mHealth

and uHealth, 9.

Schuller, B. (2018). Speech emotion recognition: Two

decades in a nutshell, benchmarks, and ongoing

trends. Communications of the ACM, 61(5):90–99.

Short, M. and Louca, M. (2015). Sleep deprivation leads to

mood deﬁcits in healthy adolescents. Sleep Medicine,

16.

Stanton, R. and Reaburn, P. (2014). Exercise and the treat-

ment of depression: A review of the exercise program

variables. Journal of Science and Medicine in Sport,

17(2):177–182.

Taniguchi, K., Takano, M., Tobari, Y., Hayano, M., Naka-

jima, S., Mimura, M., Tsubota, K., and Noda, Y.

(2022). Inﬂuence of external natural environment in-

cluding sunshine exposure on public mental health: A

systematic review. Psychiatry International.

Taylor, S., Ferguson, C., Peng, F., Schoeneich, M., and

Picard, R. (2018). Use of in-game rewards to moti-

vate daily self-report compliance (preprint). Journal

of Medical Internet Research, 21.

Thorisdottir, I. E., Sigurvinsdottir, R., Asgeirsdottir, B. B.,

Allegrante, J. P., and Sigfusdottir, I. D. (2019). Active

and passive social media use and symptoms of anxi-

ety and depressed mood among icelandic adolescents.

Cyberpsychology, behavior and social networking.

Torous, J., Larsen, M., Depp, C., Cosco, T., Barnett, I.,

Nock, M., and Firth, J. (2018). Smartphones, sensors,

and machine learning to advance real-time prediction

and interventions for suicide prevention: a review of

current progress and next steps. Current Psychiatry

Reports, 20:51.

Umberson, D. and Montez, J. (2010). Social relationships

and health a ﬂashpoint for health policy. Journal of

health and social behavior, 51 Suppl:S54–66.

WHO (2023). World health organization - depres-

sion. https://www.who.int/news-room/fact-sheets/

detail/depression. Accessed: 2024-09-16.

HEALTHINF 2025 - 18th International Conference on Health Informatics

788