Developing a Process for the Analysis of User Journeys and the Prediction of Dropout in Digital Health Interventions: Machine Learning Approach (Preprint)
Open accessThe DOI matches its Crossref record; 47 of the 47 references it cites were found in a registry or an index.
Abstract
BACKGROUND: User dropout is a widespread concern in the delivery and evaluation of digital (ie, web and mobile apps) health interventions. Researchers have yet to fully realize the potential of the large amount of data generated by these technology-based programs. Of particular interest is the ability to predict who will drop out of an intervention. This may be possible through the analysis of user journey data—self-reported as well as system-generated data—produced by the path (or journey) an individual takes to navigate through a digital health intervention. OBJECTIVE: The purpose of this study is to provide a step-by-step process for the analysis of user journey data and eventually to predict dropout in the context of digital health interventions. The process is applied to data from an internet-based intervention for insomnia as a way to illustrate its use. The completion of the program is contingent upon completing 7 sequential cores, which include an initial tutorial core. Dropout is defined as not completing the seventh core. METHODS: Steps of user journey analysis, including data transformation, feature engineering, and statistical model analysis and evaluation, are presented. Dropouts were predicted based on data from 151 participants from a fully automated web-based program (Sleep Healthy Using the Internet) that delivers cognitive behavioral therapy for insomnia. Logistic regression with L1 and L2 regularization, support vector machines, and boosted decision trees were used and evaluated based on their predictive performance. Relevant features from the data are reported that predict user dropout. RESULTS: Accuracy of predicting dropout (area under the curve [AUC] values) varied depending on the program core and the machine learning technique. After model evaluation, boosted decision trees achieved AUC values ranging between 0.6 and 0.9. Additional handcrafted features, including time to complete certain steps of the intervention, time to get out of bed, and days since the last interaction with the system, contributed to the prediction performance. CONCLUSIONS: The results support the feasibility and potential of analyzing user journey data to predict dropout. Theory-driven handcrafted features increased the prediction performance. The ability to predict dropout at an individual level could be used to enhance decision making for researchers and clinicians as well as inform dynamic intervention regimens.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Figures
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Bremer et al. (2020) focused on the analysis of user journeys and identified that time spent in earlier stages of the program, morning wake-up times (either before 4:30 am or later than 6:45 am), time to get out of bed (less than 9 minutes or more than 66 minutes), a greater wake after sleep onset, logging triggers for 18 days or more, receiving emails for 30 days or more, and no interaction with”
From Examining the role of AI technology in online mental healthcare: opportunities, challenges, and implications, a mixed-methods review (2024) · cites this paper as [83]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Additionally, the studies reporting on the use of ML algorithms found significant success in outcome prediction and feature detection to support online therapy delivery (6, 19, 28, 29, 31, 33, 83, 85, 86, 92–94).”
From Examining the role of AI technology in online mental healthcare: opportunities, challenges, and implications, a mixed-methods review (2024) · cites this paper as [83]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“As such, the relevant studies highlighted the importance of analyzing the large datasets produced by online mental healthcare interventions and the efficiency and accuracy of AI algorithms which in some cases were comparable to or better than human assessors (6, 19, 28, 29, 31, 33, 83, 85, 86, 92–94).”
From Examining the role of AI technology in online mental healthcare: opportunities, challenges, and implications, a mixed-methods review (2024) · cites this paper as [83]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“In terms of allegiance bias, because of the scarcity and novelty of AI-augmented interventions and other AI algorithms, most of the included articles developed and studied their own AI tools, instead of AI tools developed by other researchers (29, 33, 73, 74, 76, 79, 80, 82–87, 90).”
From Examining the role of AI technology in online mental healthcare: opportunities, challenges, and implications, a mixed-methods review (2024) · cites this paper as [83]
The cited source says what the claim attributes to it.
In the citing paper
“Eight studies reported on the ethnicity of the participants (30, 32, 41, 74, 76, 77, 80, 83), and in most of the studies between 31% and 91% of the participants identified as White.”
From Examining the role of AI technology in online mental healthcare: opportunities, challenges, and implications, a mixed-methods review (2024) · cites this paper as [83]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Regression shrinkage and selection via the lasso
Tibshirani, R. (1996) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/j.2517-6161.1996.tb02080.x · Cited by 51,482 · cited 1×
“…when utilizing a large number of features [37] and some types of statistical procedures (eg, linear regression) cannot be applied when the number of features is greater than the number of observations, alternative techniques such as regularization and feature selection may need to be used [46].”
BackgroundAn introduction to ROC analysis
Fawcett, T. (2006) · Pattern Recognition Letters · doi:10.1016/j.patrec.2005.10.010 · Cited by 21,879 · cited 1×
“More information about these validation procedures and their application can be found elsewhere [45].”
Chai, T., & Draxler, R. R. (2014) · Geoscientific Model Development · doi:10.5194/gmd-7-1247-2014 · Cited by 8,761
Willmott, C., & Matsuura, K. (2005) · Climate Research · doi:10.3354/cr030079 · Cited by 7,958
Supervised Machine Learning: A Review of Classification Techniques
Kotsiantis, S. B. (2007) · Informatica · Cited by 6,007
A few useful things to know about machine learning
Domingos, P. (2012) · Communications of the ACM · doi:10.1145/2347736.2347755 · Cited by 3,326 · cited 3×
“Handcrafted feature engineering often involves a trial and error phase to experiment with different features [37].”
BackgroundInfluentialEysenbach, G. (2005) · Journal of Medical Internet Research · doi:10.2196/jmir.7.1.e11 · Cited by 2,524 · cited 2×
“Although researchers have highlighted the need for a science of user attrition [13], there have been few advances in predicting dropout through advanced quantitative approaches in eHealth interventions [14].”
BackgroundThe consensus sleep diary: standardizing prospective sleep self-monitoring
Carney, C. E., Buysse, D. J., Ancoli-Israel, S., Edinger, J. D., Krystal, A. D., Lichstein, K. L., & Morin, C. M. (2012) · Sleep · doi:10.5665/sleep.1642 · Cited by 2,172 · cited 1×
“Sleep diary questions mirrored those from the consensus sleep diary [49].”
MethodCarlbring, P., Andersson, G., Cuijpers, P., Riper, H., & Hedman-Lagerlöf, E. (2018) · Cognitive Behaviour Therapy · doi:10.1080/16506073.2017.1401115 · Cited by 1,232 · cited 1×
“The efficacy of digital (ie, internet, web, and mobile) behavioral interventions to improve a range of health-related outcomes has been well documented [1-3].”
BackgroundAn analysis of four missing data treatment methods for supervised learning
Batista, G. E. A. P. A., & Monard, M. C. (2003) · Applied Artificial Intelligence · doi:10.1080/713827181 · Cited by 895
Cited by
Hornstein, S., Zantvoort, K., Lueken, U., Funk, B., & Hilbert, K. (2023) · Frontiers in Digital Health · doi:10.3389/fdgth.2023.1170002 · Cited by 134
Self-citationZantvoort, K., Nacke, B., Görlich, D., Hornstein, S., Jacobi, C., & Funk, B. (2024) · npj Digital Medicine · doi:10.1038/s41746-024-01360-w · Cited by 116
Self-citationNaegelin, M., Weibel, R. P., Kerr, J. I., Schinazi, V. R., Marca, R. L., von Wangenheim, F., Hölscher, C., & Ferrario, A. (2023) · Journal of Biomedical Informatics · doi:10.1016/j.jbi.2023.104299 · Cited by 117
Gutierrez, G., Stephenson, C., Eadie, J., Asadpour, K., & Alavi, N. (2024) · Frontiers in Psychiatry · doi:10.3389/fpsyt.2024.1356773 · Cited by 71
Enhancing health care through medical cognitive virtual agents
Mishra, S., Chaudhury, P., Tripathy, H. K., Sahoo, K. S., Jhanjhi, N. Z., Elnour, A. A. H., & Abdelmaboud, A. (2024) · Digital Health · doi:10.1177/20552076241256732 · Cited by 50
“51 The architecture of the healthcare agents (HCA) is shown in Figure 6.”
BackgroundMachine and cognitive intelligence for human health: systematic review
Chen, X., Cheng, G., Wang, F. L., Tao, X., Xie, H., & Xu, L. (2022) · Brain Informatics · doi:10.1186/s40708-022-00153-9 · Cited by 34
“A step-by-step analysis [64] indicated the feasibility of user journey data analysis in varied machine learning models to predict dropout in digital health interventions.”
BackgroundMethodResultMoshe, I., Terhorst, Y., Paganini, S., Schlicker, S., Pulkki-Råback, L., Baumeister, H., Sander, L. B., & Ebert, D. D. (2022) · Journal of Medical Internet Research · doi:10.2196/38261 · Cited by 28
“In a study of a self-help digital intervention for the treatment of insomnia, Bremer et al [46] identified individuals at risk of dropout with an area under the receiver operating curve (AUC) of 0.719 using a combination of baseline characteristics (eg, self-reported stress levels) and intervention…”
InfluentialLinardon, J., Fuller‐Tyszkiewicz, M., Shatte, A. B. R., & Greenwood, C. J. (2022) · International Journal of Eating Disorders · doi:10.1002/eat.23733 · Cited by 26
“Accurate ML prediction models for program engagement have been observed in online mindfulness programs for wellbeing (Lekkas et al., 2021), internet-based cognitive therapy for depression (Chien et al., 2020), and web-based behavioral programs for insomnia (Bremer et al., 2020).”
MethodEberle, J. W., Daniel, K. E., Baee, S., Silverman, A. L., Lewis, E., Baglione, A. N., Werntz, A. J., French, N. J., Ji, J. L., Hohensee, N., et al. (2024) · Journal of Consulting and Clinical Psychology · doi:10.1037/ccp0000896 · Cited by 25
Bricker, J., Miao, Z., Mull, K., Santiago-Torres, M., & Vock, D. M. (2023) · Journal of Medical Internet Research · doi:10.2196/43629 · Cited by 22
“…objective measures of intervention engagement (eg, number of log-ins and proportion of content completed); and even sophisticated variables of user journeys, which encompass the myriad of sequences of interactions that a user takes to navigate through a digital intervention program [21].”
Zantvoort, K., Isacsson, N. H., Funk, B., & Kaldo, V. (2024) · Digital Health · doi:10.1177/20552076241248920 · Cited by 18
“Most papers predicting dropout focus on a single intervention, 6,16,17,20,21 while one gathers two different interventions for the same symptoms, 19 and one gathers interventions for three different symptoms.”
BackgroundMethodInfluentialSelf-citationMitigating Issues With/of/for True Personalization
B, H. O.-K. A., Pohjolainen, S., & Agyei, E. E. Y. F. (2022) · Frontiers in Artificial Intelligence · doi:10.3389/frai.2022.844817 · Cited by 23
“It is a common characteristic of digital health interventions to suffer from high dropout rates (Bremer et al., 2020).”
BackgroundMetrics for Peer Counseling: Triangulating Success Outcomes for Online Therapy Platforms
Wang, T., Shah, H. K., Shah, R. S., Wang, Y.-C., Kraut, R. E., & Yang, D. (2023) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3544548.3581372 · Cited by 24
“To increase retention, platforms can leverage user journeys that capture common pathways to motivate users to continue participating in the community when they are most likely to drop off the platform [16].”
BackgroundZantvoort, K., Scharfenberger, J., Boß, L., Lehr, D., & Funk, B. (2023) · Journal of Healthcare Informatics Research · doi:10.1007/s41666-023-00148-z · Cited by 21
Self-citationMukhiya, S. K., Lamo, Y., & Rabbi, F. (2022) · JMIR Human Factors · doi:10.2196/31029 · Cited by 14
“For example, Bremer et al [33] outlined the use of ML techniques to predict dropout in insomnia interventions.”
MethodTerceiro, L., Mustafa, M. I., Hägglund, M., & Kharko, A. (2024) · JMIR Research Protocols · doi:10.2196/65099 · Cited by 13
“Intervention dropout from internet-based treatment for psychological disorders, for example, fluctuates between 30% and 50% [24].”
Simon, L., Steinmetz, L., Bendig, E., Küchler, A., Riemann, D. W., Ebert, D. D., Spiegelhalder, K., & Baumeister, H. (2025) · International Journal of Clinical and Health Psychology · doi:10.1016/j.ijchp.2025.100598 · Cited by 11
“Some evidence exists that both in insomnia (Bremer et al. 2020) and in other mental health areas, as in depression (Moshe et al. 2022), the time needed to complete the first module is associated with dropout.”
Remegio, F. M. C. (2024) · International Conference on Educational and Information Technology · doi:10.1109/iceit61397.2024.10541035 · Cited by 9
“Studies by [17]- [21] highlight a gap in region-specific studies, especially in developing countries.”
BackgroundPredicting Adherence to Behavior Change Support Systems Using Machine Learning: Systematic Review
Ekpezu, A. O., Wiafe, I., & B, H. O.-K. A. (2023) · JMIR AI · doi:10.2196/46779 · Cited by 7
“…Accuracy Random forest a [32] Dropout or nondropout AUROC a , TPR, FPR, and PRAUC j Logistic regression, SVM i , and decision trees (boosted) a [35] Relapse or not Accuracy, sensitivity a , specificity, and AUROC a Ensemble methods a [33] Adherence or nonadherence Accuracy, specificity a ,…”
BackgroundMethodInfluentialKötting, L., Derksen, C., Keller, F. M., & Lippke, S. (2023) · JMIR Pediatrics and Parenting · doi:10.2196/44701 · Cited by 7
“In addition, there was a comparably high dropout rate in IG2 that was potentially selective, which is typical for asynchronous web-based interventions [65,67,68].”
BackgroundRamesh, S. H., Jull, D., Fournier, H., & Rajabiyazdi, F. (2025) · Interactive Journal of Medical Research · doi:10.2196/66164 · Cited by 7
“For instance, predictive algorithms have been used to predict dropout rates in health care programs [57,58], such as cognitive behavioral therapy for insomnia [59], chronic disease management programs [60], rehabilitation programs [61-63], and for selecting appropriate treatment plans [64,65].”
Wang, P., Chen, H., Li, Z., Xu, W., Chang, Y., & Li, H. (2025) · Smart Health · doi:10.1016/j.smhl.2025.100565 · Cited by 5
Ryan, A. T., Brenner, L. A., Ulmer, C. S., Mackintosh, M., & Greene, C. J. (2023) · JMIR Formative Research · doi:10.2196/40104 · Cited by 5
Derksen, M. E., van Beek, M., Blankers, M., Nasri, H., de Bruijn, T., Lommerse, N. M., van Wingen, G. A., Pauws, S. C., & Goudriaan, A. E. (2024) · European Addiction Research · doi:10.1159/000543252 · Cited by 4
“Earlier research showed that machine-learning methods could be applied to fi nd predictors or correlates of eHealth intervention adherence, dropout, engagement and success [15 – 18].”
Eicher, S., Fenski, F., Behr, S., Hammelrath, L., Boettcher, J., Schaeuffele, C., & Knaevelsrud, C. (2026) · npj Digital Medicine · doi:10.1038/s41746-026-02546-0 · Cited by 3
Zantvoort, K., Matthiesen, J., Bjurner, P., Bendix, M., Brefeld, U., Funk, B., & Kaldo, V. (2025) · Internet Interventions · doi:10.1016/j.invent.2025.100828 · Cited by 1
“Standard data collection methods for questionnaire data yield no information about the user's decision process, such as indecisiveness, lack of understanding or attention manifesting in changes to an item (Bremer et al., 2020).”
InfluentialSelf-citationUso de Inteligência Artificial no Tratamento de Transtornos Mentais: Uma Revisão de Escopo
Ferreira, D. C., & Gomes, P. C. D. S. (2026) · Psicologia Argumento · doi:10.7213/psicolargum.43.123.ao14 · Cited by 1
Kronthaler, S. M., Kosman, E., Tissen‐Diabaté, T., Mühle, E., Weber-Schallauer, L., Pross, T., Karsten, M. M., Beier, K. M., & Hatzler, L. (2026) · Journal of Medical Internet Research · doi:10.2196/86042 · Cited by 2
“Early engagement predicts adherence in digital health programs [78,81,82] and may mark a critical window for retention strategies.”
Comparative Performance of Supervised Learning Algorithms in Predicting Student Dropout Risk
Wijayanti, E., Widowati, & Widodo, C. E. (2025) · Beyond Technology Summit on Informatics International Conference (BTS-I2C) · doi:10.1109/bts-i2c67944.2025.11399356 · Cited by 1
“The results support findings from previous studies[30][22] which highlight the superiority of SVM in educational predictive modeling The model’s strength lies in its ability to capture non-linear relationships and handle feature interactions efficiently.”
Sîrbu, V., Tomoiagă, C. T., & David, D. (2026) · International Journal of Human-Computer Interaction · doi:10.1080/10447318.2026.2641702 · Cited by 1
Sediva, H. (2024) · WestminsterResearch (University of Westminster) · doi:10.34737/ww762
Pohjolainen, S., & B, H. O.-K. A. (2026) · AIS Electronic Library (AISeL)
Kulasinghe, K., Savolainen, M. J., & B, H. O.-K. A. (2026) · AIS Electronic Library (AISeL)
Metadata
- Authors
- Bremer, V., Chow, P. I., Funk, B., Thorndike, F. P., & Ritterband, L. M.
- Affiliations
- Leuphana University of Lüneburg (DE)
- University of Virginia (US)
- Corresponding author
- Bremer, Vincent
- Venue
- Journal of Medical Internet Research
- Publisher
- JMIR Publications Inc.
- Published
- 2020-01-09
- Volume
- 22
- Issue
- 10
- Pages
- e17738-e17738
- ISSN
- 1438-8871 (ISSN-L)
- PMC
- PMC7657718
- OpenAlex
- W3097807896
- Cited by
- 29
- References
- 47
- Citations by year
- 2026: 3 · 2025: 5 · 2024: 8 · 2023: 7 · 2022: 6
- FWCI
- 2.94
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by
- Locations
- Journal of Medical Internet Research (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- Multilingual Matters (Channel View Publications) (publishedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- National Institute of Mental Health (R01MH086758)
- Topics
- Social Sciences › Psychology › Applied Psychology › Digital Mental Health Interventions
- Physical Sciences › Computer Science › Computer Science Applications › Online Learning and Analytics
- Health Sciences › Health Professions › Health Information Management › Artificial Intelligence in Healthcare
- Keywords
- digital health interventions, online intervention, insomnia, cognitive behavioral therapy for insomnia, handcrafted features, feature engineering, user engagement
- MeSH terms
- Machine Learning, Internet-Based Intervention, Adult, Aged, Female, Humans, Male, Middle Aged, Young Adult, Mobile Applications
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.
