Thematic Analysis on User Reviews for Depression and Anxiety Chatbot Apps: Machine Learning Approach
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
BACKGROUND: Anxiety and depression are among the most commonly prevalent mental health disorders worldwide. Chatbot apps can play an important role in relieving anxiety and depression. Users' reviews of chatbot apps are considered an important source of data for exploring users' opinions and satisfaction. OBJECTIVE: This study aims to explore users' opinions, satisfaction, and attitudes toward anxiety and depression chatbot apps by conducting a thematic analysis of users' reviews of 11 anxiety and depression chatbot apps collected from the Google Play Store and Apple App Store. In addition, we propose a workflow to provide a methodological approach for future analysis of app review comments. METHODS: We analyzed 205,581 user review comments from chatbots designed for users with anxiety and depression symptoms. Using scraper tools and Google Play Scraper and App Store Scraper Python libraries, we extracted the text and metadata. The reviews were divided into positive and negative meta-themes based on users' rating per review. We analyzed the reviews using word frequencies of bigrams and words in pairs. A topic modeling technique, latent Dirichlet allocation, was applied to identify topics in the reviews and analyzed to detect themes and subthemes. RESULTS: Thematic analysis was conducted on 5 topics for each sentimental set. Reviews were categorized as positive or negative. For positive reviews, the main themes were confidence and affirmation building, adequate analysis, and consultation, caring as a friend, and ease of use. For negative reviews, the results revealed the following themes: usability issues, update issues, privacy, and noncreative conversations. CONCLUSIONS: Using a machine learning approach, we were able to analyze ≥200,000 comments and categorize them into themes, allowing us to observe users' expectations effectively despite some negative factors. A methodological workflow is provided for the future analysis of review comments.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it