Embodied Conversational Agents in Internet-Based Cognitive Behavioral Therapy for Depression: Bridging the Gap Between Unguided and Guided Interventions
Bibliographic record
Abstract
Guided internet-based cognitive therapy (iCBT) for depression is considered more effective than unguided iCBT, which may be explained by the higher adherence rates found for the guided format. Unguided iCBT, on the other hand, is potentially more scalable and accessible as it does not require human involvement. This thesis explores whether this gap between unguided and guided interventions, i.e., effectiveness versus scalability and accessibility, can be bridged using embodied conversational agents (ECAs). ECAs are more or less autonomous virtual characters with an embodiment used to communicate with a user, and may be used to provide automated iCBT support. With endless possible variations in ECA embodiments, ways of communicating, and reasoning capabilities, there is no clear solution to designing an ECA for iCBT support. The two research goals of this thesis are to (1) design and develop an ECA for iCBT support based on what is known to work and (2) establish whether such an ECA can increase adherence to unguided iCBT. Chapter 2 presents the results of a scoping review into the use of ECAs in the treatment of common mental health disorders. We followed up on the literature review by conducting two experimental studies to further inform the design of our ECA. In the first of these, described in Chapter 3, we investigated whether feedback from a minimalistic ECA could increase adherence. In the second experimental study, described in Chapter 4, we validated a sentiment analysis algorithm for the Dutch language against human judgment in an iCBT context. In the next two studies we looked more closely into the content of real-world iCBT feedback to inform the design of our ECA dialogues. Chapter 5 presents a study that aimed to identify which therapist behaviors occur in written online feedback, assess how well therapists adhere to feedback instructions, and explore whether either is associated with patient outcomes. Chapter 6 presents another study into the content of therapist feedback and adherence to feedback instructions. The research above comes together in the development of Moodbuster Lite, a 4-week iCBT intervention for low mood with integrated ECA support. Chapter 7 describes the protocol of a pilot two-armed randomized controlled trial (Moodbuster Lite with ECA support vs. Moodbuster Lite without ECA support) that evaluates the intervention, assessing the effectiveness of virtual coach support in terms of improved intervention adherence, as well as the feasibility of a future, larger-scale trial with Moodbuster Lite. The design of the ECA that was developed to function as a virtual coach in Moodbuser Lite, was largely informed by the research conducted in Chapters 2 to 6; it has a reasonable amount of technological sophistication (Chapter 3), applies automated sentiment analysis (Chapter 4), interacts according to iCBT feedback protocols (Chapters 5 and 6), and targets a clinically relevant outcome (Chapter 2), i.e., adherence. After the subsequent integration of the ECA with the Moodbuster Lite intervention, and having obtained formal ethical approval, Moodbuster Lite is now ready for evaluation in a pilot randomized controlled trial as described in Chapter 7. The intervention and study design show how a bridge between guided and unguided iCBT interventions was built in the form of Moodbuster Lite, and the pilot RCT will inform how the bridge can be further improved.
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.
How this classification was reachedexpand
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 | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| 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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".