Exploring Customizable Interactive Tools for Therapeutic Homework Support in Mental Health Counseling
Authors
Paper Title
Exploring Customizable Interactive Tools for Therapeutic Homework Support in Mental Health Counseling
Publication Info
- Topic area: Therapist-centered AI tools for mental health counseling.
- Keywords: Therapeutic homework, mental health counseling, generative AI, therapist-facing tools, cognitive offloading, customization, clinical workflows, AI trust, human oversight, ethical AI.
Background and Problem
- Problem / challenge: Therapists face cognitive burdens in synthesizing fragmented client homework data, which is often scattered across formats like paper worksheets, verbal reports, and app-based submissions. Existing tools fail to integrate and summarize this data effectively.
- Significance: Efficiently tracking and interpreting therapeutic homework is critical for improving client outcomes, enabling therapists to focus on clinical insights rather than administrative tasks.
- Motivation and related work: Prior research has explored client-facing mHealth tools and therapist-facing practice management systems, but few have addressed therapist needs for synthesizing multi-format homework data. Generative AI has shown promise in adjacent tasks, such as progress note generation, but its application to therapist-facing homework review remains underexplored.
Solution
- Proposed approach: TheraTrack, a customizable therapist-facing tool that integrates multi-dimensional data and leverages large language models (LLMs) to generate traceable summaries and support natural-language queries.
- Novelty:
- Centralized integration of fragmented homework inputs into a single interactive dashboard.
- Customizable widgets tailored to therapists’ clinical goals and therapeutic modalities.
- GenAI-powered summaries and chat assistant for efficient synthesis and query-driven exploration of client data.
- Procedure and key techniques:
- Defining Needs: Therapists complete an onboarding survey to specify preferences for tracking homework and clinical assessments.
- Choosing Widgets: Therapists select relevant widgets, such as homework progress charts, GenAI summaries, and assessment trackers.
- Displaying Customized Dashboard: The dashboard integrates homework trends, assessment results, and optional biometric data, with GenAI features enabling traceable summaries and interactive queries.
Results
- Concrete findings:
- 79% of therapists found GenAI summaries reduced their workload.
- 64% rated the graphs and metrics as visually clear.
- 50% expressed strong enthusiasm for long-term integration.
- Advantage over baselines: TheraTrack reduced cognitive load, enabled verification of AI-generated insights, and provided a centralized view of client data, addressing fragmentation and inefficiencies in current workflows.
- Experiments / evaluation:
- Pilot study with 14 therapists using simulated client data.
- Methods included think-aloud exploration, semi-structured interviews, and Likert-scale surveys.
- Evaluated usability, perceived usefulness, trust, and customization.
- Limitations and future work:
- Small sample size may limit generalizability.
- Focused on therapist perspectives; client perspectives remain unexplored.
- Real-world adoption and long-term use may reveal additional challenges.
- Current design tailored to specific homework tracking scenarios; broader therapeutic contexts require adaptation.
Summary
TheraTrack is a therapist-facing AI tool designed to streamline the review and interpretation of therapeutic homework by integrating fragmented data into a centralized, customizable dashboard. Leveraging GenAI for summaries and queries, it reduces cognitive burden, enhances session preparation, and supports clinical sense-making. Pilot study results indicate high usability, perceived usefulness, and trust among therapists, with customization enabling flexible integration into diverse workflows. Future research should explore long-term adoption, client perspectives, and adaptation to broader therapeutic practices.
Research Questions / Practical Problems
Question signals indexed for this paper.
- 100%
Digitizing the Pre-consultation Experience: Impacts and Design Recommendations
CHI '26· Human-LLM Collaboration +2
- 100%
Towards Better Health Conversations: The Benefits of Context-seeking
CHI '26· Human-LLM Collaboration +2
- 71%
MindfulDiary: Harnessing Large Language Model to Support Psychiatric Patients' Journaling
CHI '24· Human-LLM Collaboration +2
- 71%
Toward Flexible Psychiatric History-Taking and Visualization: Exploring Clinician Perspectives with Large Language Models
CHI '26· Human-LLM Collaboration +2
- 71%
More than Decision Support: Exploring Patients' Longitudinal Usage of Large Language Models in Real-World Healthcare Settings
CHI '26· Human-LLM Collaboration +2
- 71%
Who Does What? Archetypes of Roles Assigned to LLMs During Human-AI Decision-Making
CHI '26· Human-LLM Collaboration +2
- 71%
Exploring the Future of AI in Clinical Collaboration: A Study on Tumor Board Case Preparation
CHI '26· Human-LLM Collaboration +3
- 67%
Limitations of the LLM-as-a-Judge Approach for Evaluating LLM Outputs in Expert Knowledge Tasks
IUI '25· Human-LLM Collaboration +1
- 63%
High Accuracy and Hidden Disparities: Investigating Foundation Model Performance in Clinical Cognitive Assessment
CHI '26· Explainable AI (XAI) +3
Based on Jaccard similarity of research subtopics & professions (≥60%)