TheraQuest: A Gamified, LLM-Powered Simulation for Massage Therapy Training

TL;DR

TheraQuest tackles the gap in massage therapy education related to patient communication and diverse presentations by delivering a web-based gamified simulation powered by large language models to generate diverse virtual patients. The platform supports interactive dialogue, diagnostic decision-making, and immediate assessment within a customizable therapy room, aided by real-time skill metrics and a global ranking system. The authors detail a C# .NET Core prototype, web-based interfaces, and AI-driven patient generation, and propose a two-stage evaluation plan including expert review and in-person user studies against traditional methods. If validated, TheraQuest could offer a cost-effective, accessible training alternative that bridges theoretical knowledge and clinical proficiency, with potential extensions to physical therapy and rehabilitation domains.

Abstract

Massage therapy training emphasizes hands-on techniques and effective therapist--patient communication. However, many educational programs struggle to provide realistic practice scenarios. To address this problem, we propose TheraQuest, a gamified, web-based simulation platform that employs large language models (LLMs) to generate diverse virtual patients with varying symptoms and cultural backgrounds. Through interactive dialogue, anatomical decision-making, and immediate assessment, trainees develop both diagnostic reasoning and empathetic communication skills in a low-risk environment. Unlike exclusively VR-based solutions, TheraQuest remains accessible via standard web browsers, mitigating the cost and discomfort associated with extended headset use. Preliminary testing suggests that integrating LLM-driven virtual patients with real-time skill metrics can enhance trainee engagement and help bridge the gap between theoretical knowledge and clinical proficiency.

Figures (7)

• Figure 1: Flowchart of the TheraQuest Prototype System Design (Version: 02/14/2025). trainees begin by registering or logging in [Start], then proceed to the [Main Menu] to either review their [User Profile], [Global Rankings] or [Therapy Room]. If the [New Session] is chosen, the trainee explores the patient's symptoms, leading to the [Diagnosis & Massage] step. In [Session Finalization] step any additional follow-up communication or adjustments occur. The [Final Assessment] then shows the trainee's performance and potential improvement. Finally, the trainee decides whether to "Continue?" (looping back to the main menu) or "Quit / Exit."
• Figure 2: Main menu of the TheraQuest prototype. The background image shows the default massage room setting generated by DALL-E, OpenAI. Users can customize room backgrounds.
• Figure 3: Trainees can adjust room size, colour ambiance, and furniture layout through an interactive floor plan, with options to save, reset, or randomize the setup. The final customized floor plan will be rendered into a 3D visualization for enhanced spatial representation.
• Figure 4: An AI-generated virtual patient, displaying the patient’s demographic details, lifestyle, and medical history is shown on the left side. Trainees interact via a live chat window on the right side to gather symptoms and provide instructions before proceeding to the "Diagnosis & Massage" stage for hands-on decision-making.
• Figure 5: The left panel allows trainees to select muscle groups for diagnosis, and the right panel enables setting massage intensity through sliders. Disabled sliders appear in gray until their corresponding muscle groups are selected. The original muscle map comes from ref25.