Information-providing Chatbot

The Development and Evaluation of a Co-designed, AI-based Information-providing Chatbot for Sarcoma Patients: A Pilot Randomised Controlled Trial

This study aims to develop and examine the feasibility and preliminary effectiveness of an AI-based sarcoma chatbot building on ChatGPT (ASCC) to address the information needs of sarcoma patients. We hypothesized: 1) the ASCC will demonstrate good usability; 2) the study will be feasible in terms of all feasibility indicators; 3) the experimental group will report improved satisfaction and self-efficacy, decreased anxiety and stress than the control groups upon completion of the intervention. The ASCC will be developed using a co-design approach. A pilot randomized controlled trial will then be conducted in the three oncology wards of collaborative hospitals. Seventy-eight sarcoma patients will be recruited and randomized to the experimental group (n=39) and the control group (n=39). The experimental group will use the ASCC available 24/7 via voice or text for disease-related questions for one month while the control group will receive usual care.

Study Overview

• Status: Not yet recruiting
• Conditions: Sarcoma
• Intervention / Treatment: Other: ASCC (sarcoma-specific AI chatbot); Other: Routine Clinical Care

• Study Type: Interventional
• Enrollment (Estimated): 78
• Phase: Not Applicable

Contacts and Locations

• Study Contact: Name: Qi LIU, PhD; Phone Number: 85255741349; Email: charlotte-qi.liu@polyu.edu.hk

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: No

Description

Inclusion Criteria:

• (1) diagnosed with soft tissue sarcoma;
• (2) receiving active treatment at any stage;
• (3) aged≥18 years;
• (4) able to speak Cantonese, Mandarin and English;
• (5) access to an electronic device enabling ASCC use

Exclusion Criteria:

• (1) patients with unstable physical or mental conditions;
• (2) inability to communicate logically;
• (3) using chatbots (e.g., ChatGPT, Deepseek) for sarcoma information in the last month;
• (4) mental health disorders documented by the medical system

Study Plan

How is the study designed?

Design Details

• Primary Purpose: Supportive Care
• Allocation: Randomized
• Interventional Model: Parallel Assignment
• Masking: Single

Number of Arms

2

Arms and Interventions

Participant Group / Arm	Intervention / Treatment
Experimental: ASCC; An AI-based sarcoma chatbot building on ChatGPT (ASCC) using a co-design approach.	Other: ASCC (sarcoma-specific AI chatbot); The participants in the experimental group will access the ASCC via a webpage alongside standard information from health professionals for one month (15 minutes weekly).
Other: Control; The standard information from doctor-patient consultations and hospital	Other: Routine Clinical Care; Participants in the control group will receive routine clinical care, including standard information from doctor-patient consultations (e.g., diagnosis, treatment, medical tests) and hospital provided sarcoma and treatment leaflet.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Screening rate	Calculated by dividing the number of sarcoma patients screened by the research team by the number of sarcoma patients.	During the enrollment period of this project.
Eligibility rate	Calculated by dividing the number of participants who are eligible by the number who are screened.	During the enrollment period of this project.
Consent rate	Calculated by dividing the number of participants who were randomized into intervention and control groups by those provide consent.	During the enrollment period of this project.
Randomization rate	Calculated by dividing the number of participants who consent to join the study by the number who are eligible.	During the enrollment period of this project.
Attendance rate	Calculated by dividing the number of participants who complete the intervention by those who are randomised	Baseline and immediately after intervention (T1)
Retention rate	Calculated by dividing the number of participants who remain in the study by those who are randomized. This will be calculated by groups.	Baseline and immediately after intervention (T1)
Completion rate	Calculated by dividing the number of participants who returned questionnaires by the number of questionnaires distributed. Calculated by dividing the number of participants who remain in the study by those who are randomized. This will be calculated by groups.	Baseline and immediately after intervention (T1)
Proportion of missing data	Calculated as the percentage of missing values in the dataset. Unknown or blank values will be considered missing values.	Baseline and immediately after intervention (T1)
Adverse events	Adverse events are defined as unfavourable and unintended events that are not present, or appear to have worsened during the study.	immediately after intervention (T1)

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Information satisfaction	The EORTC QLQ-INFO25 will be used to evaluate participants' satisfaction with the information received. A four-point Likert scale is used to score 25 items ('nothing' = 1 to 'a lot' = 4), with higher scores indicating greater information satisfaction.	Baseline and immediately after intervention (T1)
Anxiety and stress	The Depression Anxiety Stress Scale-21 (DASS-21) will be used to assess anxiety and stress.(36) 21 items are scored using a four-point Likert scale, with higher score indicating more severe symptoms.	Baseline and immediately after intervention (T1)
Self-efficacy	Self-efficacy will be assessed using the Strategies Used by People to Promote Health (SUPPH) with good reliability and validity. This 28-item likert scale measure self-care self-efficacy, with higher scores indicating more positive perceptions of self-efficacy.	Baseline and immediately after intervention (T1)
System Usability Scale	System Usability Scale (SUS) will be used to measure the usability of digital health solutions by 10 items rated on a 5-point Likert scale from 0-100. The chatbot will be considered highly usable if the mean score exceeds 68.	immediately after intervention (T1)

Collaborators and Investigators

• Sponsor: The Hong Kong Polytechnic University
• Collaborators: Queen Elizabeth Hospital, Hong Kong; Prince of Wales Hospital, Shatin, Hong Kong; Queen Mary Hospital, Hong Kong

Study record dates

Study Major Dates

• Study Start (Estimated): April 1, 2027
• Primary Completion (Estimated): July 31, 2028
• Study Completion (Estimated): July 31, 2028

Study Registration Dates

• First Submitted: November 18, 2025
• First Submitted That Met QC Criteria: March 26, 2026
• First Posted (Actual): April 1, 2026

Study Record Updates

• Last Update Posted (Actual): April 1, 2026
• Last Update Submitted That Met QC Criteria: March 26, 2026
• Last Verified: November 1, 2025

More Information

Terms related to this study

• Keywords: cancer; artificial intelligence; sarcoma; chatbot; health information; ChatGPT; unmeet needs
• Additional Relevant MeSH Terms: Neoplasms by Histologic Type; Neoplasms, Connective and Soft Tissue; Neoplasms; Sarcoma
• Other Study ID Numbers: ASCC Intervention

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: UNDECIDED

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No