Effects of a Brief Mindfulness-Based Chatbot Intervention (Mindbot) on Mindfulness Level and Stress Among Cancer Patients: A Randomized Controlled Trial
Effects of a Brief Mindfulness-Based Chatbot Intervention (Mindbot) on Mindfulness Level and Stress Among Cancer Patients: A Randomized Controlled Trial
This study tests a new digital tool for hospitalized cancer patients. The tool is a chatbot on LINE, a popular messaging app. The chatbot is called Mindbot. Mindbot leads a 10-minute mindfulness practice each day. The practice uses breathing exercises and virtual reality (VR).
Adult cancer patients in the hospital ward can join. Participants are randomly placed into one of two groups. One group uses Mindbot every day during their hospital stay. The other group receives usual hospital care through the hospital's existing health app.
Researchers measure mindfulness, stress, and heart rate variability (a measure of body stress) before and after each practice. They also collect saliva samples to check a stress hormone called cortisol. Researchers track fatigue and cancer symptoms too. These measurements happen at three time points: hospital admission, hospital discharge, and the first follow-up clinic visit.
Psychological distress is common among hospitalized cancer patients and, if unmanaged, may negatively affect quality of life and disease trajectory. Traditional mindfulness-based stress reduction programs require prolonged daily practice (e.g., 45 minutes over 8 weeks) and show low adherence in this population, largely due to cancer-related fatigue. Brief mindfulness sessions (10-20 minutes) delivered via widely-used instant messaging platforms may improve accessibility and adherence compared with standalone apps or websites that require additional learning effort.
This is a two-arm, parallel-group randomized controlled trial evaluating a LINE chatbot-delivered brief mindfulness intervention (Mindbot) combined with VR-guided audio-visual content, compared with usual care, among cancer patients admitted to a hospital ward at a medical center in northern Taiwan.
Randomization: Participants are allocated 1:1 using an online randomizer (Sealed Envelope Simple Randomiser) with random permuted blocks of mixed sizes 4 and 6. The allocation sequence is generated by an independent researcher not involved in recruitment, and allocation results are concealed in sequentially numbered, opaque sealed envelopes until after consent and eligibility confirmation. To reduce cross-contamination among roommates in shared hospital rooms, all participants in the same room are assigned to the same study arm.
Blinding: Because the intervention is delivered via a distinguishable QR code (chatbot vs. existing hospital app), participants and on-site research staff cannot be blinded to allocation. After data de-identification and re-coding, the statistician performing the primary analysis is blinded to actual group assignment until the primary analysis is complete.
Intervention: The experimental arm receives automated push notifications at 13:00 and 21:00 daily, prompting a 10-minute structured mindfulness session (breath awareness, body scan, and mindful acceptance of hospital ambient sounds), guided by the virtual character "Mindbot" in a VR environment. Each session includes a pre- and post-practice assessment of cancer symptom severity and a 7-minute resting heart rate variability (HRV) measurement. The control arm receives usual hospital care and access to the hospital's existing health management app, which provides general disease-related e-education without mindfulness content.
Outcome assessment: Primary outcomes (mindfulness level, perceived stress, HRV, salivary cortisol) and secondary outcomes (fatigue, cancer symptom severity) are assessed at baseline (T0, at enrollment), hospital discharge (T2), and the first outpatient follow-up visit (T3). Salivary cortisol is additionally sampled four times daily (on waking, 30 minutes post-waking, afternoon, and late evening) throughout the intervention period. Qualitative feedback shared by participants through the LINE chat interface is de-identified, transcribed, and analyzed using content analysis to complement the quantitative findings.
Inclusion Criteria:
Exclusion Criteria: