AI-Guided Personalized Relaxation for Hemodialysis Patients

Patients undergoing maintenance hemodialysis frequently experience impaired health-related quality of life, anxiety, and difficulties in coping with the long-term demands of treatment. Artificial intelligence (AI)-guided interventions may provide personalized psychological support and promote adaptive coping through tailored relaxation strategies. This randomized controlled mixed-methods trial evaluates the effectiveness of an AI-guided personalized relaxation intervention delivered through a smartphone application among adults receiving maintenance hemodialysis. Participants are randomly assigned to either an intervention group receiving AI-guided personalized relaxation in addition to routine hemodialysis care or a control group receiving routine care alone. The 6-week intervention includes individualized breathing exercises, mindfulness practices, guided imagery, progressive muscle relaxation, and adaptive feedback generated according to participant characteristics, emotional responses, and engagement patterns. The primary outcome is health-related quality of life measured using the Kidney Disease Quality of Life-36 (KDQOL-36™). Secondary outcomes include anxiety symptoms measured by the Generalized Anxiety Disorder-7 (GAD-7), coping strategies measured by the Brief COPE Inventory, and user acceptance of the AI-guided intervention. Qualitative interviews are conducted with intervention participants to explore their experiences, perceived benefits, feasibility, and acceptability of the intervention. The study aims to determine whether AI-guided personalized relaxation can improve quality of life, reduce anxiety, and enhance coping among patients undergoing maintenance hemodialysis while providing insight into the acceptability and feasibility of integrating AI-supported psychosocial interventions into routine dialysis care. The personalization system incorporate a pre-trained facial emotion recognition model using the smartphone's front-facing camera. Facial expressions will analyze in real time to classify emotional states (e.g., anxious, neutral, or relaxed) and adapt relaxation recommendations. Facial images will not stored, and the feature will used solely for intervention personalization after informed consent. Sample size will estimated a priori using GPower (version 3.1) software for a two tailed analysis of difference between two independent means within -t- test family using the following parameters; effect size (Cohen's d = 0.64) which had been reported by Zhang et al. (2025), significance level of 0.05, a statistical power of 90%, and an allocation ratio of 1:1. The minimum sample size will require 106 participants (53 for each group). To compensate for an anticipated attrition rate of 20%, the sample size will increased to 128 participants (64 participants per group).

Inclusion Criteria Adults aged 18 years or older. Diagnosed with end-stage renal disease (ESRD). Receiving maintenance hemodialysis for at least 3 months. Able to communicate and complete study questionnaires. Possession of, or access to, a smartphone capable of running the study application. Willing and able to provide written informed consent. Exclusion Criteria Severe cognitive impairment or inability to complete study assessments. Severe psychiatric illness requiring acute treatment or hospitalization. Significant visual, hearing, or communication impairment preventing participation in the intervention. Participation in another psychosocial, behavioral, or relaxation intervention study during the study period. Inability or unwillingness to use the smartphone-based intervention.