Maintained effectiveness of a smartphone-based chatbot coach (BalanceUP) for mental well-being in headache sufferers: preliminary findings from a single-arm 6-month follow-up study.

Abstract

Background: Behavioral interventions delivered by smartphones have been proven effective in improving physical and mental well-being in clinical and non-clinical samples. However, it remains unclear whether improvements are sustained after intervention completion. This single-arm study evaluated the sustained effectiveness of BalanceUP, a smartphone-based coaching intervention delivered by a conversational agent (CA), for people suffering from frequent headaches.

Methods: Following a randomized controlled trial, where an intervention group was compared with a waitlist control group, this within-group comparison examined follow-up results of all participants who participated in the coaching program. Measurements were taken pre-, post-, and 6 months after intervention start. Changes in the primary outcome mental well-being [Patient Health Questionnaire Anxiety and Depression Scale (PHQ-ADS)] and secondary outcomes (psychosomatic symptoms, stress, headache-related self-efficacy, intention to change behavior, presentism, absenteeism, pain coping) were analyzed using repeated measure analysis of variance (ANOVA) and Cohen's d. Additionally, baseline predictors of improvement were explored.

Results: Of 179 headache sufferers who started coaching [mean age 39.2 years, standard deviation (SD) 11.8 years; 157/179, 87.7% women], 117 (65.4%) completed the intervention and 67 (36.3%) provided 6-month follow-up data. Improvements in mental well-being were maintained at follow-up with a medium effect size (Cohen d=-0.54). Secondary outcomes also showed sustained improvements with effects ranging from small to medium (d=-0.24 to 0.70), including depression, anxiety, somatic symptoms, perceived stress, headache management self-efficacy, application of behavior change techniques, absenteeism/presentism, as well as pain coping skills. Higher baseline stress and presenteeism were associated with greater improvements in well-being.

Conclusions: These preliminary findings suggest the potential for CA-delivered coaching to support improvements in mental well-being and psychological functioning over time. However, the uncontrolled study design, attrition at follow-up, and other methodological limitations require cautious interpretation of the findings. More methodologically robust controlled studies are essential to establish the sustained effectiveness of digital behavioral interventions before drawing definitive conclusions.