J Med Internet Res. 2026 Sep 17;28:e91065. doi: 10.2196/91065.
ABSTRACT
BACKGROUND: Virtual agents powered by dialogue trees have emerged as promising tools to overcome limitations of text-only psychological interventions in mental health treatment.
OBJECTIVE: This study evaluated whether a virtual agent-based digital psychological intervention plus treatment as usual (TAU) reduced craving and relapse and improved emotional state, sleep quality, and treatment motivation compared with TAU alone in patients with alcohol use disorder (AUD).
METHODS: A 2-arm, randomized controlled trial with a parallel-group design was conducted at Fuyang No. 3 People’s Hospital between April and December 2023. A total of 93 patients aged 18 to 55 years with AUD were randomly assigned to the intervention (n=47) or control group (n=46) using a random number table. Outcome assessors and data analysts were blinded to group allocation. Patients in the intervention group received the 4-week Echo App V2.0 intervention, comprising 10 cognitive behavioral therapy sessions (30-45 min each). All patients received TAU, including medications, daily 30-minute exercise, and psychological education. The primary outcome was change in craving from T0 (baseline) to T1 (week 4), T2 (1 mo after discharge), and T3 (3 mo after discharge), measured by the Visual Analog Scale, a 10-cm horizontal line. Secondary outcomes were relapse and changes in emotional state, sleep quality, and treatment motivation.
RESULTS: Craving increased after discharge in both groups. The intervention group showed significantly attenuated craving increases at T2 (β=-1.81, 95% CI -3.45 to -0.16; P=.03) and T3 (β=-1.97, 95% CI -3.61 to -0.32; P=.02). At T2, the intervention group showed significantly lower relapse rates (χ21=4.09; P=.04), although this benefit was sensitive to missing-data assumptions in sensitivity analyses. Relapse rates did not differ at T3 (χ21=0.029, P=.86). Sleep quality improved more in the intervention group at T1 (β=-3.28, 95% CI -5.08 to -1.49; P<.001). Perceived stress decreased significantly over time (β=-2.91, 95% CI -5.27 to -0.56; P=.02). Group and change in Perceived Stress Scale (PSS) significantly predicted craving change at T2 (intervention: β=-2.13, 95% CI -4.03 to -0.24, P=.03; PSS: β=0.15, 95% CI 0.03 to 0.26; P=.02) and T3 (intervention: β=-2.24, 95% CI -4.07 to -0.42, P=.02; PSS: β=0.12, 95% CI 0.01 to 0.24, P=.03). Both group (β=1.02, odds ratio 2.77, 95% CI 1.11 to 6.91; P=.03) and number of previous attempts to quit alcohol (β=0.27, odds ratio 1.31, 95% CI 1.06 to 1.63; P=.01) significantly predicted remission at T2. No adverse events were reported.
CONCLUSIONS: Echo App V2.0 attenuated postdischarge craving increases and may reduce short-term relapse among patients with AUD. However, the relapse finding was sensitive to missing-data assumptions and was not sustained at 3 months. Postdischarge continuation strategies may be needed to maintain benefits.
PMID:42752444 | DOI:10.2196/91065
