The effectiveness of chatbots and its difference with usual treatment to reduce the symptoms of depression and anxiety: meta-analysis review

The review aims to test the effectiveness of chatbots in treating depression and anxiety through cognitive behavioral therapy within a short duration of treatment. This review also tries to test whether chatbot therapies are similar to usual treatments. The study pooled data from 12 studies. In the random effects model, at the significance level of α = 0.05, chatbot therapies significantly reduce the symptoms of depression and anxiety with a small to moderate effect size. For anxiety groups, the range of summarized standardized mean difference is from -0.39 to -0.1. For depression groups, the interval of a summarized standardized mean difference ranges from -0.36 to -0.08. The difference in effectiveness between chatbot therapies and traditional therapies is minimal. Therefore, chatbot-provided cognitive behavioral therapy shows promise for the future. Three studies with significant heterogeneity were removed. They differ in terms of outdated chatbots, significantly varying durations of treatments, and the special period during which the study was conducted. These implications may suggest that chatbots are undergoing rapid upgrades, so future researchers should keep up with the changes in chatbot technology. The number of related researches is limited, and moral concern should be considered in future researches.