Integration of artificial intelligence in mental health therapy: an adaptive model for psychological interventions in the digital era

Background: The rise of mental health disorders has driven the need for accessible and adaptive psychological support through digital means. This study explores user perceptions of adaptive AI in mental health therapy, focusing on interaction quality and effectiveness. Aim: This study aims to examine the relationship between the adaptive intelligence of AI systems and the quality of psychological interaction on the perceived effectiveness of interventions in AI-based digital therapy. Method: Employing a descriptive correlational quantitative approach, data were collected from 30 users of mental health AI therapy applications such as Woebot and Wysa. The instrument used was a Likert-scale questionnaire (1–5), and the data were analyzed using descriptive statistics, Pearson correlation, and simple linear regression via SPSS version 26. Result: The mean scores for all variables fell within a moderately positive range. However, no significant relationship was found between adaptive intelligence or interaction quality and intervention effectiveness (r = -0.308 and r = -0.001; p > 0.05). The regression model was also not significant (R² = 0.096; p = 0.256), indicating a low contribution of the independent variables to perceived effectiveness. Conclusion: These findings provide valuable insights for the development of more personalized and empathetic AI systems in digital psychological services, and serve as a foundation for ethical and human-centered AI design and policy integration.