Integration of Psychometric Assessment and Ai-Based Conversational Simulations To Support Social Anxiety In Higher Education

Social anxiety among university students is associated with difficulties in academic performance, classroom participation, and interpersonal interaction. In this context, this paper presents a pilot study of a web-based platform designed to support students who experience this type of anxiety. The proposed system structures the user experience into four modules: administration of psychometric assessments (LSAS and SPIN), emotional exploration through structured questions, personalized conversational simulations and My Journal module for results tracking. The architecture is based on a three-tier design that integrates a React-based frontend, backend services implemented in Express and Flask, and data storage using PostgreSQL and MongoDB, together with a language model to generate contextualized feedback. The evaluation of the proposed prototype was conducted with 45 university students, in which usability was analyzed using the System Usability Scale (SUS). The results indicate favorable usability and a positive perception of the modules, with conversational simulations receiving particularly high ratings, while the \textit{My Journal} module revealed opportunities for improvement. In general, the findings suggest that the prototype is technically viable and well accepted, providing functional evidence of the value of integrating conversational simulations into web-based digital support systems in higher education.