Ingeniería Biomédica · 2026

Herramienta digital de apoyo emocional con Chatbot de IA para personas con ansiedad y estrés en Colombia

Resumen

Currently, mental health represents a significant problem in Colombia, especially due to the increase in cases related to anxiety and stress, as well as the difficulties in accessing professional psychological care. In response to this situation, a web-based emotional support platform was developed that combines psychometric assessment and conversational artificial intelligence, aiming to provide an accessible and useful tool for users. The developed system integrates two main components. The first is a psychometric testing module designed to evaluate anxiety and stress levels using previously validated instruments. The second is a conversational chatbot based on artificial intelligence, supported by the Augmented Retrieval by Generation (RAG) mechanism. This mechanism allows the chatbot's responses to be built from information contained in a specialized mental health database. To achieve this, the information was processed through text segmentation, semantic embedding using sentence transformers, and vector indexing with FAISS. Furthermore, the results obtained from the psychometric tests are incorporated as context during the conversation, allowing the responses to adapt from the outset to the situation reported by each user. The chatbot was designed incorporating support strategies related to Cognitive Behavioral Therapy, mindfulness, and psychoeducational processes, especially focused on managing anxiety and stress. Likewise, the system has mechanisms to identify potential signs of emotional distress and, in such cases, guide the user to professional support and care lines. This was done taking into account that the tool does not replace clinical psychological care, so a responsible approach was sought from an ethical and user safety perspective. The platform was developed using technologies such as Flask, MongoDB, FAISS, and multilingual semantic embedding models. To evaluate its functionality, functional tests were conducted, and instruments such as the System Usability Scale (SUS) and the Technology Acceptance Model (TAM) were applied. The results obtained show that the system can function as a complementary support tool in contexts where access to mental health services is limited. Finally, the project integrates psychometric evaluation, semantic information retrieval, and automated conversational support on a single platform, seeking to contribute to the development of accessible digital tools for emotional support within the Colombian context.

Palabras clave