Sistema inteligente para apoyar el diagnóstico del trastorno límite de la personalidad en un puesto de salud en Tumán
Cargando...
Fecha
2026
Autores
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Católica Santo Toribio de Mogrovejo
Resumen
La presente investigación surge ante la dificultad de diagnosticar el trastorno Límite de la Personalidad en el área de psicología del puesto de salud en Tumán; debido a la ausencia de especialistas, el tiempo limitado para las entrevistas, y la alta demanda de atención, lo que resulta en un deterioro progresivo de la salud mental del paciente. Por lo tanto, se propuso el desarrollo de un sistema inteligente basado en una técnica de Machine Learning que permite apoyar al especialista, en el diagnóstico del trastorno mental. Para ello, se realizó una evaluación de cinco algoritmos de clasificación, seleccionando el más adecuado para este contexto. Por consiguiente, el modelo computacional se basó en el algoritmo de Árbol de Decisiones obteniendo un 91.55% de exactitud, 90.07% de precisión, 93.39% de sensibilidad y 89.71% de especificidad. Asimismo, se desarrolló un aplicativo web que se integró con dicho modelo, para ello, se utilizaron tecnologías como Django para el back-end, Scikit-learn para el modelo computacional, PostgreSQL para la base de datos y la metodología CRISP-DM en la guía del desarrollo de minería de datos. Por último, se obtuvo los resultados basados en la norma ISO 25010 con 91.11% en reconocibilidad, 87.77% en aprendizabilidad, 84.44% en operabilidad, 86.66% en protección de errores, 100% en involucración con el usuario, 83.3% en inclusividad y asistencia, y 86.66% en auto-descriptividad, llevando así una sólida aceptación por los especialistas en la capacidad de la interacción del aplicativo web.
The present investigation resulted from the difficulty in diagnosing borderline personality disorder in the psychology area of the Tumán health center; due to the absence of specialists, the limited time for interviews, and the high demand for attention, which resulted in a progressive deterioration of the patient's mental health. Therefore, the development of an intelligent system based on a Machine Learning technique was proposed to assist the specialist in diagnosing the mental disorder. For this purpose, an evaluation of five classification algorithms was carried out, selecting the most suitable one for this context. Consequently, the computational model was based on the Decision Tree algorithm, achieving an accuracy of 91.55%, a precision of 90.07%, a sensitivity of 93.39%, and a specificity of 89.71%. Likewise, a web application was developed that integrated with the aforementioned model. For this, technologies such as Django for the back-end, Scikit-learn for the computational model, PostgreSQL for the database, and the CRISP-DM methodology to guide the development of data mining were used. Finally, the results were obtained based on the ISO 25010 standard, with 91.11% in recognizability, 87.77% in ease of learning, 84.44% in operability, 86.66% in error protection, 100% in user engagement, 83.3% in inclusivity and assistance, and 86.66% in self-descriptiveness, thus achieving solid acceptance from specialists regarding the web application's interaction capability.
The present investigation resulted from the difficulty in diagnosing borderline personality disorder in the psychology area of the Tumán health center; due to the absence of specialists, the limited time for interviews, and the high demand for attention, which resulted in a progressive deterioration of the patient's mental health. Therefore, the development of an intelligent system based on a Machine Learning technique was proposed to assist the specialist in diagnosing the mental disorder. For this purpose, an evaluation of five classification algorithms was carried out, selecting the most suitable one for this context. Consequently, the computational model was based on the Decision Tree algorithm, achieving an accuracy of 91.55%, a precision of 90.07%, a sensitivity of 93.39%, and a specificity of 89.71%. Likewise, a web application was developed that integrated with the aforementioned model. For this, technologies such as Django for the back-end, Scikit-learn for the computational model, PostgreSQL for the database, and the CRISP-DM methodology to guide the development of data mining were used. Finally, the results were obtained based on the ISO 25010 standard, with 91.11% in recognizability, 87.77% in ease of learning, 84.44% in operability, 86.66% in error protection, 100% in user engagement, 83.3% in inclusivity and assistance, and 86.66% in self-descriptiveness, thus achieving solid acceptance from specialists regarding the web application's interaction capability.
Descripción
Palabras clave
Salud mental, Aprendizaje automático, Diagnóstico clínico, Mental health, Machine learning, Clinical diagnosis
Citación
M. Marin, “Sistema inteligente para apoyar el diagnóstico del trastorno límite de la personalidad en un puesto de salud en Tumán,” tesis de licenciatura, Fac. de Ingeniería, Univ. USAT, Chiclayo, Perú, 2026. [En línea]. Disponible en: https://hdl.handle.net/20.500.12423/10075
