Paziente birtuala simulatzeko txatbot medikoa

In this work, we will present the creation of the prototype of a Spanish chatbot that will take the role of a virtual patient in the conversation between doctor and patient, together with the results obtained. With this aim, we have developed several resources. On the one hand, a corpus of doctors’...

Descripción completa

Detalles Bibliográficos
Autores: Idoyaga Bazán, Mikel, Arana Gonzalez, Janire, Lafuente Sánchez, José Vicente, Espina Valiño, Elisa, Gojenola Galletebeitia, Koldobika, Atutxa Salazar, Aitziber
Tipo de recurso: artículo
Fecha de publicación:2025
País:España
Institución:Universidad del País Vasco
Repositorio:Addi. Archivo Digital para la Docencia y la Investigación
OAI Identifier:oai:dnet:addi________::1f47dacf7573d164c599f9469bb18686
Acceso en línea:http://hdl.handle.net/10810/79261
Access Level:acceso abierto
Descripción
Sumario:In this work, we will present the creation of the prototype of a Spanish chatbot that will take the role of a virtual patient in the conversation between doctor and patient, together with the results obtained. With this aim, we have developed several resources. On the one hand, a corpus of doctors’ questions has been compiled, each question labeled with its basic intent. Second, a set of clinical reports was obtained, along with the physician-patient interview associated with each one. Using these resources, a chatbot that can play the role of a virtual patient has been trained using artificial intelligence and neural networks. The system has been evaluated using different metrics with positive results, thus paving the way for the construction of automatic medical dialogue systems.; Lan honetan, mediku eta pazientearen arteko elkarrizketan paziente birtualaren rola hartuko duen gaztelaniazko txatbot baten prototipoaren sorrera aurkeztuko da, lortutako emaitzekin batera. Txatbotaren erantzunak benetako pazientearenak bezalakoak izateko helburuarekin, zenbait baliabide garatu dira lan honetan. Alde batetik, medikuaren galderetatik eta pazienteen txosten klinikoetatik hartutako erantzunen corpus bat baliatu da, eta pazienteak erantzungo lituzkeen esaldien corpus berria sortu da. Baliabide horiek erabilita, paziente birtualaren papera egin dezakeen txatbotaren hizkuntza naturalaren sorkuntzarako sistema bat entrenatu da adimen artifizialeko sare neuronalen bidez. Zenbait metrika erabiliz ebaluatu da sistema, emaitza positiboekin. Medikuntzako elkarrizketa-sistema automatikoak eraikitzeko bidea ireki da horrela.