A self-diagnosis method for stroke using chatbots creates enhanced retraceability (RAG).

Abstract

The invention relates to a method for self-diagnosing stroke using a chatbot that generates retrieval enhancement (RAG), including: a step to store user text information (S1) to store information about the user's text input speed and text error range of the chatbot; a step to activate the chatbot (S2) to activate the chatbot; a step to receive user text (S3) to receive the user's text from the chatbot; a step to identify anomalies (S4) to compare the received user text information with previously stored user text information in real time to determine whether the compared result is normal or abnormal; a step to request stroke check (S5) to request, if it is determined that the text error range exceeds a predefined reference value, to check for stroke via the chatbot; a step to find keywords (S6) to compare the user's text with previously stored stroke keywords to find stroke keywords;The questionnaire creation step (S7) selects queries related to stroke keywords found among pre-stored stroke queries to generate the questionnaire; the questionnaire transmission step (S8) transmits the generated questionnaire to the user's terminal; the severity calculation step (S9) compares the received response messages with the pre-stored severity score to calculate the severity score; and the user's stroke injury check step (S10) provides a stroke self-diagnosis result based on the user's severity score.

Images