Development of Neural Network Module for Autocompletion of Examination Protocol for Modular Medical Information System

The issues of building a modular medical information system, in which the possibility of connecting and replacing modules is provided, are considered. The software is developed using microservice architecture. The module of autocompletion of the examination protocol is presented. The module is based on the use of neural network algorithms for recognizing information. SpeechKit neural network is used to translate speech from audio to text, and YandexGPT is used to extract answers to questions from text. The technologies of data exchange between the modules are analyzed, the technology of remote procedure call gRPC is highlighted. The results of application of the developed module are given.