Alfanum DOO Novi Sad
Project dates
Full title of the project:
Development of Systems for Dictation of Medical Findings in Bosnian/Croatian/Serbian including Latin Expressions
Short name of the project:
MEDICTA
Date of the award:
29/01/2021
Project duration:
15 months
Lead Company:
Alfanum DOO Novi Sad (Serbia)
Project Partner(s):
University of Novi Sad, Faculty of Medicine (Serbia)
University of Zagreb, School of Medicine (Croatia)
Abstract
“The project is aimed at the development of systems for automatic recognition of dictated medical findings and their conversion into textual form in Bosnian/Croatian/Serbian, including Latin expressions, in order to increase the productivity of medical practitioners and the quality of service offered to patients.
The general project objective is to relieve highly qualified medical personnel of the task of manual creation of medical findings, which is strenuous, often includes more than one person, may take twice the time (findings dictated into a device are manually transcribed later), and is error-prone. The developed automatical human-machine interaction system based on speech technology will greatly increase the quality of the service offered to patients, thus improving entire healthcare systems in the region where Bosnian/Croatian/Serbian is spoken (Serbia, Croatia, Bosnia and Herzegovina, Montenegro). This is of great importance for this region since it will mitigate the problem of an increasing lack of highly qualified medical personnel. Project activities are aimed at the development of the language model which includes the medical vocabulary typically including Latin expressions, from existing databases of medical findings, which will be pre-processed (text proofread, non-orthographic expressions correctly interpreted etc.) and implemented into the existing framework for automatic speech recognition jointly developed by the industry partner “Alfanum” and the Faculty of Technical Sciences in Novi Sad, Serbia. It should be noted that, although Latin is being repressed by English as lingua franca in medicine, many countries still use it widely.
The developed system will include a client-server infrastructure containing client applications for voice interaction with end-users (dictation of medical findings and composition of textual forms) as well as a server application that will perform actual recognition of what has been said. The server application may be physically located within the medical institution or integrated within its information system on the cloud. In either way, the privacy of patients is much less likely to be jeopardized than in common state-of-the-art solutions often based on speech recognition services hosted on third-party servers. The proposed setup can also be integrated into smart hospital systems and overcomes the necessity for expensive client computers in a scenario where recognition is performed locally. Through this project, the industry partner “Alfanum” will improve its competitiveness and extend the range of application of its speech and language technologies as EU market-oriented technologies as well as innovative products and services based on them. By ultimately increasing the efficiency of medical staff owing to newly established links between partners from within and outside the healthcare domain, the project will contribute to the acceleration of the innovation cycle through the entire value chain.”