gms | German Medical Science

Background: Critical incident reporting in healthcare can be considered as one of the cornerstones of patient safety. Its purpose is to identify and document unusual events which take place in a healthcare setting. Information submitted through this process using a Critical Incident Report System (CIRS) is beneficial for the development of interventions to mitigate hazards [1]. In Germany, the number of reports delivered compared to other countries such as England is quite minimal [2], [3]. The reasons for this situation are different: while filling in a CIRS report, the user does not get feedback and no advice on how to enter valuable information or which information might be valuable [4]. The analysis of a reported event takes an unusually long time which leads to users getting a feedback very late. However, the need for change and advice is urgent. The time between action and reaction is lengthy. The goal of this project is to develop a prototypical interactive cross-platform which would enable users to provide promptly detailed information and feedback concerning an incident. Artificial Intelligence enabled interaction provides support to enter better information with all critical aspects to understand the problem. Based on the data- and knowledgebase of a critical incident reporting system, users get preliminary feedback about the report with a set of proposals of actions.

Methods: For this project, a cross-platform was developed using Flutter which is a Google UI toolkit for building mobile, web and desktop applications from a single codebase. Interactivity is realised using the machine learning kit for Firebase also provided by Google. As expected, about 7000 reports inquired from CIRSmedical.de and their feedback are used as a knowledgebase to train the system and minimize errors [5]. In addition to the reports, the Open-Task-Process Model (OPT-Model) which deals with the systematic description of task complexity are incorporated into the system to ensure proper analysis of the submitted report and proper guidance of the user. This model provides a step by step method which directs the user on which information is important for continuous analysis of the event being reported based on their previous entries. Additionally, to ensure that the platform functions as desired, numerous tests shall be conducted by potential users of the platform at the end of this project.

Results: A cross-platform prototype which provides a modern interface for interaction (automatic redirection of the user and pop-ups), analysis and prompt feedback for the user. In addition, the data quality of collected reports may be increased due to the artificial intelligence-enabled interaction and analysis process. Furthermore, reviews from tests provided by users of the platform shall be analysed for future improvements and compared to the previous system.

Conclusion: The application of artificial intelligence methods in patient safety allows the detection of pattern, rare events as well as the support and a guided interaction between user and system. The reporting of critical events is an indicator of patient safety culture and collects valuable information about the safety awareness in medical environments.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.