JMIR Publications

ABSTRACT

Background:

The importance of preserving and sharing medical knowledge in the shape of evidence-based medical guidelines is well recognized in the medical community. However, their use to improve patient care remains a challenge for physicians, especially during patient consultations. For epilepsy, there are several guidelines that epileptologists can use to diagnose and treat epileptic patients, but it is estimated that up to 20% of the patients are still misdiagnosed. This can have severe consequences for patients, which could have been avoided with the right diagnosis and therefore the right treatment, since around 70% of epilepsy patients who are treated can live without unprovoked seizures.

Objective:

To present the strategy to create a computer interpretable guideline (CIG) from the guidelines of the International League Against Epilepsy (ILAE) and Deutsche Gesellschaft für Neurologie e.V. (DGN). This CIG will be used in the clinical decision support system (CDSS) for diagnosis and therapy recommendation for patients with epilepsy to improve the guideline adherence in the clinical context, the EDiTh project. As this application is considered an investigation device under the European Medical Device Regulation (MDR 2017/745), we also want to demonstrate our modular architecture approach and implementation process to follow the regulatory framework.

Methods:

An interdisciplinary team created a model from the clinical guidelines. This model consists of three consecutive chained models, where the first one is the transformation of the ILAE seizure classification into a decision tree, the second is a decision table of proposed diagnosis of the DGN guidelines, and the third is a decision table of recommended drug treatments. Then, the software was developed under the regulatory requirements of the MDR and in particular the DIN EN 62304 as an investigation device, dividing the software architecture into different modules depending on the medical risk classification.

Results:

The CDSS was created using the CIG that the experts elaborated, and we transformed into a tabular structure that can be used with simple parsing directly in the software. The application can be accessed from any computer in the hospital facility through a web browser, and the results can be saved in a PDF file as a copy of the system information. The EDiTh-App is currently being evaluated in a clinical investigation.

Conclusions:

The creation of a CDSS as medical device software (MDSW) under the MDR can be achieved by modularizing the software architecture into units of different medical risk classification and minimizing the use of unnecessary third-party libraries. Dividing the CIG into several models makes it easier for medical professionals to understand. This can help in their creation and evaluation and speeds up the preparation of a clinical investigation device, thereby improving the patient’s diagnosis and treatment.