Design of a Clinical Data Management Workflow for the COMBAT-VT project

Ventricular tachycardias (VTs) are cardiac arrhythmias that lead to faster heartbeats. These VTs occur in a subgroup of patients that had a myocardial infarction (Ml) and are a major cause of morbidity and mortality. To predict the risk of VTs, a value of left ventricular ejection fraction (LVEF) <35% is used as an indication for treatment with an Implantable Cardioverter Defibrillator (ICD). However, the LVEF proved to be insufficient in identifying patients with a high risk of VT and better indicators have not been identified yet.
The aim of the COMBAT-VT project is to improve the understanding and prediction of the occurrence of post-Ml VT, by combining physiological and data models. These computational models can be used to provide optimized patient-specific therapy guidance. This requires a large amount of input data, in order to optimally train and validate the models. In this QME project a Clinical Data Management Workflow has been designed to support the COMBAT-VT project.
In this design project, the Pilot VT Study was initiated - prior to the COMBAT-VT project - to build the foundations for this workflow. It was important to perform a thorough research into the possibilities in the Catharina Hospital Eindhoven (CZE) through this pilot study, as well as into the safe and justified way to handle the patient data for the purpose of research. In the Pilot VT Study was investigated which data are available, how all data should be handled and analyzed. At the same time we aim to find the first indications for new post-Ml VT risk parameters, while using a small sample size (n = 46).
The Clinical Data Management Workflow consists of five components:
1. Patient selection: selection and inclusion of patients and data;
2. Extraction data: obtaining the data from the different systems at the CZE;
3. Database: saving the data according to the Findable, Accessible, Interoperable, Reusable (FAIR)
principles;
4. Analysis meta data: analyzing the data to explore what is exactly saved of the patients and
what is useful for the models;
5. Sharing data: pre-processing the data and sharing the data with researchers at the TU/e, while
considering laws and regulations.
To gain as many insights as possible, three different data types were collected: structured data (lab results and vital functions), imaging data (echocardiography, MRI and CT), and monitoring data (ECG).
These data were prioritized according to the technical requirements, which were based on the functional requirements (by means of the user stories) of the PhD researchers of COMBAT-VT. After collecting the data, a start was made to pseudonymize and restructure the data to comply to the FAIR principles.
A first analysis of the meta-data provides insights for the users in the major differences in the sizes of the data sets between patients, but also in the fluctuation of the number of the measurements in time.
The Pilot VT Study provided structured and pseudonymized data sets and became a test-case for two recently developed platforms to share medical data: 1) the digital research environment anDREa, as part of the Health Data Portal of e/MTIC, and 2) the Al platform of the CZE, which is a local research environment. Both will be implemented and used within a few months.
In this QME design project, a Clinical Data Management Workflow for the COMBAT-VT project has been designed. Furthermore, the concepts of the new workflow has been clearly presented and has been successfully conducted in a pilot. On the basis of the first results of this pilot, a guide has been written to further set up this new workflow and this QME design project has been finished with recommendations to further conduct the VT-pilot study and for the subsequent COMBAT-VT project.
The new workflow will require further development, especially in restructuring and pseudonymization,
before all the data can actually be accessed by the PhD researchers in the COMBAT-VT project. More challenges are expected in finding the (location of the) raw data and obtaining the patient data in bulk, as well as in the use of the digital research environments. Using a dashboard for analysis of the meta data has a lot of potential (as has been shown with a first prototype of a dashboard in this project as well), but still requires significant work in designing a proper interface. Finally, the technical requirements of the designed workflow need to be further verified and the final design requires validation in the clinical practice, preferably at the end of the Pilot VT Study, before starting the COMBAT-VT project with this new Clinical Data Management Workflow.