gms | German Medical Science

Introduction: Using telemedicine approaches to support the care of patients with conditions such as cardiovascular diseases is a well-known research target in medical informatics (c.f. [1]). The increasing prevalence of smartphones and wearables suggests to make use of these devices to generate relevant data for telemedicine approaches and to support medical research. A sub-study within the HiGHmed Use Case Cardiology [2] addresses the standard-based data integration of such data into university-hospital-IT. One challenge is to monitor the incoming data regarding its quality, i.e. study nurses need to monitor the incoming data to promptly address problems. Objective of our work was to develop an interoperable dashboard to support study nurses.

State of the Art: Dashboards for data from mobile devices and wearables typically focus on telemedicine use-cases (e.g. [3]). A dashboard to monitor if such data is of adequate quality for research has other information needs than a dashboard to monitor patient health. To the best of our knowledge there is no published work addressing the particular needs of our use case.

Concept: We derived the required functionality for our dashboard based on our clinical partners' input. Core functional requirements were to (1) identify inactive patients, i.e. patients not sending sensor data or filling out questionnaires and to (2) visualize data for single patients, to notice data quality problems, such as implausible values. Most important technical requirements were (3) interoperability, i.e. building up on standard based data representation and standardized interfaces to allow seamless roll-out to interested project partners and to (4) employ a knowledge-based approach to represent desired visualizations.

Implementation: We implemented the dashboard as R Shiny App making use of the technical standards openEHR REST-API, AQL and the openEHR clinical information models for data representation. From previous projects we knew that these are a suitable basis for roll-out to other partners supporting this technical infrastructure. We employed a knowledge-based representation [4] for visualizations in the dashboard. The dashboard is available as open source code [5] and is in active use at Hannover Medical School since February 2024. The dashboard was not subject to formal evaluation.

Lessons learned:

1.

The development was an iterative process, i.e. when the study nurses started using the first version of the dashboard, we identified things to improve. An agile software development approach seems suitable even in such small projects.

2.

Shiny was a suitable technical means for our research driven dashboard, since it enabled a quick but flexible implementation and a smooth roll-out process, aside from minor problems getting R-Studio and Shiny installed for the study nurses by our IT-department.

3.

A knowledge-based approach for representing visualizations was useful in the development process, since it eased switching between explorative development in an R-notebook and production in the Shiny dashboard. However, the knowledge-based representation did not seem sensible for one tabular visualization involving many separate data requests.

4.

The dashboard proved useful to timely notice issues, e.g. a stop in incoming data due to expired SSL certificates or falsely set date entries in incoming data.

The authors declare that they have no competing interests.

The authors declare that a positive ethics committee vote has been obtained.