gms | German Medical Science

Background: Personalized Medicine is a rapidly growing area in healthcare which fundamentally changes the treatment of patients. This is especially true for cancer therapy where more and more hospitals conduct molecular tumor boards (MTB) bringing together experts from various clinical fields to jointly discuss individual patient cases [1]. Yet the exploration and discussion of all relevant data and information poses a tremendous challenge: Because of the experts’ distinct background expertises and time constraints, all this needs to be presented in a concise form that is quick and easy to grasp and supports the consensual elaboration of sound treatment recommendations.

Study Goal: We approach this need by developing a software solution which focuses on a novel, intuitive graphical user interface that integrates and visualizes patient data from several clinical information, laboratory and imaging systems, etc., together with additional external information from cancer-research systems (e.g. cBioPortal), genomic and pathway databases (e.g. KEGG), as well as (bio)medical literature (e.g. PubMed).

Proposed Method/Discussion Points: To detect the actual needs within a MTB, we initially visited five German MTBs [2] where we used questionnaires and interviews to gather information about specific requirements of each MTB, management and workflows, perceived weaknesses and room for improvements, as well as ideas on functionalities an ideal software tool should provide. The resulting requirement analysis formed the basis of our development process which is now accompanied by regular visits to our local MTB to discuss whether our software indeed reflects the experts’ specific needs and wishes.

Our initial implementation applied a large (84 inch) multi-touch screen table that allowed the MTB experts to collaboratively and interactively explore all data and information sources and arrange the different kinds of documents, like clinical/patient data, radiologic images, research papers, etc. on the screen [3], [4]. But although the multi-touch interface was appreciated and the data visualizations deemed to be very valuable, feedback showed that the time constraints during the MTB disallow the intended interactive exploration. We therefore divided the system and interface into two parts: One for the clinical experts to use on their regular work computers when preparing a MTB to perform their explorative research and then for the MTB moderator to integrate and (visually) condense the results into a succinct presentation, and the second one to provide the means to present this during the MTB on any touch-enabled screen dynamically.

In order to observe all relevant legal regulations, we involved the hospital’s data protection officer from the very beginning in our development process and follow a strict separation of concerns in the software architecture: The user interface focuses on the presentation layer, performs no data processing at all and is realized in HTML and JavaScript using Electron (enabling platform independence/portability). It is securely connected by an encrypted REST interface with the backend services located within the secured hospital network. Those services - implemented in Java using Spring Boot and its sub-frameworks - perform the actual data fetching, processing and integration from the source systems and also provide an additional data protection layer by integrating data pseudonymization.

The authors declare that they have no competing interests.

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