A rough sketch with the required functions of the query process served as a starting point for the development of an interactive prototype. We first examined the search interfaces of already existing tools for a similar scope of application in order to get a clearer picture of how such a search tool could look in its final version. Moreover, to elaborate the sketch towards a prototype ready for evaluation, the input of GBA IT experts was needed. Thus, we planned and conducted a 2-day workshop with 8 GBA IT members, 7 men and 1 woman. Three researchers with expertise in usability moderated the workshop, in which individual steps of a UCD process were pursued in 2 groups of 4 people each [11]. First, the participants discussed the typical context of use of the Sample Locator, resulting in the design of 2 hypothetical users (hereafter referred to as personas) representing potential end users. Second, the participants defined 2 example interaction designs. For this purpose, the groups were each assigned a specific use case to describe a typical usage sequence to be executed with the tool. The use cases were derived from real requests from researchers to biobanks, which were previously collected for a requirements analysis to determine the scope of performance of the Sample Locator. The use cases (adapted from the German version) can be found in Multimedia Appendix 1. With the acquired knowledge, the participants designed low-fidelity mock-ups using Balsamiq (Balsamiq Studios) [12]. The main ideas of these mock-ups were discussed, extracted, and summarized into a comprehensive midfidelity prototype. It was modeled using Microsoft PowerPoint (Microsoft Corp) [13]. The individual slides were linked together with hyperlinks, achieving an interactive navigation through the prototypical tool. Regarding the presentation of the user interface, the website layout of GBN [14] served as a template to harmonize the look with GBN's corporate design. Subsequently, the first draft of the prototype was pretested by 3 male researchers with a biobank background who are affiliated with GBA. In total, 3 versions of the Sample Locator prototype were created in the course of this work. Following the iterative nature of the design process, the versions are based on each other. The second version, which the questionnaire described below refers to, was used for the usability tests. This in turn resulted in the third and final prototype, which was enriched by feedback from the evaluation.

Based on this preliminary work, the evaluation study was carried out. It was conducted as a usability analysis with female and male researchers aged between 18 and 67 years who work in a scientific institution in Germany. Prior to the start of the study, approval from the ethics committee of the Charité – Universitätsmedizin Berlin was obtained.

The study plan envisaged the recruitment of between 30 and 50 respondents to the survey over a period of 6 weeks. With this number of participants, a detection of 95% to 98% of problems within an application can be expected [15]. In order to address suitable participants, one contact person per GBN partner biobank (n=13) was determined in advance to personally approach potential end users with information material. The study team chose this intermediary approach to ensure the anonymity of the participants. Except for the required activity in the research environment, there were no further inclusion or exclusion criteria. Subsequently, the identified participants received an email from the contact person with a link to the online survey tool, LamaPoll (Langner/Maibaum/Notev GbR) [16]. A first introductory page served to briefly inform the participants about the study and to provide the prototype via a download link. After agreeing to participate by clicking a consent button, the participant was forwarded to the study questionnaire.

The study questionnaire for the prototype evaluation consisted of 3 parts: (1) Sample Locator tasks and related questions on feasibility, (2) questions concerning the usability, and (3) general information.

In the first part, 4 tasks needed to be solved with the help of the provided prototype as a basis for subsequent questions. The first task was to search for samples of male patients with lung cancer. The others were to register to the Sample Locator, set project information, and refine the search query (lung metastasis samples plus excluding PAXgene-fixed tissue). The last task was to start negotiations with 2 selected biobanks. The tasks were designed to guide the test users through the prototypical system so they received insights into several possible functionalities. As with formulating the use cases for the workshop, these tasks were conceived based on the previously collected real inquiries to the biobanks. The first 6 questions enabled the researchers to evaluate the prototype’s general intuitiveness and comprehensibility using a 5-point Likert scale (1=strongly disagree, 3=neutral, 5=strongly agree). The following 10 open questions aimed to elicit opinions on content and appearance of the individual steps and further comments on the prototype.

The second part contained the System Usability Scale (SUS), a widely applied and validated score for the quantitative measurement of the usability of an IT application [17]. The wording was slightly adapted by changing “system” to “application” to facilitate the comprehension of the scale.

The third part collected information on age and computer-handling characteristics with 3 questions.

The final questionnaire (adapted from the German version) can be found in Multimedia Appendix 2.

The data from the pretest were not subjected to a sophisticated analysis due to the small amount of data, so they were immediately examined and implemented by a scientist. Two scientists documented and analyzed the collected data for the usability assessment. The applied data analysis methods include a quantitative evaluation for the SUS score, the calculation of the average rating and the standard deviations for all closed questions, and a descriptive qualitative content analysis for open questions. The result of the qualitative content analysis was a categorized list of comments and suggestions for improvement by topic, which was then prioritized according to its practicability. This means that changes to the user interface can be directly incorporated into the final prototype, while technical aspects must first be discussed with the developers.

The workshop laid the groundwork for the following usability analysis. Initially, the 2 groups participating in the workshop each created one persona of a researcher. Since the results overlapped to a high degree, the 2 outcomes were combined into 1 model persona (Figure 1) for a better overview throughout the course of the workshop. The next step—the development of the interaction designs—resulted in 2 sketched workflows of different application scenarios according to the use cases provided to the workshop participants. Here, the participants’ aim was to understand the essential steps of the process that were necessary for a successful trial of the respective use case and to visualize them in a flowchart (for the results, see Multimedia Appendix 3). Finally, the groups designed Sample Locator models by building on the previous steps, enabling the reproduction of the given use cases.

The first version of the prototype was consequently created considering the graphically sketched specifications of the functionalities and the mock-ups developed during the workshop (for some impressions, see Multimedia Appendix 4). The second version was a revision based on the feedback from biobank-based reviewers from the pretest. However, the revisions were mainly individual imprecise formulations that were corrected for better comprehensibility.

A total of 27 participants complied with the call to respond to the survey and completed the questionnaire. Due to the recruitment setting, all participants were researchers who had already collaborated with biobanks and had experience in requesting samples. Thus, they represented typical end users of the Sample Locator. There were 6 persons aged 25 to 34 years, 18 persons aged 35 to 50 years, and 3 persons older than 50 years. Gender was not specifically surveyed in the study.

The third and final prototype was modified in accordance with the responses of the usability analysis with researchers. Figure 2 illustrates the initial search page with the basic functions of the prototype’s third version. The search tool is divided into 2 sections. Search parameters can be selected in the left area of the screen. A distinction is made between sample-related criteria and donor-related criteria to obtain the most specific query possible. At this point, a help function has been added, as requested, which assists the user in finding the search parameters under the respective subheadings. Once the search criteria have been selected (eg, sex=male, as seen in Figure 3), the search can be performed by clicking the “Search” button. Here, a help function can now also be accessed to inform users about the format that can be searched for. A further implementation based on the test feedback when selecting the search parameters is the more detailed specification of the diagnostic codes. In addition to the simple code, the textual description has been added. In addition, the “direct entry of the ICD code (if already known) for input” is now possible via a free text field, as desired (Figure 4). The results will then be displayed in the right area of the screen. The feasibility query merely produces an aggregated number of potential samples and donor matches. As soon as the user has logged in to the tool in the next step, the allocation of suitable samples per biobank can be viewed (see Figure 5). In addition to an example query with detailed results for biobanks, Figure 5 also shows further functions. On the left “Search” side, the “Clear” button is used to completely restart a search, while the “Edit” button can be used to adjust the previous search. In order to have a better overview of which section you can currently interact with, a feature has been added that grays out and prevents editing within the inactive page. After the search has been carried out, one or more relevant biobanks can be selected on the right-hand side and contacted using the “Negotiate” button. The individual queries are saved and stored in a succeeding component. These can then be viewed and managed via the “Project Overview“ button. In addition, general design changes have been introduced (eg, adjusting the font size and color to improve readability). In view of the intended international use, the content was also translated into English, per recommendation. For extensive screenshots of the Sample Locator prototype, see Multimedia Appendix 5.

The main goal of our work was to evaluate the usability of a federated search tool for biosamples and associated data, the Sample Locator. This paper also illustrates the work that was accomplished in advance of the usability analysis. However, the evaluation of the tool by the end users was of particular importance. The predominantly positive feedback of the usability analysis indicated an intuitive and user-friendly operability. The answers to the open questions provided useful advice on optimization potential.

In the area of medical research, especially with regard to cohort identification, several search tools are available. Commercial representatives include, among others, TriNetX [18], BC Platforms [19], and Clinerion [20]. Platforms or projects that make their work available as open source resources are i2b2 [21], tranSMART [22], and ConQuery [23]. The German Centre for Cardiovascular Research provides an in-house development for the feasibility query of biospecimens and data called the Feasibility Explorer [24]. The German Cancer Consortium (Deutsches Konsortium für Translationale Krebsforschung [DKTK]) has also implemented an in-house development (Searchbroker) as part of their comprehensive “bridgehead” architecture [8]. After reviewing these already existing tools, the deliberate decision was made to build on the DKTK codebase and their general architecture, but develop a customized, proprietary GBA Sample Locator solution. There were several reasons for this decision. Due to their developer power, industrial suppliers may have several advantages, particularly in the areas of speed and user interface design. However, especially when dealing with sensitive patient data, a solution that would be made available as an open source tool at the end of the project was preferred. In this case, all components can be developed, operated, and sustained by academic players, safeguarding the data sovereignty of biobanks. In this context, the Sample Locator, unlike tranSMART and the Feasibility Locator, follows a decentralized search approach. Consequently, the data remain locally at the biobank location, but can be queried centrally. Furthermore, we consider this open source approach to be a sensible way to ensure the continued existence of the Sample Locator beyond the end of the project [25]. In terms of further development, support, and bug fixing, the community resulting from this project (namely the Samply Community [26]) can continue to make a valuable contribution in the future. An additional functionality that distinguishes the Sample Locator from the search applications under review is that it limits the results visualization to the aggregated sample count from the whole GBA network for anonymous researchers. Only after a researcher’s authentication and log-in, the counts per biobank location are additionally displayed [6]. Further, after the search, the Sample Locator forwards the user directly to the Negotiator, a dedicated communication tool to help request material from relevant biobanks and negotiate the terms involved based on the actual planned research project. Thus, based on the considerations outlined above, none of the existing tools were suitable for direct application and deployment in GBA. However, the various approaches provided valuable input for GBA’s own search tool development in our incremental user-centered approach.

The current lack of a suitable tool for a national search for biosamples prompted GBA to strive to close this gap. Although the initial focus was to provide a technically functioning development version at an early stage of the project, ensuring the user-friendliness of the tool should not be neglected in the process. To meet this demand, a usability analysis within the framework of a UCD process, which was also the basis for the accompanying preparatory work, was conducted. The findings of this usability analysis indicate that the considerations regarding a user-friendly application that have been made in the development process so far strongly coincide with the perception of the study participants. Despite this overlap, it was important to address the steps of the UCD process carried out in the context of this work and to engage prospective end users to ensure that the previous development was going in the desired direction and that the Sample Locator will be used in the future. Moreover, the users’ comments and suggestions for improvement that were received through the open questions will be considered in upcoming iterations to refine the Sample Locator. In this way, we are confident in the delivery of a final product that offers added technical value but is also intuitive for the user. These results will also have a positive impact on the aspired Europe-wide rollout. In Europe, the Biobanking and Biomolecular Resources Research Infrastructure – European Research Infrastructure Consortium Directory is currently commonly used as an online search tool, listing 720 European biobanks with their 1621 individual collections [33]. The Directory provides a detailed overview of the collections’ contents and enables users to contact a collection’s principal investigator. However, the Sample Locator would offer added value on a more granular level, enabling the search for specific samples and the contacting of the respective biobanks. While the biobank of Masaryk University in Brno is already connected, the further integration of European biobanks is planned.