Workflow for health-related and brain data lifecycle

Abstract

Poor lifestyle leads potentially to chronic diseases and low-grade physical and mental fitness. However, ahead of time, we can measure and analyze multiple aspects of physical and mental health, such as body parameters, health risk factors, degrees of motivation, and the overall willingness to change the current lifestyle. In conjunction with data representing human brain activity, we can obtain and identify human health problems resulting from a long-term lifestyle more precisely and, where appropriate, improve the quality and length of human life. Currently, brain and physical health-related data are not commonly collected and evaluated together. However, doing that is supposed to be an interesting and viable concept, especially when followed by a more detailed definition and description of their whole processing lifecycle. Moreover, when best practices are used to store, annotate, analyze, and evaluate such data collections, the necessary infrastructure development and more intense cooperation among scientific teams and laboratories are facilitated. This approach also improves the reproducibility of experimental work. As a result, large collections of physical and brain health-related data could provide a robust basis for better interpretation of a person's overall health. This work aims to overview and reflect some best practices used within global communities to ensure the reproducibility of experiments, collected datasets and related workflows. These best practices concern, e.g., data lifecycle models, FAIR principles, and definitions and implementations of terminologies and ontologies. Then, an example of how an automated workflow system could be created to support the collection, annotation, storage, analysis, and publication of findings is shown. The Body in Numbers pilot system, also utilizing software engineering best practices, was developed to implement the concept of such an automated workflow system. It is unique just due to the combination of the processing and evaluation of physical and brain (electrophysiological) data. Its implementation is explored in greater detail, and opportunities to use the gained findings and results throughout various application domains are discussed.

Keywords: best practices; brain data; data lifecycle; health information system; health-related data; ontology; physical data; workflow.

Figure 1

DevOps (16) and DataOps in the enterprise (16).

Figure 2

DevOps and DataOps processes (17).

Figure 3

ResearchOps Community-separated resources for each of the topics. The link to community- made tools that support each aspect of these open topics can be found in the footnote below.

Figure 4

ResearchOps for health-related data.

Figure 5

“Cube” overview: each cube part can be imagined as a box with specific needs, which have to be accounted for and supported by the underlying system.

Figure 6

Body in Numbers terminology—examples of the root terms. Note that tree visualization only contains a few top degrees of the tree. For more details, see the link to the terminology in Section 3.2.2.

Figure 7

Data flow–layer details.

Figure 8

Body in Numbers ontology—the parent node “Device categories” contains links to its child and sibling nodes. Only the subclasses are visualized (without any additional defined properties). The ontology was visualized with WebVOWL: Web-based Visualization of Ontologies, version 1.1.7, available at http://vowl.visualdataweb.org/webvowl.html.

Figure 9

BiN lifecycle—For each of the data flow steps, module was developed that assisted with their respective tasks independently of the remaining architecture.

Figure 10

P300 module—The figure describes each of the steps that will be done automatically by the module, showing also the process inputs and outputs.

Figure 11

Semantic module—the figure describes each of the steps that will be done automatically by the module, showing also the process inputs and outputs.

Figure 12

Statistical module—the figure describes each of the steps that will be done automatically by the module, showing also the process inputs and outputs.

Figure 13

Evaluation module – The figure describes each of the steps that will be done automatically by the module, showing also the process inputs and outputs.

Figure 14

Publishing module—the figure describes each of the steps that will be done automatically by the module, showing also the process inputs and outputs.