Design and Development of a Linked Open Data-Based Health Information Representation and Visualization System: Potentials and Preliminary Evaluation
Bibliographic record
Abstract
BACKGROUND: Healthcare organizations around the world are challenged by pressures to reduce cost, improve coordination and outcome, and provide more with less. This requires effective planning and evidence-based practice by generating important information from available data. Thus, flexible and user-friendly ways to represent, query, and visualize health data becomes increasingly important. International organizations such as the World Health Organization (WHO) regularly publish vital data on priority health topics that can be utilized for public health policy and health service development. However, the data in most portals is displayed in either Excel or PDF formats, which makes information discovery and reuse difficult. Linked Open Data (LOD)-a new Semantic Web set of best practice of standards to publish and link heterogeneous data-can be applied to the representation and management of public level health data to alleviate such challenges. However, the technologies behind building LOD systems and their effectiveness for health data are yet to be assessed. OBJECTIVE: The objective of this study is to evaluate whether Linked Data technologies are potential options for health information representation, visualization, and retrieval systems development and to identify the available tools and methodologies to build Linked Data-based health information systems. METHODS: We used the Resource Description Framework (RDF) for data representation, Fuseki triple store for data storage, and Sgvizler for information visualization. Additionally, we integrated SPARQL query interface for interacting with the data. We primarily use the WHO health observatory dataset to test the system. All the data were represented using RDF and interlinked with other related datasets on the Web of Data using Silk-a link discovery framework for Web of Data. A preliminary usability assessment was conducted following the System Usability Scale (SUS) method. RESULTS: We developed an LOD-based health information representation, querying, and visualization system by using Linked Data tools. We imported more than 20,000 HIV-related data elements on mortality, prevalence, incidence, and related variables, which are freely available from the WHO global health observatory database. Additionally, we automatically linked 5312 data elements from DBpedia, Bio2RDF, and LinkedCT using the Silk framework. The system users can retrieve and visualize health information according to their interests. For users who are not familiar with SPARQL queries, we integrated a Linked Data search engine interface to search and browse the data. We used the system to represent and store the data, facilitating flexible queries and different kinds of visualizations. The preliminary user evaluation score by public health data managers and users was 82 on the SUS usability measurement scale. The need to write queries in the interface was the main reported difficulty of LOD-based systems to the end user. CONCLUSIONS: The system introduced in this article shows that current LOD technologies are a promising alternative to represent heterogeneous health data in a flexible and reusable manner so that they can serve intelligent queries, and ultimately support decision-making. However, the development of advanced text-based search engines is necessary to increase its usability especially for nontechnical users. Further research with large datasets is recommended in the future to unfold the potential of Linked Data and Semantic Web for future health information systems development.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Direct model labels (unvalidated)
Per-model category and study-design labels from the labeling rounds. They are machine output, unvalidated, and the disagreement between models ships as data. No study design here is MEDLINE-validated yet.
| Model arm | Categories | Study design | Confidence |
|---|---|---|---|
| gemma | no category Domain: not available · Genre: Empirical About the Canadian research system: no · About a Canadian topic: no | Bench or experimental | low |
| gpt | no category Domain: not available · Genre: Software About the Canadian research system: no · About a Canadian topic: no | Not applicable | low |
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.010 | 0.013 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.004 | 0.004 |
| Open science | 0.003 | 0.003 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.008 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedLabeled directly by 2 models reading the full record.
The models disagree on parts of this classification; every voice is preserved in the section at the end of the page.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".