TextRank-based healthcare big data analysis system for text network analysis and trend phase identification

The demand for scalable healthcare knowledge extraction is increasing. To address this demand, this study developed a TextRank-based big data analysis system focused on reusable, system-level implementation for the automation of keyword extraction, text network analysis, and trend phase identification from large-scale textual data.

The proposed system was developed based on a system engineering framework that included requirement analysis, modular architecture design, and system implementation. Healthcare-related textual documents were collected and preprocessed using domain-specific normalisation and noise-reduction procedures. Keywords were extracted using a graph-based TextRank algorithm, and weighted keyword co-occurrence networks were constructed for each analysis period. Trend phase identification was performed by calculating the weighted Jaccard similarity between consecutive periods and detecting structural changes in keyword distributions. All analytical components were integrated into an end-to-end healthcare big data analysis system with interactive visualisation capabilities.

The proposed system successfully extracted representative keywords, constructed keyword co-occurrence networks, and identified distinct trend phases reflecting changes in healthcare research topics over time. The system automated the analytical workflow from keyword extraction to trend phase identification and provided visualizations to support the exploration and interpretation of healthcare research trends.

The results confirmed that the proposed system provides a scalable, interpretable, and reusable platform for continuous healthcare knowledge monitoring, supporting data-driven research, policy formulation, and strategic decision-making. The automated functions of the system thereby enable researchers to efficiently monitor evolving healthcare topics and discover meaningful insights over time.