JMIR Publications

ABSTRACT

Background:

Permanent pacemaker (PPM) implantation is a core treatment for brady arrhythmias. In China, the number of PPM implants reached 158,000 in 2023 and continues to rise. However, long-term postoperative management is crucial for ensuring favorable outcomes. Patients after PPM implantation face risks of complications and physiological and psychological changes associated with the device. These subjective experiences directly affect quality of life and treatment adherence, making them central issues in postoperative care. Current research primarily relies on clinical follow-ups and questionnaires, which can capture objective indicators but fail to fully address patients’ subjective experiences, limiting the precision of nursing interventions in meeting core patient needs. Temporal and spatial constraints in traditional settings also hinder timely and continuous data collection, making it difficult to track evolving patient needs. Social media platforms such as Zhihu offer new avenues for capturing authentic patient experiences, generating vast amounts of unstructured health data that complement traditional methods. However, manual analysis struggles to process this volume of data effectively, preventing systematic identification of shifts in patient concerns and psychological adaptation mechanisms. Natural language processing (NLP) technology offers a promising solution by enabling deep mining of unstructured data, accurately identifying patients’ emotions, key concerns, and potential risks, and revealing patterns of dynamic change. This study leverages NLP to analyze self-reported texts from PPM patients on Zhihu, exploring the evolution of their concerns and psychological adaptation mechanisms, thereby providing data support for developing lifelong care strategies. By integrating theories of self-management, impression management, and chronic disease management, we propose a phased, precision-based nursing model that advances patient-centered care. This approach offers a practical pathway for clinical translation of digital health data, fills gaps left by conventional methods, and provides novel insights for data-driven decision-making.

Objective:

This study aims to identify latent needs of pacemaker patients through natural language processing (NLP) mining of social media data. By addressing the limitations of traditional questionnaires in terms of timeliness and continuity, we seek to provide robust data support for precision medicine and health management.

Methods:

Text data related to pacemakers were collected from the Zhihu platform and subsequently screened, resulting in 2,424 valid records post quality control. Techniques employed included Latent Dirichlet Allocation (LDA) topic modeling, sentiment analysis, t-distributed Stochastic Neighbor Embedding (t-SNE) dimensionality reduction, and keyword co-occurrence network analysis.

Results:

Five primary themes emerged: core medical care, device technology, consultation interaction, surgical rehabilitation, and life recording. Sentiment evolution indicated an initial predominance of negative sentiment, transitioning into a complex structure with stable negative and increasing positive sentiment during the middle period, and culminating in a recent significant surge in positive sentiment that surpassed negative sentiment. Fine-grained analysis revealed that scores for worry, trust, and financial pressure remained below 20 across all groups. Network co-occurrence analysis identified three major clusters centered on "heart," "implantation," and "physician," with "heart" serving as the most central hub term.

Conclusions:

Utilizing 11 years of longitudinal data, this study uncovers non-linear development trajectories in the needs of post-permanent pacemaker (PPM) implantation patients. In conjunction with self-management theory, impression management theory, and chronic disease management theory, a phased precision nursing intervention model was developed, offering an actionable pathway for translating digital health data into clinical practice.