藤野 正之

Abstract Heart rate variability (HRV) is a well-established, noninvasive measure of autonomic nervous system activity and is associated with clinical outcomes. Although real-time monitoring of HRV is valuable in clinical practice, its effectiveness is often compromised by major challenges: high inter-individual variability and frequent data contamination from procedural artifacts. To address these challenges, we developed and validated a computational framework for robust and personalized real-time HRV analysis oriented toward clinical application. The framework performs simultaneous analysis and visualization of both time- and frequency-domain HRV indices and incorporates an adaptive alert algorithm that personalizes alert thresholds using the interquartile range of each patient’s own data. A workflow-integrated mechanism for manually annotating and excluding artifact-prone periods prevents procedural artifacts from skewing the statistical baselines, and a multi-scale visualization module provides a unified view of short-term fluctuations and long-term trends. While existing HRV tools are powerful for research or offline analysis, they often lack the integration of personalized alerting and workflow-oriented artifact management needed for bedside care. The proposed system uniquely combines personalized alerting, care-linked artifact exclusion, and multi-scale bedside visualization within a single real-time software package. The framework was validated using open-access electrocardiogram (ECG) databases and synthetic noise-contaminated signals, confirming robust R-wave detection across pediatric and adult recordings and under low signal-to-noise conditions. In addition, the framework was operationally validated at the bedside using ECG data from 24 newborn patients. By systematically addressing the core challenges of personalization and artifact management in a clinically integrated manner, this work represents a significant step toward translating real-time HRV analysis into routine vital sign management and, ultimately, improved patient outcomes.