Page 115 - AIH-2-4

Artificial Intelligence in Health Autonomic nervous system patterns in men

activity, with implications for both physical and mental
well-being. Reduced HRV is generally associated with a
49
higher risk of CVD and physiological stress, 3,50,51 whereas
elevated HRV is typically indicative of good health. 38,52-54 The
advent of wearable technology has made HRV monitoring
more practical, enabling early detection of cardiovascular
dysfunction and supporting proactive health management
strategies. However, to date, no studies have examined the
55
use of combined PCA and K-means clustering techniques
to identify individual risk profiles based on HRV data.
Pasquini et al. applied PCA to identify five PCs in
54
ANS time series data that collectively explained 75% of
the variance during emotional reactivity tasks. K-means
clustering revealed five distinct ANS states corresponding
to specific emotions, such as awe and sadness.
Figure 3. Silhouette plot for identifying the optimal number of clusters However, the present study validates a methodological
framework integrating PCA and K-means to identify
distinct patterns of autonomic regulation in healthy men
using HRV data. This approach demonstrates variability
in autonomic balance, even within a healthy population.
The findings highlight the potential of these techniques
to stratify individuals based on cardiovascular autonomic
function and suggest that lower HRV profiles may indicate
individuals who could benefit from targeted interventions
to improve autonomic balance.
A critical component of the analysis involves selecting
the optimal number of clusters. Notably, both the elbow
method and hierarchical dendrogram support k = 3 as a
plausible solution, whereas the silhouette index favors a
more conservative k = 2 solution. Rather than interpreting
this discrepancy as a contradiction, it provides valuable
insights into the hierarchical structure of the data. The
higher silhouette score for k = 2 indicates the presence of
Figure 4. Hierarchical cluster analysis of the heart rate variability dataset two primary, well-defined groups. However, selecting k = 3,
as recommended by the elbow method, provides greater
methods, highlighting a nested and interpretable structure physiological granularity by revealing an “intermediate”
within the dataset.
group that would otherwise remain undetected in a binary
5. Discussion classification. This methodological approach allows for a
more detailed and clinically interpretable classification of
The primary aim of this study is to develop and validate autonomic regulation.
a method for identifying distinct patterns of autonomic
regulation in healthy men, using a combined PCA and In addition, these findings offer promising potential
K-means clustering approach applied to HRV data. The for clinical application. The ability to stratify healthy
individuals into potential autonomic “risk” profiles provides
findings demonstrate that this approach is both feasible
and effective in stratifying cardiovascular autonomic insights into the development of personalized preventive
function, revealing significant physiological heterogeneity strategies. A post hoc analysis of the three-cluster solution
within a homogeneous and healthy sample. identified distinct, hierarchically ordered autonomic
profiles. Cluster 3 (high-vagal-tone profile; n = 33) exhibits
HRV is an important biomarker for assessing the highest values across all HRV parameters, indicating
cardiovascular health, serving as a non-invasive indicator of enhanced parasympathetic modulation. This profile is
48
the ANS and its modulation of cardiac function. It reflects typically associated with good cardiovascular health,
the balance between sympathetic and parasympathetic greater stress resilience, and efficient physical conditioning.

Volume 2 Issue 4 (2025) 109 doi: 10.36922/AIH025050006