# Clinical validation of comprehensive time- and frequency-domain photoplethysmography features from a single-sensor system for non-invasive assessment of vascular load and systolic blood pressure

## TL;DR

This study shows how a single-sensor PPG device can detect subtle vascular changes linked to high blood pressure by analyzing waveform and frequency patterns.

## Contribution

A novel single-sensor PPG system is validated for non-invasive vascular load and systolic blood pressure assessment using combined time- and frequency-domain features.

## Key findings

- Waveform sharpness indices and harmonic ratios decline progressively with increasing systolic blood pressure, indicating vascular stiffening.
- Time-domain features like prolonged systolic phases and higher Ts/Td ratios correlate with elevated blood pressure levels.
- Effect sizes for waveform sharpness were the most pronounced, suggesting strong clinical relevance for hypertension monitoring.

## Abstract

Hypertension remains a silent yet powerful driver of cardiovascular disease—one of the world’s leading causes of death. Despite being treatable, it often goes undetected until complications arise. Photoplethysmography (PPG), a low-cost, non-invasive tool, holds promise for identifying vascular changes linked to high blood pressure. However, current single-site methods primarily focus on time-domain signals, often missing rich spectral data that could enhance early detection of hypertensive changes.

We designed a compact, noise-resistant PPG acquisition system (Taiwan Patent No. 114206342) aimed at real-world usability and recruited 590 adults from cardiology outpatient clinics. Using a validated oscillometric device, systolic blood pressure (SBP) was measured and categorized into four groups (≤120, 121–139, 140–159, ≥160 mmHg). Ninety-second fingertip PPG recordings (500 Hz) underwent careful pre-processing—filtering, normalization, and artifact rejection. We extracted 19 features covering waveform shape, sharpness, heart rate variability (HRV), and frequency-domain harmonic ratios. Group differences were assessed using ANOVA or Kruskal–Wallis tests, with post hoc analyses and effect size reporting (epsilon squared, ε2).

Several features demonstrated strong associations with SBP levels. Waveform sharpness indices (1_10, 1_8, 1_6, 1_5, 1_3, 1_2) and harmonic ratios (H2/H1, H3/H1, H4/H1) showed a progressive decline as SBP increased—signaling a loss of higher-frequency components likely due to vascular stiffening. Time-domain features revealed that individuals with elevated SBP had prolonged systolic phases, shorter diastolic intervals, and higher Ts/Td ratios. Peak amplitudes (P1, P2), systolic and diastolic slopes, and overall waveform area also differed significantly by group. Effect sizes ranged from small to moderate, with the most pronounced changes seen in waveform sharpness.

Our results demonstrate the promise of a single-sensor PPG-based approach for monitoring hypertension-related, subtle vascular alterations by combining morphological and spectral information. Progressive decline in harmonic ratios provides new information on arterial stiffening and changed wave reflections. This level of physiological depth combined with a remarkably simple and portable approach is the key that could allow wearable devices to provide immediate blood pressure screening for unprecedented widespread user engagement. These aspects are facilitated by making the test easier to operate than standard PCR and combining it with types of analyte detection that expand its performance capabilities in this specific disease context.