Khalifa University Unveils Four-Wavelength Photoplethysmography Dataset for Non-Invasive Hemoglobin Monitoring in Nature

🔬 Breakthrough in Non-Invasive Health Monitoring: Khalifa University's Hb-PPG Dataset

Khalifa University of Science and Technology in Abu Dhabi has made significant strides in biomedical engineering with the release of the Hb-PPG dataset, published in the prestigious Scientific Data journal, part of the Nature portfolio. This four-wavelength photoplethysmography (PPG) dataset represents a pivotal resource for advancing non-invasive hemoglobin monitoring, a critical tool for detecting anemia and assessing cardiovascular health without the need for painful blood draws. Photoplethysmography, commonly abbreviated as PPG, is an optical technique that measures blood volume changes in peripheral tissues by detecting light absorption and reflection variations caused by arterial blood flow. Traditional hemoglobin (Hb) measurement relies on invasive venous or capillary blood sampling using hematology analyzers, which, while accurate, poses challenges in resource-limited settings, frequent monitoring scenarios, or remote areas.

The Hb-PPG dataset addresses these limitations by providing high-quality, multi-wavelength PPG signals paired with clinical reference data, enabling researchers worldwide to develop and benchmark machine learning models for accurate, real-time Hb estimation. Collected from 252 adult subjects aged 21 to 90 years, the dataset includes 1,008 PPG segments captured at four strategically chosen wavelengths: 660 nm (red), 730 nm (deep red), 850 nm (near-infrared), and 940 nm (infrared). These wavelengths were selected based on hemoglobin's distinct absorption spectra and tissue penetration depths, allowing for depth-resolved signal analysis to mitigate motion artifacts and improve estimation precision.

Understanding Photoplethysmography and Its Evolution

PPG technology, first introduced in the 1930s, has evolved dramatically with advancements in light-emitting diodes (LEDs) and photodetectors, powering features in smartwatches like heart rate monitoring. In single-wavelength PPG, light at one frequency (typically green or infrared) illuminates the skin, and reflected light is detected to capture pulsatile blood flow. However, for complex parameters like hemoglobin concentration—which varies with oxygenated (oxy-Hb) and deoxygenated (deoxy-Hb) forms—multi-wavelength approaches are essential. Different wavelengths interact uniquely with blood chromophores: shorter wavelengths (e.g., 660 nm) are absorbed more by deoxy-Hb, while longer ones (e.g., 940 nm) penetrate deeper with less scattering.

Research reviews highlight that multi-wavelength PPG outperforms single-wavelength methods by providing ratio-based features (e.g., AC/DC ratios across wavelengths) that correlate strongly with Hb levels, achieving mean absolute errors as low as 1 g/dL in controlled settings. Yet, the scarcity of public datasets has hindered progress. Prior works often used two or three wavelengths, but four-wavelength configurations, as in Hb-PPG, offer richer physiological insights, capturing contributions from superficial and deeper vascular beds.

• 660 nm: High sensitivity to deoxy-Hb, shallow penetration.
• 730 nm: Bridges visible and NIR for balanced absorption.
• 850 nm: NIR for motion robustness and oxy-Hb sensitivity.
• 940 nm: Deepest penetration, isosbestic point minimizing oxygenation effects.

The Hb-PPG Dataset: Composition and Acquisition Process

Data collection occurred at Guilin University of Electronic Technology's facilities, with Mohamed Elgendi from Khalifa University's Department of Biomedical Engineering overseeing processing and validation. Subjects rested for 5 minutes before fingertip PPG measurements using a custom reflectance-mode sensor. Each session yielded four 30-second PPG recordings per subject, denoised via empirical mode decomposition (EMD) and assessed for quality using signal quality index (SQI) metrics like skewness and kurtosis. Reference Hb was measured invasively via capillary blood analyzed on a Mindray BC-6000 hematology analyzer, alongside fasting glucose and brachial blood pressure.

The resulting dataset spans Hb concentrations from low (anemic) to normal/high ranges, ensuring representation across clinical thresholds (e.g., WHO anemia cutoff: <13 g/dL men, <12 g/dL women). De-identified CSV files include raw/filtered PPG waveforms (500 Hz sampling), timestamps, and metadata. Publicly hosted on Figshare (download here), with GitHub code for SNR calculation, statistical analysis, and baseline models (GitHub repo).

Addressing Anemia: Global Burden and UAE Relevance

Anemia remains a public health crisis, impacting 1.92 billion people globally per WHO 2019 estimates—40% of children aged 6-59 months, 37% of pregnant women, and 30% of women 15-49 years. In low/middle-income countries, prevalence exceeds 50% for non-pregnant women. In the UAE, anemia affects 28.2% of reproductive-age women (2023 World Bank data) and 20.4% of children under 5, linked to nutritional deficiencies, chronic diseases, and expatriate demographics.

Non-invasive tools like Hb-PPG could transform screening in UAE's diverse population, supporting Vision 2031 health goals. Frequent Hb checks are vital for at-risk groups (e.g., laborers, pregnant expatriates), where invasive tests deter compliance. Integrating PPG into wearables aligns with UAE's digital health push, as seen in Abu Dhabi's smart city initiatives.

Technical Advantages of Four-Wavelength PPG Over Existing Methods

Prior studies used dual/triple wavelengths, but four-wavelength setups reduce errors from skin tone, motion, and perfusion variations. Reviews show multi-wavelength ratios (e.g., R660/R940) yield Pearson correlations >0.8 with Hb. Hb-PPG's inclusion of hemodynamic references (BP, glucose) enables holistic models, addressing confounders like diabetes (prevalent in UAE at 12.3%). Compared to datasets like MAX30102-based ones (two wavelengths, smaller N=35-127), Hb-PPG's scale and diversity stand out.

• Improved depth selectivity minimizes fat/muscle interference.
• ML-ready: Supports CNNs, transformers for waveform analysis.
• Benchmark potential: GitHub baselines for MAE/RMSE evaluation.

Khalifa University's Biomedical Engineering Excellence

Funded by Khalifa University grant FSU-2025-001, this work underscores KU's leadership in healthcare innovation. Ranked top in UAE for engineering (QS 2026), KU's Healthcare Engineering Innovation Group, led by Prof. Mohamed Elgendi, pioneers PPG applications. Collaborations with Chinese institutions highlight UAE's global research hub status. For aspiring researchers, explore higher ed jobs at KU or similar via UAE academic opportunities.

Applications: From Wearables to Telehealth in UAE

Hb-PPG paves the way for smartwatch Hb monitoring, vital for chronic care. Potential: Anemia alerts for construction workers (UAE's 90% expatriate workforce), maternal health apps. In telemedicine, post-COVID UAE platforms could integrate PPG for remote Hb screening, reducing hospital visits. Studies project wearables could cut anemia diagnosis costs by 70% in field settings.

Challenges, Limitations, and Future Outlook

Challenges include motion artifacts (addressed via SQI) and skin tone bias (Hb-PPG diverse cohort mitigates). Future: Hyperspectral PPG, integration with ECG for multi-vitals. UAE's AI/health tech ecosystem (e.g., MBZUAI partnerships) positions it for leadership. Researchers can extend via federated learning on public datasets.

Discover career advice in biomedical fields at higher ed career advice.

Photo by Usman Yousaf on Unsplash

Conclusion: Empowering Global Health Research

The Hb-PPG dataset from Khalifa University marks a milestone in non-invasive monitoring, freely available for innovation. Download today and contribute to anemia eradication. Explore faculty positions at higher ed faculty jobs, rate professors on Rate My Professor, or seek advice at higher ed career advice. UAE universities like KU offer prime university jobs.