Real-Time Hemoglobin Measurement Using Smartphone Video and Artificial Neural Network

Normally, clinical assessments of hemoglobin are performed by evaluating blood samples collected through venipuncture in laboratories. These techniques are uncomfortable, painful, and costly for patients. A non-invasive, affordable, accurate, and point-of-care Hb test is required everywhere. This research provides a unique non-invasive real-time technique for measuring Hb using a smartphone PPG signal derived from fingertip video. Fingertip recordings were acquired from 88 people utilizing a smartphone's primary camera and a near-infrared illumination source. The video was used to create PPG signals, and a Butterworth bandpass was applied to remove the noise and motion artefacts. As a result, the preprocessed PPG signal and its derivatives yield 34 time-domain characteristics. Finally, artificial neural network model was developed to estimate the Hb level accurately. The suggested approach outperformed in estimating Hb level, with a coefficient of determination (R2) of 0.909, suggesting a strong link between hemoglobin level and retrieved characteristics. Experiments demonstrate the proposed approach can be utilized to measure Hb level without extracting blood.