Smartphone-based non-invasive haemoglobin level estimation by analyzing nail pallor

Blood haemoglobin concentration is a vital physiological parameter for assessing disease conditions, such as Anaemia in the human body. Anaemia is one of the most common diseases that affects billions of people worldwide. Especially, in developing countries, like India, the prevalence is quite high according to WHO statistics. Traditional invasive process for screening of anaemia is exorbitant and is hard to orchestrate in a resource-poor environment which raises the demand for cost-effective, reliable, portable, non-invasive solution. Hence, in this work, we aim to design a solution for non-invasive screening of anaemia by combining the cutting-edge Artificial-Intelligence techniques with the conventional practises for rough assessment of anaemia by observing pallor in the finger-nail of hand of a human subject. The proposed method first induces color changes in nail-bed by application of suitable amount of pressure and release thereafter through a customized hardware device. Finally, the captured video by a smartphone camera is analysed to measure the rate of change of color to quantify the blood haemoglobin level after correlating with the clinically determined value of blood haemoglobin. The proposed model with a fusion-based prediction approach outperforms the state-of-the-art solutions, ensuring a mean RMSE and MSE error of 0.63 and 0.61 and standard deviation +/- 0.46 g dL-1 and +/- 0.73 g dL-1 respectively as compared to clinically tested haemoglobin level for a total of 220 subjects.