Design and Implementation of Low-cost Hemoglobin Level Prediction using Machine Learning

Emerging nations have an especially pressing need for a low-cost haemoglobin measurement solution, but this demand is global. The mobile phone, which is widely used in both affluent and impoverished regions, is an excellent candidate for this remedy. Non-invasive quantification of haemoglobin levels is proposed in this Project. Methods of collecting data, theoretical underpinnings, the retrieval of photoplethysmogram (PPG) signals and functionalities, machine-learning computation, and modeling techniques for determining haemoglobin levels were also especially in comparison. Based on this evaluation, practical methods for developing a mobile phone point-of-care device for non - intrusive haemoglobin assessment were proposed. The overarching objectives of this study were about (Aim 1) test the viability of creating a technique to analyse smartphone image data of fingernails in order to estimate haemoglobin (Hgb) levels, (Aim 2) create an anaemia predictive model and legitimise it with a sizable health diagnosis (
$\mathrm{n}=337$
), and (Aim 3) conduct testing process and integrating this automated system into a translational, user-friendly, and low-cost smartphone app. Self-Hgb measurement by individuals with persistent anaemia using this image processing algorithm implemented in a mobile platform as just an app improves access to medical care generally and in reduced settings in particular.