Diagnosis of sickle cell anemia using AutoML on UV-Vis absorbance spectroscopy data

TL;DR

This paper presents an AutoML approach to classify UV-Vis spectroscopy data for early, cost-effective diagnosis of sickle cell anemia and trait, achieving high accuracy and suitability for resource-limited screening programs.

Contribution

It introduces a novel AutoML-based method for analyzing UV-Vis spectra to diagnose sickle cell conditions with high sensitivity and specificity.

Findings

Achieved 100% sensitivity in detecting sickle hemoglobin.

Achieved 93.84% specificity in classification.

Demonstrated potential for mass screening in resource-limited areas.

Abstract

Sickle cell anemia is a genetic disorder that is widespread in many regions of the world. Early diagnosis through screening and preventive treatments are known to reduce mortality in the case of sickle cell disease (SCD). In addition, the screening of individuals with the largely asymptomatic condition of sickle cell trait (SCT) is necessary to curtail the genetic propagation of the disease. However, the cost and complexity of conventional diagnostic methods limit the feasibility of early diagnosis of SCD and SCT in resource-limited areas worldwide. Recently, our group developed a low-cost UV-Vis absorbance spectroscopy based diagnostic test for SCD and SCT. Here, we propose an AutoML based approach to classify the raw spectra data obtained from the developed UV-Vis spectroscopy technique with high accuracy. The proposed approach can detect the presence of sickle hemoglobin with 100% sensitivity and 93.84% specificity. This study demonstrates the potential utility of the machine learning-based absorbance spectroscopy test for deployment in mass screening programs in resource-limited settings.