Artificial Intelligence (AI) Based Risk Prediction System for Sickle Cell Anemia using Clinical Data

Abstract

SCA, or sickle cell anemia, is among the most prevalent hereditary blood disorders globally, and is most prevalent in the Middle East, Africa, and India. Early diagnosis and accurate risk classification will be critical to avert potentially fatal events in patients with SCA. For this purpose, we present an Artificial Intelligence (AI) based risk prediction system, which makes predictions using clinical data. We developed a custom target column that classified risk as low, medium, or high, using the clinical parameter thresholds defined from a Kaggle dataset. After evaluating several machine learning classifiers, With an accuracy of 94.45% (F1 = 0.95, AUC = 0.95). The Random Forest classifier outperformed the others. After training the model, we wrapped it in a Streamlit app that enables real-time prediction, data visualization through graphics, and preparation of automated PDF reports. Evidence from the experiments indicated that the system has the potential to help medical professionals monitor their patients and conduct early screening for SCA.

Ähnliche Arbeiten