0.02% of US Citizens Have Sickle Cell Disease: Here’s How We Balance the Data

In the United States, only 0.02% of the population, or about 100,000 out of 334.9 million people, are diagnosed with sickle cell disease. This creates a significant class imbalance in datasets used for medical classification problems. When predicting diseases like sickle cell, where abnormal hemoglobin levels (6-11 g/dL) are a strong indicator, models struggle to identify meaningful patterns due to the rarity of the condition. To address this, the Synthetic Minority Oversampling Technique (SMOTE) is employed. SMOTE generates synthetic samples of the minority class to balance the dataset, allowing machine learning models to better learn from and predict rare events.

Source: towardsdatascience.com

Related Videos

Key Findings from the Fatigue in Sickle Cell Disease project

Sickle Cell Disease: Challenges From the Payer Perspective

Gene Correction in Sickle Cell Anemia – Mark Walters

National Education Session: Management of Pregnancy in Sickle Cell Disease

Management of Sickle Cell Disease in 2023

Global Epidemiology of Sickle Cell and Newborn screening with Point of care testing