Symptom-Based Machine Learning Approach for Early COVID-19 Detection

Since the global spread of COVID-19 began in November 2019, it has become a widely searched term online. COVID-19 remains a severe public health issue requiring ongoing efforts until a permanent solution is established. Asymptomatic individuals pose a significant risk as they can unknowingly transmit the disease and may suffer from severe lung damage. To combat the disease and save lives, accurately predicting and monitoring asymptomatic patients is crucial. This research introduces a machine learning model to detect symptomatic patients based on data such as fever, blood oxygen levels, heart rate, cough intensity, and motion activities. A deep dense neural network (DNN) model was developed to identify COVID-19 infections, achieving high performance with 98% accuracy and a loss value of 0.1 on both training and validation datasets. Additional improvements could be attained by extending training duration or adjusting the learning rate. This AI-based solution demonstrates promise in accurately and efficiently detecting COVID-19 infection using input data such as fever, SpO2 percentage, cough intensity, and heart rate. The proposed model can be particularly beneficial in hospital settings, especially during quarantine periods.