Computational Algorithms Could Transform COVID-19 Diagnosis and Care, Finds Study

The study conducted by researchers at the University of Eswatini (Kwaluseni, Eswatini) found that computational algorithms could transform COVID-19 diagnosis and care for low-income countries.

With the emergence of new COVID-19 variants, and infections continuing to rise, effective diagnostic software systems are urgently required to support the increasingly overstretched virus testing services. In addition, these software systems can help to stem the spread of COVID-19 by enabling early identification of new cases. This can be particularly important in low-income countries where medical personnel and facilities are limited.

Computational algorithms can also play a valuable role in identifying cases that traditional clinical diagnosis methods may miss; for example, infections in people with certain underlying diseases. While artificial intelligence models are already available to support diagnosis of COVID-19, most are used in interpreting X-ray image data; and are not always effective in early-stage diagnosis, when the patient’s respiratory and cardiovascular systems may show few signs of the virus.

In the study, the researchers tested the ability of seven computational algorithms to diagnose COVID-19 at an early stage, based on the following common symptoms: fever or chills; cough; shortness of breath or difficulty breathing; fatigue; muscle or body aches; headache; loss of taste or smell; sore throat; congestion or runny nose; nausea or vomiting; and diarrhea. The researchers found that the algorithms Multilayer Perceptron, Fuzzy Cognitive Map and Deep Neural Network outperformed Logistic Regression, Naïve Bayes, Decision Tree and Support Vector Machine. The researchers believe that these findings could guide future software development.

“This information could be adopted to develop intelligence-based software that both medical personnel and patients can use for early diagnosis of COVID-19 when these symptoms are present. At the time we were conducting this research, we could not find any other studies that had applied any of the listed intelligent techniques for COVID-19 diagnosis using these common symptoms,” said Boluwaji A. Akinnuwesi, an Associate Professor in the Department of Computer Science at University of Eswatini. “Using these algorithms is a better option than exposing patients to X-rays, which, in addition, are not always easily accessible. The three best performing algorithms have the potential to be developed into widely-available software, increasing access to quick and affordable diagnosis of COVID-19 infection, which is particularly important for low-income countries, like Africa.”

Related Links:
University of Eswatini