AI has huge potential for analyzing vast amounts of data quickly, an attribute that can have a big impact in a situation of the ongoing COVID-19 pandemic.
FREMONT, CA: For patients with COVID-19, dreadful shortness of breath can set in virtually overnight. In several cases, it is caused by an aggressive infection in the lungs. Detecting these severe cases early on is essential for treating them successfully. At present, the only way to tell whether the coronavirus causes a patient's illness is by conducting a test. However, this test takes up to 2 days to complete. Serial testing may be needed to rule out the chance of false-negative results. There is a shortage of RT–PCR test kits, underscoring the urgent need for alternative ways for rapid and accurate diagnosis of patients with COVID-19.
AI can provide a method to augment the early detection of coronavirus infection. Researchers are designing an AI model that can detect infection based on initial chest CT scans and associated clinical data that could rapidly identify COVID-19 patients in the early stage. The collected chest CT scans and corresponding clinical information obtained at patient presentation is helping in accomplishing this. Clinical data includes travel and exposure history, leukocyte counts, symptomatology, patient age, and patient sex.
The deep convolutional neural network to learn patients' imaging characteristics with COVID-19 on the initial CT scan combined with support vector machine (SVM), random forest, and multilayer perceptron (MLP) classifiers can classify patients with COVID-19 according to clinical data. MLP marked the best performance on the tuning set. This means that a neural network model combining radiological data and clinical information can predict COVID-19 status.
Using AI and machine learning techniques, investigators can categorize patients into mild and severe groups based on clinical data, allowing them to unearth more information about the virus's impact. In the future, AI could help researchers discover disease progression across different populations, including patients with chronic lung conditions and long-term smokers. Coupled with a growing body of tools, AI has the potential to expand the role of chest imaging beyond diagnosis to empower risk stratification, treatment monitoring, and discovery of new therapeutic targets in this race to contain and treat COVID-19.