Image credit: Accrad
Healthcare tech companies are employing artificial intelligence to advance illness detection and prevention. The pandemic pushed development into overdrive, and the results are very promising.
Companies are creating and adapting AI-based diagnostic tools, adding remote and contactless capabilities to accommodate COVID-19 restrictions. At the same time, many added COVID-19 detection capabilities in an effort to stem the spread of the novel coronavirus. The new technologies help reduce backups due to limited staff and enable doctors to treat patients more quickly.
A New Spin on Xray Vision
In Kenya, for example, radiologists are in short supply and high demand, with only 200 radiologists available to meet the needs of the country’s 43 million people. Medical software company Accrad, based in South Africa, has developed CheXRad, a chest X-ray solution that uses artificial intelligence to diagnose at least 15 pulmonary diseases.
CheXRad is an Intel-based desktop application designed to help radiologists manage workflow and diagnose patients quickly before the disease progresses or spreads to others. It uses machine learning to predict which patients might need medication or a ventilator and highlights areas of concern for additional doctor review. The software can accurately identify and label pathologies in chest X-rays in just 90 seconds, 160 times faster than a human radiologist.
Accrad developed the CheXRad system using the Intel® DevCloud with its access to the Intel® AI Analytics Toolkit and the Intel® Distribution of OpenVINO™ toolkit. The machine-learning algorithms are set to diagnose COVID-19 and viral pneumonia, but new learning models can be developed using Intel’s Optimization for TensorFlow in the AI Analytics Toolkit.
Image credit: Accrad
Sniffing Out Disease
Other technologies are emerging to diagnose respiratory ailments as well. Researchers at Moscow's Skolkovo Institute of Science and Technology (Skoltech) are testing an “electronic nose” that can diagnose chronic obstructive pulmonary disease (COPD). The non-invasive tool uses an array of sensors to identify the presence of gaseous compounds in exhaled human breath. In tests, the breath analysis took about three minutes.
The Skoltech researchers used additive technology to print nanocrystalline films of eight metal oxides onto a multielectrode chip. Using machine learning pattern recognition algorithms, the e-nose was able to sniff out separate but chemically similar alcohol vapors in thin concentrations. The researchers claim the algorithms could be altered to identify gases that indicate lung cancer, kidney disease, diabetes, and cystic fibrosis.
All Ears for AI
Researchers from MIT have created AI software that can detect whether a person has COVID-19 by “listening” to a cough. MIT used more than 70,000 samples of spoken words and forced coughs from healthy and infected subjects to train the machine-learning algorithms. The MIT study found that the AI model was able to correctly identify 98.5 percent of people with symptomatic COVID-19 cases, and 100 percent of people with asymptomatic cases.
The MIT team is working to integrate the technology in a mobile app, which could be used as a free, non-invasive diagnostic tool, if approved by the Food and Drug Administration. The technology was initially designed to detect the presence of pneumonia, asthma, and even Alzheimer’s, which is characterized by memory decline and weakened neuromuscular functionality, including diminished vocal cords.
With the expanded use of AI, the healthcare industry can expedite the diagnosis and treatment of an increasing number of illnesses.