Deep Learning Detects Coronavirus
Chinese researches in the Hubei Province just published their work using deep learning, a branch of machine learning to detect the 2019 novel coronavirus or Covid-19. A team of nearly 20 scientists from The National Natural Science Foundation of China have been using high resolution images of patients' lungs that contain the virus to train their deep learning to diagnose the virus with an accuracy of about 95%.
An expert radiologist would take nearly 2 minutes to read an image of a patients lungs with the coronavirus. However, the Chinese team's deep learning took 41 seconds per patient.
If a phenomenal outbreak occurs throughout the world and healthcare infrastructure is pushed to its maximum capacity, this could prove to be a huge breakthrough for the medical community. Not only is the diagnosis time cut down tremendously, but the ability to use a deep learning to make diagnoses and frees up radiologists to attend to patients makes for a much more efficient system.
Computed topography is the method most radiologists will use for examining and diagnosing a patients coronavirus condition. The Chinese team trained their deep learning models on over 21,000 images of coronavirus infected patients' lungs along with over 5,000 images of lungs that did not contain the virus. However, one negative fact to note about the research is the very small sample set. Despite having thousands of images to train their models, these images are derived from just 40 total patients. 27 with the coronavirus and 13 without.
Deep learning is perfectly situated to detect the coronavirus and other medical ailments. The branch of machine learning is trained to identify images from unsupervised learning of those images. Dr. Andrew Ng, a famous Ai researcher from Stanford is the first person to prove that deep learning works. In 2009, the professor fed his deep learning model 4 million images of cats. The model was then able to accurately detect images with cats, despite having no direct human coding to do so. Deep learning is an open-ended technology that can be applied to a vast array of potential data sets.
The Chinese government has come under fire from the rest of their world regarding their handling of the coronavirus. At first China ignored the whistleblowers in their health community, going so far as threatening arrest to arrest the first doctor who found the coronavirus. Then China did not quarantine Wuhan and reports of some 5 million citizens leaving the city after the infection started spreading put fear into the whole world.
“The danger for Xi Jinping is that as the virus spreads globally, the role that China’s system of governance played in delaying a timely response will face growing scrutiny and criticism from the international community,” Elizabeth C. Economy, a senior fellow and director of Asia studies at the Council on Foreign Relations, told the New York Times.
China is trying to deflect this criticism by focusing attention on their ability to build multiple dedicated hospitals to care for their virus-infected population or the immediate government funding of scientific groups like this one to research clinical methods to work with the coronavirus. President Xi has already released a book in 6 languages showcasing his response to the outbreak.
Government spin aside, this study is a positive step towards combating a virus that has petrified the world and caused the fastest 10% correction in the history of the S&P 500.