The use of mechanical ventilation can save lives. But the lung is a complex organ, and the amount of pressure necessary to keep all parts of the lung open to airflow can actually cause damage to some parts through overdistention of the tissue. “The real crux of the problem is that when we’re treating a patient using mechanical ventilation, up until now, there hasn’t been any way to detect overdistention of the lung tissue," said Prof. Wolfgang Wall, Professor for Computational Mechanics at the Technical University of Munich (TUM).
Wall and his team have harnessed the computing power of AI to develop a "digital twin” model of the patient’s lungs. The digital tool can “map” a patient’s lungs – and which can even be used for early detection of COVID-19 infections. “More than 80 percent of COVID-19 deaths are the result of acute lung failure,” said Wall. “And with long-term mechanical ventilation, the survival rate for our most critically ill patients drops to only 50 percent,” he added. “The goal of our work is that in the future, at each ventilation site a digital lung model helps to optimize the ventilation to the patient’s needs so that we can significantly increase the chance of survival.”
Researchers at The Korea Advanced Institute of Science and Technology have developed a deep learning-based cough recognition model to classify a coughing sound in real time. The coughing event classification model is combined with a sound camera that visualizes their locations in public places. The research team said they achieved a best test accuracy of 87.4 %. Professor Yong-Hwa Park from the Department of Mechanical Engineering said that it will be useful medical equipment during epidemics in public places such as schools, offices, and restaurants, and to constantly monitor patients’ conditions in a hospital room.
To develop a cough recognition model, a supervised learning was conducted with a convolutional neural network (CNN). The model performs binary classification with an input of a one-second sound profile feature, generating output to be either a cough event or something else.
The German tech company audEERING is in the process of developing an app aimed at detecting coronavirus using the sounds of coughing and sneezing as well as voice recognition. The firm is banking on audio artificial intelligence as a means of containment and improved analysis of the disease.
With the app, users are asked to record audio of themselves coughing, sneezing or speaking. The technology begins to learn from such vocal characteristics, and compares the data to recordings of COVID-19 patients. Age, gender and health condition can all be automatically detected by the recordings, the company claims.
CT scans offer a deep insight into COVID-19 diagnosis and progression as compared to the often-used reverse transcription-polymerase chain reaction. These RT-PCR tests have high false negative rates, delays in processing and other challenges. In addition, CT scans show COVID-19 in people without symptoms, in those who have early symptoms, during the height of the disease and after symptoms resolve. However, CT is not always recommended as a diagnostic tool because the disease often looks similar to influenza-associated pneumonias on the scans. Researchers at University of Central Florida have now demonstrated that an AI algorithm could be trained to classify COVID-19 pneumonia in CT scans with up to 90 percent accuracy, as well as correctly identify positive cases 84 percent of the time and negative cases 93 percent of the time. “We demonstrated that a deep learning-based AI approach can serve as a standardized and objective tool to assist healthcare systems as well as patients,” says Ulas Bagci, an assistant professor in UCF’s Department of Computer Science.
To perform the study, the researchers trained a computer algorithm to recognize COVID-19 in lung CT scans of 1,280 multinational patients from China, Japan and Italy. Then they tested the algorithm on CT scans of 1,337 patients with lung diseases ranging from COVID-19 to cancer and non-COVID pneumonia. “We showed that robust AI models can achieve up to 90 percent accuracy in independent test populations, maintain high specificity in non-COVID-19 related pneumonias, and demonstrate sufficient generalizability to unseen patient populations and centers,” Bagci says.
FluSense, a compact device about the size of a large dictionary, uses AI to predict trends in infectious respiratory illnesses, such as influenza, by detecting cough sounds and counting people in public spaces. This new edge-computing platform, envisioned for use in hospitals, healthcare waiting rooms, and larger public spaces, may expand the arsenal of health surveillance tools used to forecast seasonal flu and other viral respiratory outbreaks, such as COVID-19. "I thought if we could capture coughing or sneezing sounds from public spaces where a lot of people naturally congregate, we could utilize this information as a new source of data for predicting epidemiologic trends,” said Tauhidur Rahman, computer scientist and mobile sensor expert at University of Massachusetts Amherst.
Engineering computer scientists at Northwestern University are aiming to speed up treatments and vaccines for COVID-19 — by making researchers’ jobs easier. They have developed a new tool called CAVIDOTS (Coronavirus Document Text Summarization) that searches through scientific literature, predicting the most useful results for each user. After pulling documents of interest, the tool then uses AI to generate a short, easy-to-skim summary of each paper. “Researchers can spend hours combing through documents and reading peer-reviewed papers,” said Ning Wang, a graduate student who developed the tool. “Our tool provides the most salient details for academic articles rather than simply retrieving them. We hope this will be a time saver for researchers, guiding them to key information.”
To use CAVIDOTS, users can visit a web-based application and enter search terms. They can first enter large categories and then more specific keywords. The tool then searches through 30,000 documents in the COVID-19 Open Research Dataset (CORD-19), a free database housing scholarly articles related to the novel coronavirus.