- Most coronavirus cases are mild or moderate, and some people might not even experience any symptoms before they recover from COVID-19.
- The elderly and those suffering from other medical conditions may experience more severe COVID-19 cases, which could require oxygenation and ventilation.
- Researchers used artificial intelligence to find three signs that can accurately predict a severe outcome, including two parameters that are routinely tested in hospitals.
- Visit BGR’s homepage for more stories.
The novel coronavirus outbreak is far from being under control, with COVID-19 ravaging several countries at the moment. The US alone has accounted for nearly 175,000 cases of the more than 818,000 cases worldwide at the time of this writing. Italy is topping the casualty list with over 11,500 deaths and a fatality rate of 11.39%. Social distancing measures and good hygiene habits should significantly flatten the curve, but the results won’t be seen for several more weeks. The lower the curve, the less crowded the hospitals will be. That way, the severe COVID-19 cases will have a better chance of surviving the disease, which still lacks an effective treatment or vaccine.
Several such drugs are in testing, with some of them showing promise in limited trials. Plenty of vaccines are in the works as well, with at least two trials already underway. But there might be another untapped resource that could help doctors create therapies that could assist critical COVID-19 patients: Artificial intelligence. A new study shows that AI has been able to highlight three COVID-19 symptoms that are indicative of severe COVID-19 complications. Interestingly enough, it’s not the most common coronavirus symptoms that signal rapid deterioration after infection. If the discovery can be scaled up to more patients, it could potentially save more lives in the months ahead.
You’ve heard it a hundred times before, but COVID-19 has no universal symptoms. If you’re experiencing a fever, coughing, and shortness of breath, you might be infected. But first you have to rule out the flu. Other common symptoms may include throat pain and fatigue. Also, doctors observed that some patients reported a loss of smell and taste, which is the only COVID-19 symptom that really stands out. But many people who contract the virus will not show any symptoms at all, or at worst will simply experience mild discomfort.
According to a new study, only one of these symptoms can be indicative of severe disease, but only when combined with two other signs which require hospitalization. Per AFP, researchers from the US and China used AI to analyze data from 53 coronavirus patients across two hospitals in Wenzhou, China.
The algorithms discovered three changes in the body that precipitate severe illness: Body aches, levels of enzyme alanine aminotransferase (ALT), and hemoglobin levels. ALT is a liver enzyme that’s tested to measure liver function and diagnose liver failure. Hemoglobin testing is part of the standard blood work you get when admitted to the hospital.
The AI figured out these three features were the most accurate at predicting a severe COVID-19 case. The algorithm showed a 70%-80% accuracy at predicting the risk of acute respiratory disease syndrome (ARDS). ARDS is the COVID-19 complication that fills the lungs with fluid and kills some 50% of patients who get it.
Other symptoms, including particular patterns in lung imaging, fever, and strong immune responses, were not useful at predicting whether a mild case could worsen to ARDS:
The model highlights that some pieces of clinical data may be underappreciated by clinicians, such as mild increases in ALT and hemoglobin as well as myalgias. Key characteristics predictive of diagnosis, including fever, lymphopenia, chest imaging, were not as predictive of severity. Likewise epidemiologic risks such as age and gender were not as predictive; all ARDS patients in this study were male but most males did not develop ARDS.
“It’s been fascinating because a lot of the data points that the machine used to help influence its decisions were different than what a clinician would normally look at,” physician and professor at New York University’s Grossman School of Medicine Megan Coffee told AFP.
The team is looking to further refine the data, and the AI tool might be ready to deploy sometime in April. The full study is available in Computers, Materials & Continua.