As a leader in the technological space, Haval Dosky is always keeping tabs on how emerging technology such as artificial intelligence and machine learning is being utilized to make society better. Given the impact that the current COVID-19 coronavirus pandemic is having on our communities, every segment of our economy is heeding the call to step up and pitch in to create solutions and save lives, following the medical profession’s lead. Solutions from the technology sector for harm reduction and mitigation are more important now than ever. In this article, Haval Dosky discusses several ways that AI technology is being implemented to help us fight the current crisis and give us the tools required to lessen its detrimental effects, with the side-effect of making humanity better-equipped to fight the next emergent disease or crisis.
AI Project for Diagnosing COVID-19 Via Voice Analysis
Haval Dosky notes that perhaps the most important current difficulty in trying to keep up with the crisis is the limitations on supply and speed of COVID-19 testing. As necessity is the mother of invention, this scarcity has inspired several teams to find novel approaches to diagnosing the virus.
For example, Carnegie Mellon University researchers are currently developing a voice analysis system for use in diagnosing COVID-19. One of the benefits of this approach is that if successful, testing could be administered remotely, including from the comfort of the patient’s home. This would be enormously helpful in alleviating hospital crowding, decreasing the potential increased rates of spread, and keeping medical professionals healthy. While such a method will likely never be as accurate as a full test, if successfully implemented, it would provide another layer of screening to keep people with other, less serious ailments home safe and out of the risk of infection. Currently, researchers are asking healthy and infected individuals to share a recording of their voices to benefit the algorithm and increase its success.
The test is only a few minutes long, and requires the patient to cough three times, say ‘a’, ‘o’, and ‘e’ sounds for as long as possible, count to 20, and say the alphabet. While the app still in its early stages, and in no sense should be used to replace a regular medical examination, the prospect is exciting and could help many people if successful, not to mention serve as a building block for future advances.
AI Tool for Predicting Critical Illness in Patients
NYU researchers have been enlisting the help of AI to assist with predicting which COVID-19 patients may later develop severe respiratory disease such as Acute Respiratory Distress Syndrome or ARDS. To do so, researchers collected lab, demographic and radiological findings from 53 diagnosed patients and used that data to train AI models. In an interesting development, the researchers found that many characteristics previously thought to be indicators of COVID-19 were not great indicators for predicting if these individuals would later develop other severe ailments. The AI tool determined that Alanine Aminotransferase or ALT levels in the liver, as well as myalgia and hemoglobin levels, could be used for predicting the risk of ARDS with 80 percent accuracy. There is still work that must be done to confirm the validity of the NYU model and perfect it, but Haval Dosky is incredibly impressed by the ingenuity that researchers have displayed in adding to humanity’s arsenal for fighting the COVID-19 virus and future diseases as well.