How AI Could Aid The Coronavirus Crisis
We’re now deep into pandemic territory as most of the world has been in some form of lockdown for at least a week. The world’s health systems are being put to the sword as they attempt to cope with the disease. To ease some of the pressure, the President of United States, Donald Trump, announced an expansion of telehealth services last week. From this an emphasis on technology in health has blossomed with results we can all be positive about.
AI Makes A Positive Impact
We have previously discussed the potential, and positive effect Artificial Intelligence could have within our health system. Studies have demonstrated that using AI as a triage service could effectively predict mortality rates. Therefore, aid decisions on which patients to prioritise and where.
Although this study showed fantastic results, we did not anticipate a pandemic such as Covid-19. This has fast tracked the use of AI in healthcare. From this, researchers from New York University have partnered with two hospitals in China to start trialling AI technology.
What Does This AI Do?
The AI acts as a decision support tool, similar to our own skin search AI. The tool helps ER physicians decide which patients to admit and which should be sent home. This is a complex and critical decision that ER physicians are forced to make more frequently with rising Coronavirus cases. This, in turn, is pushing hospitals way beyond their capacity and resources.
The decision on which patients to prioritise is made particularly difficult by the apparent unpredictability of the disease. Some people only ever suffer from mild symptoms, others it hits much harder. It is for this reason that scientists hoped AI could bring some predictability to the disease for ER physicians.
How Did They Perform The Test?
The tool was tested using demographic, laboratory and radiological findings from 53 patients who tested positive for the coronavirus at two Chinese hospitals. These patients had presented typical coronavirus symptoms, a mild cough, fever and stomach upset that, for some of the patients, developed into severe pneumonia or acute respiratory distress syndrome (ARDS).
Interestingly, patterns seen in lung imaging, age, sex and other markers were not helpful in predicting which patients would end up being the sickest. The AI found that three physiological metrics were actually the best predictors of severe disease. This was: elevated hemoglobin levels, deep muscle aches (myalgia) and slightly raised levels of a liver enzyme called ‘alanine aminotransferase’.
By using these three readings plus other factors, the AI tool was able to predict the likelihood of ARDS with up to 80% accuracy.
What Does This All Mean?
To summarise, the AI discovered two interesting results when testing. Firstly, is that factors such as age, gender, lung function etc. were of no use when trying to predict the severity of a patients infection. Secondly, factors that were not previously considered such as hemeglobin levels, allowed for 80% accuracy in predicting Coronavirus severity.
The plan is to test this AI with more patients to improve accuracy and measure results. If it continues to perform as it is currently, we could be seeing more of this technology in our hospitals very soon. You can read more about this study here.
If you’re interested in AI and want to see what all the fuss is about, you can of course give our skin image searcher a try for free. It’s able to establish skin conditions with 80% accuracy on 43 skin diseases.
Ask a Dermatologist
Anonymous, fast and secure!
Ask a Dermatologist Now
Anonymous, fast and secure!
Specialist doctor from the University Hospital in Gothenburg, alumnus UC Berkeley. My doctoral dissertation is about Digital Health and I have published 5 scientific articles in teledermatology and artificial intelligence.