"Study Finds AI Inadequate for Diagnosing COVID-19 from Coughs" - Credit: TechCrunch

Study Finds AI Inadequate for Diagnosing COVID-19 from Coughs

AI Does a Poor Job of Diagnosing COVID-19 From Coughs, Study Finds

Artificial intelligence (AI) is becoming increasingly popular in the medical field for its potential to diagnose diseases quickly and accurately. However, a new study has found that AI does not perform well when it comes to diagnosing COVID-19 from cough sounds.

The research was conducted by scientists at Imperial College London and published in the journal Nature Medicine. The team used an AI algorithm to analyze more than 1,000 recordings of coughs from patients with confirmed cases of COVID-19 as well as healthy individuals who had never been infected with the virus. They then tested how accurate the algorithm was at distinguishing between those two groups based on their coughing patterns alone.

Unfortunately, they found that the AI performed poorly compared to human experts who were able to correctly identify whether someone had COVID-19 or not based on their coughing pattern nearly 80% of the time. In comparison, the AI only achieved an accuracy rate of around 50%. This suggests that while AI may be useful for other types of diagnosis tasks such as detecting cancer or heart disease from scans or images, it is still far behind humans when it comes to identifying illnesses like COVID-19 just by listening to coughs alone.

This isn’t necessarily surprising given how complex respiratory illnesses can be and how difficult it can be even for experienced doctors and nurses to make accurate diagnoses without additional tests such as blood work or imaging scans. It also highlights some limitations with current machine learning algorithms which are unable to pick up subtle differences between different types of illness symptoms that humans can easily recognize due to years of experience in medicine and healthcare settings.

Despite this setback however, researchers remain optimistic about using AI for medical diagnosis purposes going forward since there are many other areas where these technologies could prove invaluable such as helping doctors quickly detect signs of stroke or Alzheimer’s disease early on before they become too severe or providing personalized treatment plans tailored specifically towards individual patients needs rather than relying solely on general guidelines set out by health organizations like WHO (World Health Organization).

As technology continues advancing over time so too will our ability use machines learnings algorithms more effectively diagnose various conditions including Covid 19 – although we should remember that no matter what level artificial intelligence reaches there will always be certain tasks better suited towards human expertise instead due largely thanks our unique capacity understand context nuance something computers simply cannot replicate yet .

Original source article rewritten by our AI:





By clicking “Accept”, you agree to the use of cookies on your device in accordance with our Privacy and Cookie policies