COVID-Detecting Algorithm Developed in MIT


Ever since the first case of COVID-19 broke out, the world was in turmoil, with a lot of people wondering if they will be next in catching this contagious virus. Despite the many different measures governments set in place, the death toll still seemed to be rising with second waves in other areas.

This had MIT students and professors thinking about what they could do to help out in these tough times. Their ponderings resulted in a very smart algorithm that manifests the sheer power of IoT. What algorithm did they make and does it work? Here are more details about MIT’s COVID-detecting algorithm:

The algorithms created to fight the spread of Covid-19

A lot of countries all over the world have adopted apps that are implemented nationwide to try and curb the spread of COVID-19. In the United States, there were more apps that use various AI/ML models such as the Covid cough algorithm.

Apart from the newly MIT cough detector, there are other apps that have been launched. Here are a few of them:

  • CovidSafe
  • coEpi
  • How We Feel
  • ProjectCovid
  • TeamSense

These are only a few of the apps being used in the U.S but the new MIT algorithm is projected to change the game.

How was AI used to develop this algorithm?

A widely used subset of AI, natural language processing, was used in developing this app. The Covid cough detector is based on this AI model because it processes sounds originating from your vocal cords.

It processes the intensity, sound, and other audio elements to make a diagnosis. The technology used in this Covid algorithm is cutting edge and there has not been any other software with these capabilities developed for this purpose.

What data backs this research?

There is a treasure trove of data that had been used to develop this Covid-19 algorithm. For example, MIT collected 70,000 audio samples with different types of cough and amongst them, some were contributed by people who had COVID-19.

More than 378 individuals contributed a dataset of 459 coughs and breath when this coronavirus algorithm project was initiated by MIT. Its data collection has improved this app’s accuracy and since their samples are increasing in number daily, the MIT algorithm will be far more accurate soon.

What type of COVID-19 is detected by this algorithm?

The spread of COVID-19 can be dramatically reduced by using this MIT algorithm Covid. Specifically, MIT developed this algorithm for asymptomatic COVID-19, which is the hardest to detect because it does not present any clear symptoms.

That is why this algorithm for Covid-19 is a game-changer. It has features that were unimagined before. Since it is getting more fine-tuned with more data available, this algorithm will be more accurate in detecting asymptomatic COVID. Undoubtedly, that can go a long way to curbing the spread of this virus.

Where can this algorithm be implemented?

This Covid AI-algorithm can be used at a lot of places if it can be integrated into user-friendly and possibly mobile apps. Employers might use it as a further screening mechanism to detect asymptomatic COVID-19 that employees might potentially have.

At the same time, this app could also do great work as hospitals or clinics in detecting this virus when it does not indicate symptoms. Therefore this AI model Covid detector has a variety of applications that could utilize it at its fullest potential.

How effective is this algorithm?

This algorithm has improved greatly, especially since its data sets are continually widening. That has been reflected in the accuracy when detecting COVID-19.

Back in July, it did exceedingly well even with the limited data it had. In July, it was reported to have 80% accuracy. Later on, MIT’s algorithm has been reported to have a 98.5% accuracy rate.

The bottom line

The uprise of COVID-19 has inspired a lot of initiatives and most of them are aimed at reducing the spread of this virus. Although a lot of these initiatives are successful, this research conducted by MIT has changed the game. By detecting and diagnosing asymptomatic COVID using audio only, this has really proved how much AI could do for the healthcare industry as a whole.