To detect a deadly disease, this Google AI has an original method

In partnership with specialized companies in the medical sector, Google is working on the development of artificial intelligence capable of detecting a fatal disease without any invasive examination.

Going to the doctor with a fever, cough, or other common symptoms is usually enough to get you out with a diagnosis and possibly a treatment. But sometimes it is necessary to carry out further examinations. So you have to go to a laboratory or in clinical to take a blood test for example. An invasive procedure whose results take some time to be analyzed. Theartificial intelligence wants to change all that.

In terms of disease detectionwe are seeing the emergence of methods that do not require pricking or puncturing, while still allowing us to quickly determine what is affecting the patient. To do this, AI looks you in the eye, but it can also be interested in the shape and color of your tongue, two factors that can identify several pathologies. The model on which Google currently working in partnership with specialist firms adopts another approach: that of bioacoustics.

Here’s How This Google AI Easily Diagnoses A Deadly Disease

Bioacoustics, as its name suggests, consists of studying sounds in order to interpret them in order to give them their biological meaning. Google focuses on detecting the world’s deadliest infectious disease : there tuberculosis.

According to the World Health Organization, it is approximately 4,500 deaths per day. It can be treated, but the problem is that a huge number of patients are not diagnosed in time. The company has partnered with an Indian startup, Salcit Technologiesto remedy this. In the country, nearly 250,000 people die of tuberculosis every year.

Read also – AI detects this disease in children in 100% of cases

In order to train its AI model, 300 million sound clips from around the world were used: coughs, sniffles, sneezes, breaths… They come from non-copyrighted and publicly available sources. On YouTube for example, but also from hospitals that take sound recordings of tuberculosis patients. Using these samples, artificial intelligence can spot subtle differences in people’s coughs and make an accurate diagnosis. This requires nothing more than a smartphone.

Google and partners want to make tuberculosis detection accessible to everyone

In India, Salcit Technologies is already using its own AI to detect tuberculosis. It is using Google’s model to improve it and the results are very encouraging. coughing for 10 seconds near a cell phonethe application gives a response with a 94% accuracy rateThe advantage of such a device is that it is very easy to use in places where there is no access to more sophisticated devices.

For Shravya Shetty, director of engineering research at Google, the goal is also to train as many people as possible to use it. Ultimately, Access to a smartphone will be enough to diagnose someone, or even yourself. Obviously, this is not without raising several issues.

Read also – This AI sees the future, it detects if you risk a heart attack in 10 years

Healthcare professionals will have to accept this new tool that radically changes the way they do their job. People who are not comfortable with technological tools may also have difficulty, for example, not thinking about going to a quiet place to avoid background noise while recording a cough.

Very promising, Google’s AI is still in the development phase and commercialization is still far away. The Mountain View firm is not at its first attempt in this area. In 2018, it announced that it was working on an artificial intelligence capable of detecting breast cancer in 99% of cases.

Source: Bloomberg