Health Tips: Miracle of Artificial Intelligence, will detect lung disease by listening to the sound of cough!

Sometimes due to cold and fever and sometimes due to dust and dirt, the problem of cough often occurs. But do you know that in future, just by the sound of your cough, it can be known whether there is any disease in your lungs or not. Yes, scientists have developed a technology using Artificial Intelligence (AI) which can tell by analyzing the sound of cough whether you have a serious disease like Tuberculosis. Let us know about this new technology in detail.

A team of AI researchers from Google Research, in collaboration with the Center for Infectious Disease Research in Zambia, has developed a machine learning system to detect lung disease. This system is capable of detecting the disease by analyzing the sound of cough. Researchers have used YouTube videos for the study. Google researchers have named this new system Health Acoustic Representations (HeAR).

In fact, during the Corona epidemic, health workers had told that many times they used to detect corona only by the sound of patients' cough. This became the basis for the development of the Heer system. Other research teams are also trying to develop systems to detect various diseases based on the sound of cough, but Google's approach is a little different. Other teams train AI systems using recordings labeled with the names of diseases, while Google has used an approach used to build large language models (LLMs) such as ChatGPT.

According to a study published in the journal Nature , in the HeAR system, various types of sounds like breathing, gasping and coughing received from YouTube are first converted into spectrograms. Next, the researchers hid parts of these spectrograms and instructed the AI ​​to complete the missing parts. This process is similar to the way LLM learns to predict the next word in a sentence. This process created a base model, which the researchers say can be adapted for different tasks.

Result of the study:

Researchers used this model to detect TB and corona. They found that in the case of Corona, the HeAR system achieved a score of 0.739 on one dataset and 0.645 on the other. At the same time, its score for TB was 0.739. This is a better result than existing systems. However, researchers say that more research is still needed in this area. But, it is possible that in the future doctors will also be able to use sound tests to detect lung disease.

Photo Credits: Google