With everything that’s happened in the last year and a half, it’s easy to forget that AI isn’t a new trend, and its potential extends far beyond removing people from our photos or generating emails in Gmail. It’s also easy to forget that Google was the leader in applying AI to all types of jobs and industries to make our lives easier, if not save them.
Generative AI, such as Gemini or ChatGPT, is just the tip of the iceberg; the latest project presented by Google shows that this technology can be truly revolutionary. It is an AI capable of detecting lung diseases simply by listening to the patient through a microphone; like with our smartphone.
The system is already being used in India forto touch diseases such as tuberculosis at an early stage; In this country, tuberculosis kills nearly a quarter of a million people each year, in many cases because diagnosis comes too late.
Although there are diagnostic methods and medical devices capable of detecting tuberculosis before it is too late, the lack of medical personnel and, above all, the high price of the devices prevent further expansion of detection. This is where the Google project comes in, which has collaborated with an Indian startup, Salcit Technologies, to develop an AI model for detecting the symptoms of this and other diseases.
To do this, Google trained its AI model with more than 300 million sound files
Once downloaded to a smartphone, AI is able to use this data to detect subtle differences in coughwhich can indicate the presence of one disease or another. According to experts, the sound of coughing is equivalent to a blood sample, and it is enough for the person to cough near the mobile phone for about ten seconds to obtain an accuracy of 94%.
The advantage of this system is obvious: not only is it a much cheaper and faster system, but also a mobile phone can be taken anywhere in the world without the need for special transport. It also reduces the training required, allowing inexperienced people to carry out checks independently.
However, this system is not perfect; requires an internet connection, as current mobile phones are not yet fully prepared to run AI locally and require the power of the cloud. In addition, problems have been noted in the acceptance and adoption of the system by local doctors; But above all, the big obstacle is typical for AI-based projects: obtaining enough data, or in this case, obtaining cough sounds without background noise and that can be used by these projects.
