Indian Scientists Build App To Detect Tuberculosis Instantly With Just Sound Of Your Cough
The app is called Timbre (developed my medtech company Docturnal) and it detects TB in a patient with the help of a cough audio recording.
Tuberculosis, often referred to as TB is one of the most dangerous diseases that's not just easy to spread but also takes a long time to get detected. As per WHO, India tops the number of patients suffering from TB -- around 26.9 lakhs in the year 2018.
Economic Times
The conventional screening process involves the sputum test which asks for the patient to provide a sample of their phlegm that is then tested for hours to come to figure out if he/she is affected by TB or not.
However, now, the IIT Calcutta Alumni team has created an app that defects TB with the help of artificial intelligence and machine learning.
The app is called Timbre (developed my medtech company Docturnal) and it detects TB in a patient with the help of a cough audio recording.
According to Rahul Pathri, a biotechnologist by profession and founder of Docturnal, "It uses artificial intelligence and machine learning to make the cough interpretation. It is easy, non-invasive, affordable and an easily accessible procedure for TB diagnosis."
A doctor or medical practitioner records the audio of cough of the patient through a series of microphones. This audio is loaded on the app along with other necessary information like cough pattern, when it started, age, previous medical history etc.
Economic Times
After adding relevant data machine learning runs its sorcery and gives out final results. The machine learning algorithm is trained with a data set of over 7000 cough recordings to generate accurate results.
According to Pathri, When we cross-validated our solution with existing solutions, we found an accuracy of 85 percent. Once the patient is detected TB positive, he/she will be referred to medical experts, partner physicians and diagnostic centres.
He further added, "TimBre¡¯s applications is not only limited for TB screening but can classify the cough into other ailments as well, such as bronchitis, pneumothorax, smoker¡¯s cough, etc. The algorithm contains an entire spectrum of lung-based diseases.¡±
TimBre is in currently deployed as a pilot project in a few chest hospitals in Hyderabad. Docturnal¡¯s clinical trials are going on at Narayana Hrudayala in Bangalore.
With inputs from ET
