Hyderabad: Ever thought of a specialised app which will ‘listen’ to your cough, and within seconds diagnose the health of your lungs. Sounds futuristic, but a Hyderabad-based company has developed an award-winning software to do just that.
Meet, ‘Swaasa,’ an AI driven software that can assess if human lungs are abnormal using an algorithm. And all it takes is a user to cough thrice near the phone. Swaasa, which won the prestigious Anjani Mashelkar award for innovation in November, is developed by Hyderabad-based Salcit Technologies.
Salcit founder Narayana Rao Sripada’s interaction with a professor at All India Institute of Medical Sciences (AIIMS) five years ago was the foundation for this innovation. It was then he realised the need in the medical sector for devices to screen and diagnose lung health in real time with non-invasive technology. It pushed him to think along the lines of developing Swaasa.
“I met Prof. Anand Krishnan (head of Community Medicine at AIIMS) in 2017 as part of another college research project. It led to a discussion on how there was an immediate need for such a technology which is non-invasive and can help identify a problem. Say if it can be attributed to lung airways or lung parenchyma, that itself can be of significant value for the health worker to make an appropriate intervention and that is how the origin of this invention took place,” Rao, who is also the Chief Technology Officer of Salcit Technologies, told ThePrint.
It took him three years, but in 2020 Swaasa was ready to roll out.
Currently, spirometry is one of the most common and standardised methods used to test lung health. But it requires expensive spirometers and trained technicians.
In rural and remote areas, where accessibility to medical care and diagnostics labs is no guarantee, this poses a challenge. Swaasa eliminates all of these concerns, said ManMohan Jain, Chief Operating Officer of Salcit.
Also Read: 4 patented HIV, hepatitis & TB drugs now on govt price control list. Controversial ranitidine cut
So, how does Swaasa work?
The team refers to it as an Artificial Intelligence-based software that can be accessed through an app on a phone.
Swaasa, however, is not a ‘consumer-based’ software, meaning anyone cannot randomly download the app and start using it. The software is designed for health workers or medical teams who can check anyone’s lung health using a smartphone.
Once the software is installed on the phone, health workers can create a profile of the patient, fill in basic details such as name and age and then either generate a link to the user or test a person’s lung health on their own phone.
“They can do it on their phone using the link sent to them by the health worker, which justifies their tagline ‘Respiratory Health Unbound’,” said Rao.
Based on the result, the health worker can take the reports to a doctor if it is necessary and the next step of diagnosis can be done. The whole idea is early detection.
“We collaborated with a team of doctors and determined a Lung Health Index, and devised parameters on par with spirometry standards. Normally spirometry is used to assess lung health and that is also an input for the doctor to decide on the diagnosis further based on the lung condition. So, we thought without using spirometry through the cough, i.e. can we give a similar kind of inference for the doctor,” Jain said.
He noted that the software was able to provide the same kind of inference.
In a clinical study, doctors compared results of a patient’s lung condition assessed through spirometry and also inferences from the software to test its efficiency.
“The software also asks some questions such as if the person has mucus, trouble breathing. All these have been devised as per St.George Respiratory Questionnaire standard,” he added.
An index of 1-3 means the lungs are normal, anything above 8 means high risk of possessing abnormal lungs, Rao said.
From catering to corporate clients who’ve used the technology at offices to check their employees’ health before they return to office, to non-profit organisations, their services are across sectors. The software is also being used at Andhra Medical College’s rural health centre.
“During the validation process of the software, earlier this year, we tested it on 100 people and also used it on patients after the clinical validation. The efficacy was as expected. Such kind of software helps in early screening which can help in quicker diagnosis and here the advantage is the screening process is also without any hassle,” Dr. Gayatri Yellapu, who was formerly with Andhra Medical College and also was part of clinical validation of the software at AMC, said.
With the software, the team was able to diagnose 10-12 new cases of pulmonary tuberculosis, especially among people from tribal communities.
“And this was due to extensive damage of lung parenchyma, which could be due to delay in diagnosis on account of lack of awareness. Such software will help in early detection. But it does not end there, the next diagnosis is important,” she added.
Swaasa, however, is not the first such software providing such a service. There are similar prototypes in the market, focusing majorly on Tuberculosis detection.
Acoustics of cough sound
What started off as a research project, when Rao was advising a group of engineering students, turned into a full-fledged idea and Manmohan Jain and Chief Executive Officer Venkat Yechuri also joined Rao in early 2020 to develop it further. The project received funds from the Central Government’s Department of Biotechnology and also the Department of IT.
To develop the algorithm, the team used ‘acoustics’ of cough sounds. Different types of cough sounds were collected from Nizam Institute of Medical Sciences, which was also part of the project, to understand what kind of information a certain type of cough sound carries.
“Cough is an explosive expulsion of air to clear the airways. It carries a lot of information about the respiratory system. It behaves differently for various respiratory conditions. The cough comes out differently for Asthma. It is different for conditions such as COPD (a chronic inflammatory lung disease), when lungs are healthy it comes out in a certain way,” Rao said.
Jain said that gave the idea of using cough patterns for detecting symptoms. “So, we thought if the cough is unique, why not try an algorithm to understand the uniqueness of it? Parameters such as energy, frequency, and amplitude in sound waves of cough are analysed. It is similar to speech analysis and we built Artificial Intelligence based on it,” he added.
One of their biggest clients in the medical industry is Apollo Hospitals group which during the pandemic integrated Swaasa technology into their portals for people to test their lung quality at home and around 2,00,000 assessments were done from 2020 to 2021, Jain said.
Swaasa, which also has a no-objection certificate from India’s regulatory body Central Drugs Standard Control Organisation (CDSCO), has completed five clinical validations. It is working with the Odisha government to test the software as a pilot project in a few villages.The software also has patents from the Indian Patent Office and in countries such as the US, and Australia.
“CDSCO has updated guidelines on the regulations for ‘software as a medical device’ category which Swaasa falls under. We already have an NOC from them. We have applied for a licence as well. We have also written to several state health secretaries, but are yet to hear from them. PM Modi had spoken about TB mukth Bharat by 2025; we feel our innovation can be a great contributor to it,” Jain said.
(Edited by Tony Rai)
Also Read: Advanced AI can now end global TB crisis