Tuberculosis (TB) is India’s most severe health crisis. According to the National Strategic Plan for Tuberculosis Elimination 2017–2025 report, TB kills approximately 480,000 Indians every year. That is a whopping 1,400 every single day.
Not just in India, Tuberculosis (TB) is one of the top 10 causes of death worldwide. The World Health Organisation (WHO) estimates that over 95% of TB deaths happen in low and middle income countries. India leads the count here.
This strongly points to the fact that poverty and lack of access to healthcare are the leading causes of TB deaths. Various drugs have been tested and rolled out. But again, the problem remains access. According to infectious disease specialist Judy Stone’s report in Forbes, Janssen - a division of Johnson and Johnson - rolled out a drug called bedaqualine in 2012. However, the drug has only reached 1,000 people. Affordability of these drugs is a whole issue in itself.
In conclusion to her report, Stone says that while it is exciting that new drugs are coming into the market, better diagnostic tests are a critical need. Only then can there be more guidance for effective treatment.
This was the same problem Rahul Pathri recognized as well. Pathri had a bitter experience when he had to be tested for TB in relation with another health problem. His mother, too, was diagnosed with TB and he also lost a friend due to a TB-induced lung collapse. A series of unfortunate events made Pathri want to explore a way to find a quick, easy and affordable method of detecting TB.
Hence was conceived the idea of a smartphone application that records the sound of a person’s cough to detect TB. The app, called TimBre, does a spectral analysis of a recorded sound file of a person’s cough. It then uses the machine/deep learning algorithms developed by Pathri and his team to classify the cough pattern.
“When a cough is pathological in nature, it has a certain pattern and characteristic to it. Analysing the sound wave, its energy or amplitude in a low or high frequency can determine if it is pathological in nature,” Pathri says.
Once a user downloads the app, they will have to enter their basic information, some information like your habits, a medical history, sleep and cough patterns. You then record your cough as per the given protocol. The data then gets recorded, processed and an SMS is sent to the person confirming whether the cough is TB related or not.
The cough is analysed in conjunction with clinical data like pre-existing health conditions, demographic data like where the patient lives and in what conditions, and finally, his/her socio-economic background to determine the kind of environment exposure he/she has had. The algorithm then determines whether the condition of the patient is TB or not.
Once the app was developed, Pathri filed for a provisional patent in April 2016 and soft-launched the app in December. A provisional patent is what you initially file and get a priority date till the next year by which a final patent needs to get approved. TimBre’s patents are filed by IP law firm, Lexorbis, in Delhi.
Pathri calls TimBre a screening tool and not a diagnostic tool as it is yet to receive necessary approvals. The team, which includes doctors, is currently running clinical trials in Gandhi Medical Hospital in Hyderabad to improve the accuracy of the app.
“Currently the accuracy of our app is 70-75%. When the accuracy levels reach 98-99%, we will file for an approval with National Accreditation Board for Testing and Calibration Laboratories (NABL). We are working with Gandhi Medical to collect training data and put our app to test. Once we reach that level of accuracy, our tests will be able to determine whether you do have TB or not and suggest a pulmonologist for treatment,” says Pathri.
While the app is free for general users, a package which includes more advanced tools for detection is sold to clinics and hospitals for a starting price of Rs 15,000.
Pathri wants to make the app a screening tool for the masses. The idea is to make it available in schools, railway stations, airports, in addition to hospitals, where regular testing can happen. This will enhance TB detection.
The company is also talking to the large diagnostic giants in India like SRL diagnostics to partner with them and include TimBre as a diagnostic tool for TB.
Started with an initial investment of around Rs 30 lakhs, the parent company Docturnal, also raised seed funding of Rs 10 lakhs from IIIT, Hyderabad. In fact, Docturnal was incubated under IIIT Hyderabad’s incubator Avishkar and was a part of Avishkar’s deep tech accelerator program.
But how effective is a mobile app to detect a disease as serious as TB?
Dr Mohan Natarajan, Director in-charge, National Institute for Research in Tuberculosis, says that while TimBre is a step in the right direction, the symptoms found in patients are not always uniform. “The methodology is fine. But TB as a disease has many variations. Some patients will have a cavity, some wont. So it is not easy to quantify it just by a cough. Sometimes even the body structure affects the way a person coughs. The idea is good but it will have its limitations because of the variability involved with TB. A lot of testing and development is required,” he adds.
Pathri, too, acknowledges certain challenges. For example, TB can also be extra pulmonary, when it is not related to the lung. Globally 10% of the TB cases are extra pulmonary and those with this form of TB cannot use TimBre to detect the disease. But for the remaining 90%, Pathri wants to make TimBre the go-to diagnostic tool.
Once Docturnal is able to improve the accuracy and receive NABL’s approval, the next step is to implement TimBre in all Revised National Tuberculosis Control Programme (RNTCP) centres, in major government and private hospitals and diagnostic labs in India.
Docturnal is also working on expanding its algorithm to detect other lung-based diseases like COPD, Bronchitis, Pneumonia, Whooping Cough and Asthma. It is also developing the app in more Indian and global languages. The aim is to expand TimBre to other foreign countries such as Indonesia, China, Africa, and Russia where TB rates are high.