Nexus-supported start-up Docturnal provides screening and diagnostics for various diseases through smartphone apps.
Tuberculosis (TB) is one of the most severe public health issues faced by India. The country has over 25 percent of the total TB and multidrug-resistant TB (MDR-TB) cases in the world. In 2016, approximately 2.8 million people suffered from TB and 435,000 died from the disease. There are also an estimated 850,000 cases of TB each year that go either undetected and untreated, or diagnosed and treated with potentially substandard drugs and treatment regimens. Such drugs and treatments, according to the U.S. Agency for International Development (USAID), not only fail to fully eliminate the TB bacteria, but they also contribute to an increasing incidence of drug-resistant TB.
In this scenario, Rahul Pathri saw a need for better TB diagnostic tests and, in 2016, founded Docturnal to make the detection of TB and other diseases easier, accessible and affordable. Docturnal’s flagship products include TimBre and retiNNapp smartphone apps to screen TB and diabetic retinopathy, respectively, using artificial intelligence non-invasively.
Docturnal is supported by the Nexus Incubator start-up hub at the American Center New Delhi, a collaboration with the IC2 Institute of The University of Texas at Austin. The company was one of the winners of the UberEXCHANGE start-up mentorship program, announced at the 2017 Global Entrepreneurship Summit in Hyderabad.
Excerpts from an interview with Pathri.
What is the story behind Docturnal’s origin?
I experienced the journey of undergoing TB tests, like IGRA [Interferon Gamma Release Assay] and Mantoux. The cumbersomeness and inaccuracy of the tests were a starting point for me to bring a change.
My first big break was in 2014-15 at the Grand Challenges in TB Control initiative in Bengaluru, where I made a presentation about my idea, even before TimBre was a reality. After that, I was able to connect with a team of like-minded people that believed in the aural or acoustic nature of cough. And that was the genesis of Docturnal.
What led you to target TB? What opportunities did you see to improve its diagnosis and treatment?
Some close family members and friends had been diagnosed with TB. I realized that the current TB diagnostic methods have many of the same issues. First, the tests are not fast. Some tests take several days to process, and the patient has to make multiple trips to the diagnostic lab. Second, many TB tests lead to results that are either inconclusive or subjective, making it difficult to get an accurate diagnosis. Last, many of these tests require resources to allow access to the proper technicians and medical infrastructure. With Docturnal, we wanted to address all of these issues by creating a solution that is accessible, affordable and accurate.
When I worked in the U.S. for seven years, I gained exposure to bioinformatics, which helped me understand the statistics behind developing a possible prediction algorithm. We also had remarkable success with our deep learning solution for diabetic retinopathy, which paved the way for implementing a deep neural net on a spectrogram of a cough file with TimBre.
Could you briefly explain how TimBre works? How accurate is it and how are you working to improve it?
TimBre basically does a spectral analysis of a recorded sound file of a patient’s cough. The app then uses our algorithms to classify the cough pattern. Then, when considered in conjunction with clinical data, like the patient’s pre-existing health conditions and demographic data, the app can help determine if the user is at risk of TB. Currently, the accuracy of our app is about 85 percent. We are working with Gandhi Medical College & Hospital in Hyderabad as well as Revised National Tuberculosis Control Program and World Vision to collect data and pilot our app further. When the accuracy levels reach 90 percent, we will file for approval.
What feedback have you received from doctors and patients about TimBre?
So far, TimBre has been endorsed by the Indian Institute of Information Technology, Hyderabad; AIRmaker and AIM Smart City [accelerators]; Nexus [incubator]; and the Biotechnology Industry Research Assistance Council. After piloting the app in Hyderabad, we plan for our products to be implemented in the hospitals in the city of Taj [Agra]. Our hope is to pilot our products in 50 hospitals within the next six months.
Are there any new projects Docturnal is currently working on?
Docturnal has been working on expanding its algorithms to detect other lung-based diseases like COPD [chronic obstructive pulmonary disease], bronchitis, respiratory syncytial virus, whooping cough and asthma. We also want to develop the app in more Indian and global languages. The aim is to expand TimBre to countries such as Indonesia, China and Russia, where TB rates are also high.
Jason Chiang is a freelance writer based in Silver Lake, Los Angeles.