Though data is tough to come by, AI companies are having a healthy run

BENGALURU: The odds were stacked against Rohit Kumar Pandey, cofounder of artificial intelligence-based healthcare startup Sigtuple. Pandey and his cofounders were up against tough questions like legal requirements, privacy issues, data management and data annotations. All of this for acquiring medical data.

Artificial Intelligence (AI) in healthcare is expected to be the next big thing, but crucial medical data required to build AI models are usually locked up in hospitals and labs. The only way to crack this problem was to take the arduous route of tying up with hospitals and labs.

Medical institutions are very choosy about sharing data and rightly so, as any patient data is private information. Any data sharing goes through the ethics committee and only after various approvals does it reach startups. Startups in most cases even sign non-disclosure agreements with hospitals before getting their hands on the data.


“We had to go through a litmus test to prove our seriousness before we were given the first data set,” said Pandey. Sigtuple’s product Manthana can read medical data such as blood sample images, and classify objects of interest and detect diseases.

Pandey isn’t the only one. A clutch of AI-based healthcare startups are making their presence felt in the growing healthcare startup ecosystem in the country. The startups are actively entering into partnerships with hospitals and labs, both to get crucial data and expertise from doctors. Overtime, this has become a symbiotic relationship with startups learning from the experts and hospitals getting a peek into the future. They have co-developed products for the nascent market.

“We spent a lot of our early days tying up with medical institutions as well as setting up a group of radiologist advisors. Both groups were instrumental in defining the problems we should solve,” said Prashant Warier, CEO of Qure.ai, a healthcare AI subsidiary of analytics service provider Fractal Analytics. The company uses algorithms to automatically interpret chest X-rays and head CT scans.

Adarsh Natarajan, CEO of cervical cancer screening startup AIndra, has entered into similar tie ups. For example, Dr Vani Ravikumar, cofounder of Karnataka-based diagnostic chain RV Metropolis, has been crucial in testing the ‘staining’ capabilities of AIndra’s product. Normally, every slide that is checked by a pathologist is stained, so as to enhance the visualisation under a microscope. Ravikumar lent her expertise to see how well the product could compete with humans when it came to efficiency.

“I accessed the quality of the stain. I scored the slides for various parameters and compared it to manual staining. We give points to each. They were blind tests,” said Ravikumar.

AIndra’s product CervAstra is a point-of-care device to help screen cervical cancer in tier-II and III cities. AIndra works with Bengaluru-based Kidwai Memorial Institute of Oncology, Rajarajeshwari Medical Hospital, RV Metropolis and Manipal-based KMC to validate its product. Natarajan believes another big challenge staring at AI-based healthcare companies is to make hospitals and doctors understand that AI would not replace them but make their jobs simpler and effective. “We had to convince them that our proposition was not to eliminate the need for a pathologist but to make screening faster by reducing the redundant work of the pathologist,” said Natarajan.

Natarajan and team had to train doctors on software that they had specifically built for annotating the data.