Artificial intelligence is slowly becoming a gamechanger in medical diagnosis. Google recently announced its partnership with a Dutch university to use the deep learning branch of AI to improve diagnosis of breast cancer and afford better treatment to patients. An Indian startup, SigTuple is doing something similar with its flagship product Manthana. This AI solution uses AI to analyse most screening tests and visual medical data like MRI and CT scans, X-rays, blood smears and semen.
Co-founder and CEO, Rohit Kumar Pandey speaks to Networked India about the role of AI in assisting with timely and improved medical diagnoses, the challenge to scale modern solutions to reap cost benefits and the AI Hackathon they hosted to engage college youth in helping SigTuple solve actual problems.
Networked India (NI): What prompted you to start SigTuple? What gap in medical data diagnosis is SigTuple aiming to bridge?
Rohit Kumar Pandey (RP): We were excited by the vast amount of data available in the medical industry and its potential. We are solving three problems –
- There is an ever-widening gap between the growing number of patients and doctors inspite of the patients’ ability to pay.
- Analysis of visual medical data is a function of medical professionals’ expertise, skill set and even the state of mind. Analysing visual data under a microscope is a very strenuous and fatigue driven process therefore we see a lot of variability and inaccuracy in the reports.
- Adoption of new features and parameters requires a change in hardware and makes it difficult for the labs and hospitals to adopt new technology to scale.
We are looking to solve the above three problems for common screening tests which require visual medical data analysis. It includes analysis of peripheral blood smear analysis, urinalysis, semen analysis, chest x-rays and retinal scans. We want our solutions to reach the masses. Hence we have chosen screening tests as our focus area.
NI: Tell us more about your core product – Manthana. What are its main features?
RP: Manthana is a home grown continuous learning AI platform. It powers solutions for peripheral blood smear analysis, urinalysis, semen analysis, chest x-rays and retinal scans.
Manthana provides the following salient features –
- Ingestion of visual data i.e. images/videos etc.
- Classification of various objects using the AI models
- Sniffers powered by AI models to detect various diseases e.g. Malaria
- Longitudinal memory to use the historical data to predict disease conditions
- Reports supported by visual evidence which enables telemedicine
- Interface for feedback on the reports generated to facilitate continuous learning
NI: What is the scope of Manthana in terms of the types of methods of diagnosis – MRIs, scans, tests, etc as well as range of diseases?
RP: Manthana can power analysis of any screening test that requires analysis of visual medical data. At this point of time our focus is on the analysis of peripheral blood smear slides, urinalysis, semen samples, chest x-rays and retinal scans.
NI: What kind of challenges did you face in terms of convincing hospitals and medical experts to try out Manthana? Were there any other challenges?
RP: Manthana is the AI platform that supports various solutions. It is agnostic to the labs and hospitals as they don’t directly interact with Manthana but through the solutions supported by Manthana.
During our discussions with most of the hospitals and medical experts we realised that most of them believe that quality healthcare delivery to the masses is only possible through adoption of scalable technology. Having said that, adoption of any new technology takes its own time therefore one of the biggest challenge for us will be adoption and change management.
NI: Give us some background into the founding team.
RP: Myself, Apurv Anand & Tathagato Rai Dastidar are the Co-founders of SigTuple. Apurv is the CTO and Tatha is CSO. All three of us are computer science engineers with a collective experience of 36+ years in the industry. Our area of expertise includes forming R&D team, developing large scale systems, machine learning, image processing and big data. We met in 2012, when we were hired by American Express Big Data Labs to form the team and build their big data platform from scratch. While working together for American Express, we developed the chemistry amongst us and when it came to starting something on our own, it was a no brainer to come together.
NI: What is the next step for Manthana? Are there plans to provide access to the common man – patients and their family to understand medical diagnosis as well?
RP: We are looking to scale Manthana so that we can churn large amount of data at a higher speed to derive intelligence out of it and power various solutions that we are currently working on.
Our solutions are going to improve the efficiency of the medical experts which in turn will facilitate quality healthcare delivery to the common man. We are a B2B company and we strongly believe that even though AI can generate the report, it should be validated and approved by a medical expert before it is handed over to the patient.
NI: How far can AI assist with diagnostics? There are a number of startups exploring a range of applicability in India as well as abroad. What is your opinion on this?
RP: AI can play a significant role in diagnosis i.e. basic screen tests for diagnosing very intricate disease conditions. Various startups are working on solving different kind of problems and I strongly believe that as the ecosystem matures, we will be able to see various solutions in the market that are powered by AI.
NI: SigTuple recently hosted an AI challenge on the HackerEarth platform. What was the goal here? Tell us about the experience interacting with other smart young people working to solve actual problems in building your Shonit solution?
RP: We were looking to onboard talented and visionary individuals who would like to be part of this journey therefore we hosted the AI challenge. The response was overwhelming and we invited the top 10 performers to our office to hear their views on the problem. We also discussed the work we are doing at SigTuple to seek their opinion.
NI: SigTuple has enjoyed some success in securing funding recently. What are the future plans for the company?
RP: SigTuple will use the latest round of funding to expand the team, bullet-proof the platform and the product for user adoption followed by commercials, and regulatory clearances for global markets.