We are extremely thrilled to present to you our conversation with Reshma Suresh, the Head of Operations at Qure.ai.
In this conversation, Reshma talks about building, scaling, and delivering mission-critical data products using AI, the symbiosis between technology and business, her vision for AI-enabled healthcare products, and more.
SDD: Reshma, currently you are working as the head of operations at Qure.ai. What was your first encounter with healthcare product development like and can you tell us a bit about your journey in this regard?
RS: During my college days, I witnessed some incidents that showed me that quality healthcare was accessible only to those who could afford and few unfortunate events of medical negligence. I felt technological innovations could play a significant role in bridging these underlying gaps. Ever since, I’ve been very interested in disruptive innovations and approaches in healthcare. I then learnt about an interesting set of professionals who were building remarkable tools for point-of-care diagnostics, with a vision to democratize healthcare in India. I joined them as a founding member and together we built the first non-invasive anaemia scanner in India aimed at improving maternal and child health. We went to communities with high maternal and infant mortality rates to screen the adolescent girls, pregnant women and children, who would otherwise have to travel miles to get this done or not get tested at all. I’ve had first-hand experience screening them with our product alongside community workers, seeing them put on treatments, and witnessing their health conditions improve.
Years later, at Qure.ai, I’m on yet another exciting journey with a diverse team having a similar vision, creating powerful impacts. We use deep learning techniques for early detection and diagnosis of conditions like stroke, tuberculosis and now COVID-19.
SDD: Based on your experience, can you talk to us about building, scaling and delivering mission critical data products using AI? What are some of the key features of the product development cycle?
RS: Artificial Intelligence in healthcare, especially diagnostics, is still regarded only to augment human capabilities, and I believe it will continue to be so for another few years. One key reason for this is the lack of evidence to be completely “trustworthy”, while factoring in risks to human lives. If these tools can consistently and reliably outperform human experts, they may be seen as viable alternatives. However, in the current scenario, hybrid workflows with AI in conjunction with humans can result in better decision making. The accessibility is such that every other day I learn about a new/ budding AI company in some part of the world.
For such AI products to be considered as reliable, the algorithm should be trained with a large amount of data, much more than a human would encounter in their lifetime. Another significant factor is the accuracy claims backed by real-world validations in independent, recognised and peer-reviewed journals. At Qure, we were cognizant about these facts since we started building our platforms with proprietary deep learning algorithms. Our product platforms are trained on millions of scans and the validation results are published in prestigious journals like The Lancet, Plos One, Nature Scientific reports etc. These are and should be a continuous process throughout the product lifecycle, as algorithms get stronger and data proliferates.
SDD: While translating technologies to products, what are the biggest challenges you faced in terms of the symbiosis between business management and core technology?
RS: Well, AI application in healthcare is fairly recent. Building a robust technical capability is very important, it’s not easily achievable, and it is a resultant of several factors, especially world-class human talent and data. The irony is, we need humans to build machines that would eventually replace humans. It’s equally important to translate these technical capabilities to real-world solutions that integrate into the existing health care systems – that’s as challenging as building the technology. We can think of all possible scenarios on how the product will be used and design for it, but there is a limitation to one’s imagination on how the technology can make an apparent difference in the end user’s daily workflow for the desired outcome. Our customer is the first source of truth. There had been many instances where I couldn’t succeed in realizing solid technologies, simply because it was not “usable”. We’ve been fortunate to have some great customers who believe in a collaborative approach and an agile team willing to adapt to these feedbacks, with a quick turn-around. It’s also important to understand that it takes time to see the impact of the value created.
SDD: As you know the world is banking on technology to overcome these exceptional times. Can you tell us a bit about the relevance of Qure.ai’s work at this time?
RS: We at Qure have been working on AI solutions that aid in disease detection for other epidemics like Tuberculosis. The beauty of platforms built at Qure is that its capability could be extended to multiple dimensions in a short span. When the COVID-19 outbreak happened, within a few weeks we re-purposed our technologies to add additional features to our product to detect radiological manifestations of COVID 19, predicting likelihood and severity of the disease from chest X-rays. We also took a step beyond the existing tech, to build qScout- an AI powered contact management and chatbot integrated virtual monitoring application.
Our software is now been used in over 40 sites across the world for triaging, mass screening in containment zones and disease progression monitoring for COVID-19. You can read more about how we achieved this here.
SDD: What is your vision on the future of AI enabled healthcare products?
RS: AI has endless possibilities; In the near future AI with Brain computer interfacing (BCI) could be revolutionary and would dramatically change lives of individuals affected with severe cases of cerebral palsy/ traumatic brain disorders, individuals with impairments or help paraplegics to lead a normal life. There are few recent start-ups in this domain building products for individuals with speech impairment to communicate their thoughts.
One of the questions I have been asked many times is “Will AI products replace humans” and is debatable. When wheels were invented, it transformed the entire transportation and agricultural industry that time. Centuries after, automobile was invented in late 1800s replacing horse-driven carts and now we have self-driving cars – this is disrupting the industry and what this did to mankind is upgrade their standard and quality of living. It’s a matter of generating enough evidence, and in a few years, AI-powered healthcare products will change its role from augmenting to replacing humans, enabling them to contribute more in evolving medical science. What these products would do is upgrade normal technicians to specialists and specialists to super specialists. And when that happens, we might even be able to predict, detect and create vaccines for pandemics that may occur in future much before its onset.
SDD: Are there any projects or papers related to AI enabled healthcare products that caught your eye recently?
RS: I read a recent Forbes interview about a start-up named Cognoa, that uses AI to detect Autism Spectrum Disorder (ASD). The article talked about how children with ASD can be diagnosed within first two years and currently pediatricians refer them to specialists with a waiting period of over 1 year before they could be any diagnosis. It’s interesting how their technology with a combination of questionnaire and video-based techniques empowers pediatricians to diagnose ASD in children as well as provide therapeutic for treatment in the early development window of five years. They have a recent publication on nature scientific reports here.
SDD: What inspires you every day?
RS: Each day comes with a different set of exciting challenges and resulting in new opportunities. What excites me most, is how the solutions we design, and build can transform lives. I’ve been fortunate to have associated with organizations that create real impact on human lives through healthcare delivery.
I’ll narrate an incident that’s still fresh in my memory- I’ve always made it a point to visit the grass root level where our products are being really used. One day I was out in the field for a Tuberculosis (TB) screening activity using our products for individuals at risk of contracting the infection. A mother and child approached me, asked why people subjected to Chest X-ray and being guided to the clinic out-patient department. She then told me how she had lost her husband, the only earning member of the family, due to late diagnosis of TB. With her, was her 13-year old child, who was not attending school anymore since she was going though preventive treatments for TB. That moment, I was thinking how access to accurate and early diagnosis through technologies could save many more lives and wished we had done what we are doing now, a little earlier.
So, every day, when I see the number of chest X-rays we process at Qure and enable the early detection of a potential tuberculosis individual, who may have been otherwise missed, I feel content. Because that’s the number of lives we’ve touched and made a difference, without even them knowing about it!
Join our She Drives Data community on SHEROES (You can also download the app on iOS and Android) to connect with thousands of women data enthusiasts across the world, hear from data science experts and get updated on the latest tech news every day!