Chiranjiv Singh is the Chief Commercial Officer at Qure AI, a healthcare startup developing AI algorithms for medical imaging. Their products are built using deep learning technologies and millions of images, helping detect anomalies in CT scans, XRays and MRIs. The startup has particularly been hailed for its technology solutions qXR and qScout during the COVID19 crisis for accurately and speedily detecting abnormalities in chest Xrays, and reducing the downtime for radiologists and healthcare staff around the world. The tech has been used in India, Italy, Mexico, Oman and other nations to screen COVID patients, is garnering praise for its quick turnaround time, specificity and conclusiveness in detecting ailments, and paving the way for tech-driven healthcare, esp in limited or low resource settings. Chiranjiv is no newcomer to the field of healthtech - prior to joining Qure AI, he spent 13 years at GE Healthcare, where he was instrumental in launching the first ever Xray device with an embedded AI algorithm. He joined us from Wisconsin, USA in this podcast with Sindhuja Balaji.

Listen to this episode on Indiaai's Youtube channel.

Here are some excerpts

How have the past six months really been at Qure AI? 

You rightly said, and I agree. This has been a time where the first time in our lifetime, we’re all going through the same experience no matter where you are. Be it India, Australia, USA, Brazil. Everyone has been in lockdown, figuring out how the government is going to limit the spread of the virus. It’s a unifying experience. For us at Qure, it’s a super active time…probably the busiest six months since I joined. The speed of adoption and pace has impacted us in many ways. Our core team is in Mumbai, which is one of the COVID hotspots. Everything is under lockdown. The entire team is remote – the wall of a Mumbai apartment has been converted into a board with Post-It notes, with everything happening virtually. Our sales team is working 24/7. We’re doing sales calls at 2 am or 3 am, reaching customers around the word, spreading word about the product and helping them tackle the pandemic. Its been an amazing experience for our deployment teams. We’re lucky it’s all software so it can be remotely deployed but it has been a completely new experience in term of iterating solutions for COVID, presenting to customers, actually closing orders, getting deployments done and seeing the product impact. I loved what you said - eventually AI is important for the impact it can create and were seeing that happen in the last six months.

Tell us about the product journey of qXR and qScout, its AI and DL capabilities and the accuracy level of the results? 

So there are two distinct products – qXR, where a deep learning algorithm interprets chest X-rays. It was launched three years ago and it has since been an evolving product, getting more data and customer feedback. It can read a chest X-ray and detect at the current standpoint 29 different findings – it can see a nodule, rib fracture, if a feeding tube has been misplaced, different clinical applications. At one level, it has also fond adoption as a workflow management tool, X-ray is the most commonly done imaging exam in the world. Because of high volumes, there is a shortage of radiologists who can read X-rays. A lot of hospitals we went to said “We’re doing 200 X-rays a day. Can you tell me the 20 X-rays that are important or that I have to read first so I don’t delay patient care? Some of our early success with the product in India and outside was with tuberculosis screening. TB is still a big burden disease for India and in other markets, and we found a very good solution to the potential delay in diagnosing TB. The disease is a 100 years old, the cure is known, medicine is available too, but how do you detect it early enough? We realized the workflow for this process is very inefficient. People would take an X-ray and wait for a week for it to be read. In that week, if a person has TB and is walking around, he’s infecting others. Its also contagious like COVID. That time period of a week to diagnose has been cut short to a few mins. We enabled our partners in India and other countries where a person does an X-ray. Even before they leave the room, we give an alert that they suspect TB, mask them, and then take a swab sample for confirmation.

When COVID happened, we realized we have data and expertise to detect findings in the X-ray indicative of COVID. We can see opacity and understand if its COVID or something else. We repurposed that tool and very quickly started giving it to our customers. Once we reached countries like Italy, we realized there is more value we can add by not only detecting and diagnosing but monitoring the progression of patients with COVID. As you would know, there is no known cure for COVID but there are a lot of trials going on. The critical task for hospitals is to manage the workflow right now – so how do you decide if a person tested COVID positive should be admitted or allowed to go home? What’s the right time for ICU, the right time for ventilator? A daily bedside chest X-ray that an algorithm like ours can measure and quantify is proving to be a very valuable tool in clinical decision making. Its been a very interesting journey, not just as a workflow or detection tool but helping clinical partners in actual decision making.

qScout was a new product. We didn’t have something like this before COVID. End February was early days when cases were rising. As a team we thought, if this continues to grow, our existing hospital infrastructure around the world will prove to be insufficient. We don’t have enough beds, emergency rooms or testing facilities to take care of everyone. We said, why not build something that would help institutions and countries manage their populations remotely? A lot of people have mild symptoms. Since there is no known treatment, maybe all they need is isolation and continuous monitoring. Why not do that remotely through a mobile chatbot and using some AI and prevent everyone from rushing to the hospital? The genesis of this product lay in TB. Even in TB, we had contact tracing – if someone has TB, we have to test everyone in the household and workplace to see if they’re also carrying this disease. We repurposed some of these learnings and tools from our TB partners to build this new platform called qScout. This is the benefit of a startup like Qure. From the first discussion of “listen we can build this” to our first sale - was 30 days. The pace of work was amazing. Our engineering team was working 18-20 hours a day, writing code and testing it because we didn’t want to release something that doesn’t work. And we’ve have great success. For qXR, we had enough research and validation that on every test, our account number was in the high 90s. For qScout. we are measuring in terms of adoption. Like in Oman, we started it with 1,000 patients under quarantine and now we’re at a scale where the entire country of Oman is using it as a safe way to come back to work. Obviously, every country wants to open up the economy and safely determine who is ready to come back to the shopfloor or factory or bank.

Your solution is being used in various countries now -- how was it like getting these hospitals to adopt your tech? In times like this where in person demos are a challenge, how did you manage to convey the power of your tech?

COVID has made our obstacle of convincing hospitals to adopt tech easier. There was a realization that not every physician could come to work, not every patient needs to come in either. You are also concerned about workflow of healthcare workers but their safety as well. A lot of barriers to adoption of tech were “how will we integrate IT? What is the security behind it?” We have seen a lot of openness when we proposed solutions. For example, most IT depts of hospitals historically have had a view that they want to protect data within their premises and not release it to the cloud. During COVID, we have been able to get more cloud adoption because hospitals realized engineers cannot come to hospitals to deploy servers. In terms of our demos, all been virtual. We have become natives of Skype and other VC platforms. Now, all our demos are virtual, trials are virtual. We sign contracts digitally. We have worked hand in glove with customers and IT departments to do this. Some hospitals in the UK – NHS wasn’t allowing them to move data to the cloud or outside the country but we still deployed with their datacenter with an engineer sitting in Mumbai. In a pre-COVID time, we would have flown someone down. We are using tech for its true purpose. We have seen openness, awareness, alternative adoption methods, deployment…and things have worked.

Qure AI is a great example of an Indian product company going global - what do you think is the mantra to this success? What cues can other product companies take from Qure AI?

Great question. While we have had a lot of success of India software companies like Infosys, TCS and Wipro, there are very few product software companies that have gone global. One of the first things we do well is focus. Its very easy when you’re a small company to say – “hey I don’t want x, I want y.” and the attraction is to build Y too. What many people don’t realise is that to build a good product, you need depth, you need to refine the product to deliver on customer promise. It requires understanding and feedback as well. In Qure, we have kept a narrow focus on solving a few problems at depth. It’s different if you’re a large corporation with many years of experience, Especially when you’re small, it’s very important for product companies to understand their customers, workflows and go deep into problem areas and solve them else you’re touching the surface and not solving them well enough. The other mantra for success is focus on adoption. Until I understand my customer workflow, I will not grow. I give people the analogy of Netflix – as a streaming service, if I’m not happy with it, I can shut off subscription. It’s not like a hardware product – if I buy a car, even if I don’t like it, I’m stuck with it for atleast 6-7 yrs. All our products are subscription-based software means unless we spend time getting customer adoption and if its not integrated into day to day usage, we will never get a happy customer who leads us to the next one. Those are the two very keen mantras – getting a focus, going deep in a product and making sure its getting adopted the right away. We also knew for a fact that while India has a very good name for its engineers and clinical talent, there are very few Indian AI healthcare companies and we had to build credibility which is not less than any US or Israeli company or any other company from anywhere. A lot of validation and research we did - while we started in India and a lot of our partners were Indian – very early, we started reaching out to academic hospitals in the US too – and we were able to get research and publications with top academic institutions and top publications in US, UK and western markets, which gave us credibility in the eyes of the clinicians or healthcare institutions which was not less than any company coming out of San Francisco or the Bay Area. That plays a different role – I’ve been in conversations where when I first say I’m from an Indian company, I see that look in their eye. The moment I tell them we have done this kind of validation and research with top US institutions and got published in The Lancet, they look at you differently. Building credibility in the right global forum has been key to our success.

Coming back to COVID, where do you think we're at in this pandemic? How will AI help us flatten this curve? 

Where are we on the curve? Wish I had a crystal ball. Every week I hope and pray that we’ve beyond the peak. But every time I think that, I see more cases coming up across the world. The way I see AI playing a role is – I draw inferences from the seasonal flu in the US. There is a vaccine for the flu, but that vaccine changes every year because of the strain of the virus. In our own family, we took the vaccine and two months later, my wife got the flu. The virus is smarter and a few steps ahead of the vaccine. I see the similar thing playing again in COVID. There are 300+ drug trials and everyone is waiting for the vaccine, The vaccine won’t be a silver bullet. Everywhere else, we need to see how to manage the population, how to get the economy back and running in the presence of the virus. There will be vaccine but we don’t know how effective. Will it last 3 months, 2 months or 6 months a year? We don’t know. Some of the basics of monitoring people remotely, maintaining social distancing, doing early diagnosis, treatment for the right people at the right time – already are and will remain critical to countries flattening the curve. And AI will play a role for all. It is humanly impossible to do even contact tracing- physically you cannot go to the streets of Mumbai or Delhi or any other city or go into every household, knock doors and check symptoms, and definitely not three times a day! AI can do this. You can get into longitudinal records of a patient. Next time patient comes with a cough, using AI, looking into their past health records, whether there was an incidence and likelihood of their immunity getting developed so the protocol and testing may be different. Healthcare systems are realizing you need to be smart about this. It’s a 21st Century virus and needs tools and tech to help battle this. That’s where the product we are building in terms of detecting disease, managing patients, remotely managing them will be critical components to flattening the curve.