Techtonic 2018

Health 2.0

Visit’s AI-driven chatbot is looking to change the way patients consult doctors

People will likely be more truthful with a bot than a person,” wrote Biz Stone, the co-founder of Twitter, Medium and Jelly, in an April 2018 blog post, soon after investing in Delhi-based chatbot-based health advice app, Visit. Stone’s blog is aptly titled, “Why I’m investing in Visit?”

The two-year-old start-up, founded by four batchmates from the Birla Institute of Technology and Science (BITS), Pilani, is an attempt to democratise healthcare and make up for the shortage of doctors, by using artificial intelligence (AI). 

According to the World Health Organisation, the world suffers from a shortage of at least 7.2 million healthcare workers. Stone writes, “In India, for every doctor there are 2,000 patients lined up in clinics, waiting for hours. Accessibility to quality health advice is an overwhelming problem in a country with over 200 million people affected by lifestyle problems such as stress, chronic conditions, obesity, skin conditions, and more. This is where the technology approach by Visit comes in.”

But before we get to the nitty-gritty of Visit, let’s understand what it does. Users interact with a chatbot by adding their symptoms and the chatbot follows up with a series of questions. These questions are generated by a machine learning model, trained using a knowledge base of probabilistic relationships (over 20,000) between variables such as conditions, symptoms, risk factors and past history.

The app also offers users a free chat with a general physician for initial guidance and if required, the Q&A-based chat records are made available to specialist doctors for advanced consultation over a video call (which is a paid feature). The chatbots work in unison with the doctors. Almost 50% of queries are addressed at the chatbot level. 

Visit was incubated by MapmyIndia in Delhi in January 2016, and besides Biz Stone, it has also got the attention of other investors such as Kunal Bahl and Rohit Bansal (founders of Snapdeal), Muthu Murugappan (Muruguppa Group), Alagu Periyannan (co-founder of the US-based BlueJeans Network) and Karthee Madasamy (former MD of Qualcomm Ventures).

Instant consulting

Visit is the brainchild of Anurag Prasad, Chetan Anand, Shashvat Tripathi, and Vaibhav Singh. “We were all part of Centre for  Entrepreneurial Leadership at BITS,” says Anand.

The club members would toy with start-up ideas. Around March 2015, a swine flu epidemic broke out in some parts of Rajasthan and Haryana. “There was a lot of panic in the campus. That’s when we started a toll-free number. We got a few doctors, who were parents of our friends, to interact with the students. This helped in reducing the chaos,” reveals Prasad.

That is when they realised healthcare is not just about visiting the doctor in person because in Pilani, the resources were scarce with just one clinic and a doctor. “Learning from that experience, we started building a virtual care model. In the summer of 2015, during a two-month vacation, we built a prototype,” says Anand.

They conducted a market research and met doctors. Around this time, they met Rakesh Verma, founder of Delhi-based MapmyIndia at a BITS Pilani start-up event, who liked the idea and the team. “They were in the final year of college. I really felt these guys could do something. I am from the same alma mater, and I, too, went through that phase. What mattered most was that the guys appeared genuine and hardworking,” says Verma. He invested an undisclosed sum as seed money and also provided office space to Visit.

They started with a psychologist-on-call function. “At first, there was no chatbot. We thought people needed emotional help and support with privacy,” says Anand. But within two to three months, they recognised it wasn’t good enough. “We realised that users were looking for a service that provided them round-the-clock health advice. Every day there are 250 million health searches and Google gives you scary search results. A simple chest pain can pop up as a symptom of serious heart problem. A chatbot can distinguish between the two with the help of follow-up questions. Our vision was to make Visit the first port of call for any health issue — not Google or your friends who would provide vague answers,” says Prasad.

So a chat service was launched in the app. A doctor would be available throughout the day for users with a two to three minute response time. By June 2016, the app became a doctor-chat and appointment platform.

But the patient-doctor chats were leading to something more interesting — a possibility to introduce AI. “For a year-and-a-half, we ran this chat service that generated about 250,000 chats. We gathered 2.5 million messages. Using this data, we built the chatbot and rolled it out in February 2018,” informs Anand. The chatbot helps in diagnosing nearly 800 ailments (with 1,600 symptoms) including hypertension, cardiac and skin problems.

With this AI solution, Visit aims to address a major structural issue: there are only 600,000 doctors to attend to 1.3 billion Indians. Average consultation time in India is two minutes, for which one needs to wait for 40 minutes outside a clinic. “There are two things here: first, the user understands his problem better, and second, the data generated by the questions is given to a doctor. The bot is doing 80% of the work. Hence, the conversation starts with detailed questions rather than basic ones. The whole concept is geared towards helping doctors be more efficient,” says Prasad. Stone feels that Visit facilitates a symbiotic relationship between AI and doctors, making a positive impact on patients.

Meanwhile, over the past year-and-a-half, users had been asking similar and repetitive questions. So, there were clear patterns lending themselves to machine learning. “We were able to create a decision tree. On top of this, we used medical journals to supplement this effort. Doctors helped us develop correlations between symptoms and diagnoses,” says Prasad. 

The chatbot can ask you 30 questions, at the end of which it presents the medical condition with the probability. If the results are inconclusive, the chats are sent to a doctor for a second opinion. “Something I’ve learned from working at Google and from my own search service, Jelly, is that the majority of the queries are the same. We’re all just people — the same stuff happens to all of us, over and over again. Visit’s AI will learn from the interactions between patients and healthcare professionals and continue to get better at diagnosing common ailments on its own,” writes Stone in his post.

Visit started with 50 empanelled doctors two years back, but they already have 2,200 doctors on board today. Apart from the panel, there are 14 in-house doctors, who offer their services on a full-time basis. The co-founders understand that the quality of consultations can make or break their business. “We treat doctors as our partners. We offer them software and empanelment for free and provide them with patients. They usually charge between Rs.300 and Rs.700 per consultation. We track everything for quality and have a lot of checks in place, such as notifications, consultation duration, response time, and patient review and feedback,” says Anand.

Sustainable model

Instinctively, Visit appears as a clear B2C business, but it isn’t yet. The wise advice of their incubator, Verma, helped the young founders tread on the right path.

“Fundamentally, though it lends itself to being a B2C business, everybody can’t be a Flipkart or an Ola which can keep bleeding, and yet invest $20-$30 million in marketing and customer acquisition. At MapmyIndia, we relied on the B2B route first, reached out to customers through automotive companies, and then took the B2C plunge,” says Verma.

While the app is open to use for any customer, most of their customers come from the B2B side for now. “We work with corporate and health insurance companies which offer Visit to end users and employees as an online outpatient service”. Visit counts companies such as Tata Metaliks, HDFC, Rolls-Royce, Times Group, Wipro Consumer Care, Religare Health Insurance and Apollo Munich as its clients. It currently has about 130,000 users from about 7,000 two years back. Visit tailor-makes the product to suit the client requirement and charges on a monthly basis for unlimited usage by employees and their families.

“Most of our customers are in cities such as Mumbai, Bengaluru and Delhi. Once you have a corporate user endorsing the product, you can proliferate better. Customers in metros are early adopters of technology-based products and we are reaching out to them through their employers,” says Prasad. Given the initial success, Verma feels the time is right to start Visit’s branding process. He is currently advising the team to raise about $2 million over the next year to help launch a brand campaign as a prelude to their foray into the B2C segment by the end of next year.

The founders say, thanks to their focus on the B2B segment, they are already making money. “We have a healthy contribution margin. We are not profitable at the net level because we are investing that money for growth. You need to choose at what scale you want to become profitable. Right now we are investing in making the product better,” says Prasad.

Snapdeal’s Kunal Bahl and Rohit Bansal participated in the last round of investment. Bipin Saha, who represents the duo of Snapdeal as an observer on Visit’s board, says, “We saw value in the solution they are offering for an existing problem. Firstly, people in small towns don’t have a good cardiologist, but now, even they can consult a well-known cardiologist via a video chat. Secondly, in cities, people are so busy that they don’t have time to visit the doctor. So, Visit functions like an OPD.”

Further iterations

Any app platform would ideally like to have daily visits from its users. But the need for a doctor consult may not be that often. So, how does Visit tide over this constraint? The team is trying to build chatbots that help with nutrition, fitness, and wellness on a daily basis. Currently, these features are being bundled with the overall package and sold to corporate clients, which increases the overall engagement. “We are working on conversations with chatbots for chronic conditions and lifestyle management after tracking how well you sleep and what your activity levels are,” says Anand.

With chatbots, they want to offer preventive care solutions also. The idea is to warn you in advance that you are not sleeping well or not active enough. “A lot of your phone data is tracked on Google Fit. We take permission to use that, since Google is already tracking your sleep and activity data,” says Prasad.

Visit faces competition from other apps such as Lybrate and DocsApp, which offer online doctor consultations and patients are charged per consultation. But Visit is the only app that offers chatbots for consultation. While the business model is still evolving, the founders are confident that they are moving in the right direction. “We have evolved from 50 chats a day about two years back to 1,000 chats a day,” concludes Prasad. Looks like visiting the doc has become just a tad more easier.