Near policy paralysis in Indian healthtech, says doctor who developed country's first physician chatbot 

In private practice, I witnessed the problem of late diagnosis. Some of my old friends would call me asking for advice if somebody in their family had a stroke, for example. There were instances when such stroke patients would be in local hospitals for days and families would not know that such patients need to go to a neurosurgeon within three-four hours.

So I realised that lack of access, affordability and awareness are major problems in the healthcare domain. Because of these things, people go to a local chemist or a quack, which leads to mistreatment, and misdiagnosis as well as delay in getting the right treatment, which can lead to disabilities and deaths.

Then I started thinking that there should be something or someone who is able to give advice to people on their health in a timely manner. There are very good doctors around and there is telemedicine too.  But there is only a limit a human doctor can go to, there are limited numbers of hours a human doctor can commit to their duty or their work. So it is nearly impossible that all people would be able to get the right advice.

So I thought that if I can create a virtual agent of sorts that would be able to help people understand what is wrong with them and guide them on what they should do next, it would be beneficial.

After doing two-three startups in healthtech, I met the right people, built the right team, and we were able to commit our resources, our knowledge and our skills to build an innovative solution. That’s how Zini came about, to help people evaluate their symptoms, get answers to their health concerns and get the right advice in a timely manner and be guided in the right direction from the very beginning. It is aimed at helping them make the best decisions for the best health outcomes.

How exactly can Zini assist doctors and patients?

One format is in hospitals or clinics where information is collected from the patients when they are in the waiting area before seeing the doctor, and the doctor is able to see them for some time.

Another format is where we give this to telehealth platforms or e-health platforms. In that case, the patient is advised based on their symptoms and redirected to a doctor.

With HDFC Bank, under a CSR (corporate social responsibility) initiative, we've started Zini clinics which are nurse-led. There, the nurse is able to gather a patient’s vitals using medical devices. But Zini is able to have a thorough conversation. So one preliminary report is created by the bot itself to guide the patient. But later on, through telemedicine and video calls, connect with real doctors as well. And they get assigned a prescription from doctors within five to 10 minutes. So these are like different formats, applying the same technology.

Also read: India has more doctors than WHO standards, says Centre. Is that a valid claim?

What more does Zini offer than, say, search over the internet?

Over the internet, it's a passive, one-way input of information from what you feel is important from your side. If you have noticed, if you are telling your symptoms to a doctor, sometimes they will interrupt you and that’s because the doctor knows what is relevant.

Similarly, Zini is a two-way conversation and she will ask relevant queries. For example, people sometimes start with very vague symptoms that are actually not the real cause of concern. The chief of Google Fit had himself written that it is a bad idea to Google your symptoms and instead, people should go for this new aid of symptom evaluation, bots or symptom checkers that are now coming up.

On the internet, searching for a disease based on symptoms is like a lottery ticket, you may end up at the right spot but you may end up at a very wrong spot as well.

Now that the chatbot is ready for use, what is on your radar?

We’re now trying to teach Zini empathy, which technically speaking, is cognitive behavioural therapy as patients also have emotional needs. We are also trying to build personalised health recommendations. And then, since we are attaching every patient to a global health ID, or a 16 digit GHID, so in the Android app, we collect their health vitals, the people store their health records, they share their symptoms and health history.

Over time, we want to give personalised predictive health analytics to people. They can have a personalised health system and personalised suggestions or recommendations. For some diseases like obesity, diabetes, hypertension, people get standard protocol treatment, but there are diseases that require personalised treatment, dosage and attention. So I think with data and AI, over the next 8-10 years, personalised care for specific ailments may be possible. As more research and data is published, governments may even allow primary care treatment by bots in some countries.

Are you also planning to tie up with hospitals to expand its usage?

We are in mid-stage talks with Paras Hospitals in Delhi. The intent of the Zini clinic initiative is to screen cases and redirect them to the right doctors through telemedicine or tie-ups with hospitals. Hospitals, in such cases, will have the data from the very beginning.

Let me share an interesting statistic. In India, a recent study showed that the average time to diagnose a tuberculosis patient from their initial symptoms is 60 days. We are here to fix these major gaps.

As we screen patients, it can help government bodies understand what are the health burdens in specific areas or age groups basically and allocate resources accordingly. We are also looking to partner for government programmes such as Ayushman Bharat and eSanjeevni (Centre’s telemedicine project).

Also read: Private healthcare suffers trust issues in India, we want to change that: Himesh Joshi, CEO, Ayu Health

In a country like India, do you think there would be acceptance and comfort level with a chatbot for patient interaction?

As of now, some Zini clinics are operational through CSR and therefore, patients are not paying us anything. They are getting an evaluation done by a bot and then talking to a doctor on video call but are really happy. In these clinics, nurses thoroughly check patients and ask them every detail about their symptoms and later they get a call-back from qualified doctors.  So far, all our patients have given us very positive and happy feedback.

But beyond this, heavy technologies in rural or poor socio-economic demographics will need a CSR or government input for sure. We have some plans where we integrate this into existing health setups like dispensaries or local chemist shops. But then again, the poor guy would want to save as much money as they can. So I think the best way would be CSR or government support in this domain.

What do you think about government intervention in the health technology space?

 There is no doubt in saying that government policymaking is very slow. There are four pillars of health-tech sector: provider or doctor, policymakers, payer or insurer and patients who should be the main focus. But, sadly, everybody is in a silo and nobody wants to lose their power.

So doctors are always under this assumption that some technology would, maybe, reduce their value. Many, not all but many, are apprehensive about these innovations in the sector.

Policymakers, on the other hand, are afraid that if they make some policy, maybe the insurance providers or the doctors will get upset. So there is a sort of policy paralysis there. It was only during COVID-19 that guidelines for telemedicine were released in 2020 while e-health platforms like Practo have been there since 2010—so they were technically not legal before 2020.

There is no guideline yet on AI solutions in healthcare but AI solutions themselves require very thorough policies. In the new guidelines on telemedicine that are yet to be public, there is probably a line mentioned about AI.

I would say the policymakers are very slow; there are either no policies or they are not detailed enough. Also, government bodies need to be active on issues like how data has to be used, secured and how solutions targeted at different groups should have checks and balances in place.

In the US, for instance, the Food and Drug Administrator has labeled AI solutions as SAMD (software as a medical device).  So they call these AI solutions software that are acting as a medical device on patients and based on their criticality or usage, some solutions are auto-passed, some solutions require some scrutiny and others require very intensive scrutiny. All this requires very thorough policymaking which is not happening here, to be honest.