The importance of data in powering AI for healthcare

During a PCQuest HealthTechTalk on "Key role of AI in healthcare", Sunil Rajguru, Editor, PC Quest, Dataquest, and CiOL, said that healthcare requires greater technology, no-touch technology, etc. Universal healthcare should come to India within this decade. AI can be the central authority that can make it part of universal healthcare in future.

Reshma Suresh, Head, Business Operations, Qure.ai, said many HealthTech companies came forward during the times of Covid-19. We are an AI-based company. We have solutions that can look at clinical findings. We also repurposed our existing software and quickly rolled them out. We worked with several government and non-government institutions. We look at X-rays, and can look at Covid-19, using them. In Mumbai, there are stadiums converted to quarantine facilities.

Earlier, PCR kits were also not available. Lot of optimization happened. There were X-ray machines in large slums. X-rays were taken of patients at frequent intervals. Rural India was largely ignored. VCs also created funds, and came forward and started supporting institutions like ours. There were Telegram-based apps, as well, that helped the healthcare people.

Baljit Singh, Co-founder and CEO, Simbo.ai, said we have seen the advancement of AI-based diagnosis over the last two years. Putting medical diagnostics into rural settings was a problem. We needed to figure out who needed attention. AI takes over and tries to help where people cannot be put in place.

Subhashish Banerjee, Co-founder and CIO, ARTPARK, said we launched X-ray Setu. To create any kind of AI, you need to have high-quality database. We built the X-Ray Setu based on the database. To get even better accuracy, we need to build databases in the country. Once I have the initial infrastructure, the marginal cost for an X-ray is hardly anything. Talking about rural healthcare, the marginal cost is very little.

Looking at radiologists in India, they currently don't exist, and are mostly based in urban areas. Radiologists, and anesthesiologists are required. It is about cost and availability. We can take healthcare to third-world countries if we can bring in AI. India is a diverse country. Thakurs in UP and Brahmins in Tamil Nadu are genetically diverse. If we develop AI using diverse database, that will work very well. India has high-level talent and diverse data. Healthcare revolution can take place in the future.

Arun Mallavarapu, Co-founder and CTO, Fedo.ai, said we are not looking at imaging and radiology. Looking at the implication of AI on healthcare, the focus is on the diagnosis side. We have to focus more on customers. What can the consumer get without visiting the doctor. Over time, we need to reduce cost and pressure on the healthcare systems in the country. India can be an AI-entity. We need to be business- and consumer-centric. Adoption of AI in India is going to be low. Once, the consumers start adopting AI-based processes, things can change. AI needs to focus more on consumers, where we can tell them you can take care of yourselves using A, B, and C, etc. AI in healthcare has been there. We need to look at where we focus.

Rajguru asked whether there were any misses that happened?

Baljit Singh, Simbo.ai, said one area we need to look at is digital data. If one wants to build solutions to have a look at patients and say there could be future trouble, how do we get that data? This is one area missed for EMR and other perspectives. We have created voice-based assistants for doctors. Someone has to work toward fueling that data. This is a biggest miss that India had, while some countries have mastered.

Mallavarapu, Fedo.ai, said health data is going to be a goldmine if it is successful. In our academics and medical schools, we are not seeing knowledge being built. Adoption and finding use cases where AI can work is important. Doctors can create use cases and feed them back to data scientists. That will make the adoption of AI much faster. We should catch up.

Banerjee, ARTPARK, said Harvard medical students can take part in MIT, and vice-versa. We are building a school. IIT Kanpur has also announced plans. We can also have data on vaccines. We can do it in India. We can get global leadership in healthcare data. We need to create databases. Doctors also cannot create databases. Data has to be done properly so that the private sector also gets to use that. In the early days, we needed to follow protocols. Cultural and political systems also need to gear up.

Reshma, Qure.ai, said, we need to start thinking about creating databases. We can do things differently. There are technologically-driven tight contact tracing activities. Alerts come about contacts. We could have done this. In Oman, there are wearables that track the location of contacts. They ask questions about medical conditions regarding their health. They also alert the healthcare systems appropriately. If we had brought in something like this, it would have helped. We are talking about booster doses now. We can also monitor vaccine side effects. India is a vast country. There is no one size fits all model. We also need to have interconnected mechanisms in India.

Pitfalls of AI

Banerjee, ARTPARK, said any cultural activity can predict things much in advance. Things do happen! Youngsters today are more comfortable with AI. We also brought a bunch of robots for kids, who said they have seen better versions in Hollywood movies. Now, healthcare is an important sector. Pace will be slower, and regulation will be there. If you have used AI in your publicity, the stock price may go up.

Reshma, Qure.ai, said that she has been inspired by Hollywood movies. Digital assistants today do the same thing. They control home automation, etc. Technology is designed to make human lives easier. Amazon has a completely automated warehouse. It has improved the quality of our lives. AI brings limelight! It is accessible to a lot of people. It is easy to build an AI algorithm. We need to have peer review publications. We need to factor in medical devices and whether they have been peer reviewed. We need to adhere to all the factors, and that can be challenging.

Adoption is another challenge. Specialists also do not wish to adopt AI. AI and humans can achieve a lot together. Innovators need to be responsible to help and build trust. We can also look at AI to enhance the specialists.

Baljit Singh, Simbo.ai, said there are movies that show AI can help. There is also a fear if AI takes over. It can replace jobs and create disasters. AI is highly mis-rated. AI is helping reduce costs and take things to the next level. AI is nowhere close to near-human intelligence. We need to create awareness around AI. There should be a lot more responsibility layers that can be put. We can gain the trust of more people.

Mallavarapu, Fedo.ai, said there has been some misuse of AI. The abuse has been so much. We need to be more transparent about what we are doing. It has remained a black box for many cases.

Parting shot

Banerjee, ARTPARK, said nurses give medicines to patients. There needs to be jobs that can be outsourced. Others can do many new things.

Baljit Singh, Simbo.ai, said there needs to explainable AI in healthcare. That's the next step in AI. Data is also important for innovation.

Mallavarapu, Fedo.ai, noted the biggest use case can be meaningful use of AI. It should be used ethically in healthcare systems.

Reshma, Qure.ai, added we are now seeing an integrated screening and prevention approach. We need to have an approach where we are screened for everything. We need government, technology partners, and academia to come together. We need to have one-stop screening.

Check out the complete panel discussion...