“When you have lots of information, there is a pattern that is evident in that data and that pattern might not be easily understood by the human mind because there is only a limit to which we can process information. So, what we are aiming to do is analyse the huge amount of data that is available here.”
Dr Anthony Vipin Das is an ophthalmologist by passion and an innovator by choice. He is a consultant ophthalmologist, comprehensive ophthalmology and cataract service at L V Prasad Eye Institute (LVPEI), Hyderabad. He has presented widely both on national and international platforms. He is team principal and chief architect of eyeSmart EMR, an award winning electronic medical record and hospital management system developed in-house at LVPEI. The system is currently operational across 174 centres in the LVPEI Network.
Dr Das is a member of the International Task Force for Emerging Technologies for Teaching and Learning at the International Council of Ophthalmology (ICO) and is passionate about developing meaningful ophthalmic educational content on the internet. He has served as an advisor on healthcare innovation to the Ministry of Health, Medical & Family Welfare, Government of Telangana. He is a TED Senior Fellow and is named among the Top 35 Innovators under 35 years of age in the world (TR35 2012) by the Massachusetts Institute of Technology, US.
In an interview with India Medical Times, Dr Anthony Vipin Das talks on behalf of the technology and innovation team of LVPEI.
Before we go into the main part, please introduce yourself doctor.
I’m Dr Anthony Vipin Das, consultant ophthalmologist at L V Prasad Eye Institute. I’m also associate director and head of technology & innovations strategy for the institute.
Artificial intelligence has penetrated almost all the fields today, but using it to fight blindness is quite new. Please elaborate on this part.
To start with the explanation about what we do here, it’s not exactly artificial intelligence in general but we are specifically dealing with big data and machine learning. We are not trying to use another intelligence to help us but what we are trying to do is — identifying patterns in data. When you have lots of information, there is a pattern that is evident in that data and that pattern might not be easily understood by the human mind because there is only a limit to which we can process information. So, what we are aiming to do is analyse the huge amount of data that is available here. At LVPEI, over the past seven years, we have about 3.5 million consultation records. In that big data set, we are trying to identify the patterns of diseases in terms of progression or surgical outcomes that are modifiable. What we are trying to do is make our delivery of care much more precise and much more better. So, I would say artificial intelligence is a very broad term which might not be suitable here but we are specifically looking at deep learning and machine learning for analysis.
So, can you give an example for the type of surgeries?
We have shared some examples in the public space also, in the past year. So, the first most successful modular that was found in our clinic is the prediction of refractive errors in children. The first question that any parent of a child with refractive error asks the doctor is: will the power change; will it increase or decrease? So, what we are doing is that we are predicting how the power will change in the next two years based on the patterns that we are getting on the existing database. We are seeing patterns that exist in the data collected in real time and we are using it to train the model in real time and predict the data that would occur next or the possible outcome. The other successful example is the prediction of refractive surgery’s outcome. This helps the surgeon on the table to make the necessary modifications to make the outcome better. And also in chronic diseases like diabetic retinopathy and glaucoma, we are trying to see how the disease will progress.
What are the potential advantages that you feel is there in clinical research using AI in comparison to the conventional methods?
Conventional methods are published research where people are sharing information. It basically gets published and it has to get written down in textbooks and then you get the benefit of that research when people buy the books and gain that knowledge. Here, it gives us the opportunity to learn in real time. So, there is a lot of data, we connect it to the machine and the machine analyses it in real time and we get the results. So, this is very different, more real time and more relevant.
What are the disadvantages in this?
No. I would not use the word disadvantages. I would only say that only time will tell the value and validity of the model. The machine is telling you a prediction and we are also validating it in longitudinal correlation. This is more important and this will bring out whether the prediction is good or not.
What are your future prospects in this venture?
I think we are basically starting to change the way we look at research and trying to ensure that people have access to information in real time and also use the information to review with the past data. All of these things will actually help in the delivery of better and efficient care and improve the treatment and management of ailments.
Do you think unsuccessful surgeries can be prevented by these early predictions?
We will be able to identify the people at risk and probably do some therapies and modalities to prevent the severe complications. About surgeries, we will be able to change the precision with which a surgeon has to operate a particular patient. This is not yet live, still in production.
Cataract is the most common ophthalmic disorder prevalent, what are the prospects of this in cataract surgery?
Prediction of cataract surgery outcomes is also an area which we are working on. We are still exploring the data in that.
Electronic record and data management system that you are developing, its use is not just limited to ophthalmic issues, right? Primary care and other sectors can very well use it, what is the potential of this system in other clinical fields?
At the end of the day, it is all about having structured information and data, which might have various uses. We can use in cases like oncology, gastroenterology, studies on aging. We need to identify the current patterns and apply what we find in the clinical use. If we don’t apply it clinically, none of what we do would make sense.
What is the position of this venture in comparison to the global standards?
It’s not about the comparison with the global standards; it’s a very new thing — like a paradigm shift in research. So, it is not going to be compared. Moreover, we are in an era where people want more information in real time like internet, social media etc. Like everything is instantaneous, what we are trying is to use the instantaneous data in clinical and research sector.
What would you like to tell the doctors reading this?
I would only say that, discipline in documentation and transparency in maintaining records is very important. The discipline of following up a particular patient with the findings is very crucial for understanding the response to our treatment methodology and progression of disease. For young doctors, the most important thing is the discipline of clinical documentation — be it on paper or through electronic means. Once the information pool is there, we can use it effectively.
One last question sir, how did you come up with this idea?
No, it was not a Eureka moment. It was more of an evolution of our works. Of how we had expanded our electronic medical record systems, was our start. It was operational in 174 centres, which basically provided all the information that we are working with now. We realised that with such a huge amount of data, we could actually do something useful. That is how we started the process of machine learning for our models.
by Usha Nandini
|More from INTERVIEWS|