India's big data hunt
for cures to its mental,
ageing-related diseases

One trend is important to point to here: India’s population bulge in the middle today – two-thirds of 1.3 billion Indians are younger than 35 – will change rapidly in another three decades. UN data projects one in six Indians will be over 60 years come 2050, up from about one in 20 about 25 years ago. That’s a dramatic transition in the world, especially if you consider that India will overtake China's population in less than five years from now. More on Indian demography later.

But why the focus on mental health? CBR sets the context: “...we need to understand the normal aging process in the brain, which often results in a decline of specific cognitive functions, heterogeneously in the population. Such an understanding can help promote normal aging and delay dementia, which is important considering the increasing life span.”

In developed countries, CBR continues, 10% of the over-65 years population has dementia. “The prevalence doubles every five years after the age of 60 and reaches nearly 50% after the age of 85 years. An estimated 35.6 million people worldwide were living with dementia in 2010... The numbers in developed countries are forecast to increase by 100% between 2001 and 2040 on average, but by more than 300% in India, China, and their south Asian and western Pacific neighbours.” Worse, risk factors such as hypertension, diabetes, obesity, smoking, and alcohol abuse in India doesn’t help and “results in cerebrovascular disease-vascular dementia,” CBR adds.

The Srinivasapura playbook has previous successes to build on. The Framingham Heart Study, which started in 1948 in Framingham, Massachusetts, has helped establish correlations between cigarette smoking and increased risk heart diseases, among several research milestones over seven decades. Others, such as the Religious Orders Study, which included around 1,000 nuns and priests, also involved the cohort voluntarily donating their brains after death. It opened a goldmine of insights because many brain-related studies cannot be conducted through MRI and other imaging scans.

David A. Bennett, the principal researcher behind the Religious Orders Study, is a CBR advisor. Now the director of Rush Alzheimer’s Disease Centre in Chicago, Dr Bennett has been guiding the team of researchers and scientists at CBR led by Vijayalakshmi Ravindranath of IISc. Vijayalakshmi, 64, used to chair the Centre for Neuroscience at IISc and was also the founder director of the National Brain Research Center at New Delhi between 2000 and 2009.

“Literally every drug for Alzheimer’s has failed in the last few years, we have no medicine. What we are now trying to do the world over is to try and understand how people age,” says Vijayalakshmi. “And we cannot extrapolate data from the western world to ours because illiteracy is a risk for dementia, for instance, which is a big factor for us.”

All the medical data captured from 10,000 people will be first stored in a yet-to-be-named Android app developed by the CBR team at IISc. The data collected by field agents and volunteers over the year will be channeled to a central supercomputing server at IISc.

And this is where the data sets will start getting bigger, and more complex. For instance, each individual’s data could go up to few Terabytes in size. Genetic data alone can be as big as 2 Terabytes per person. MRI data can go up to 200 Gigabytes depending how much of processing is done.

The site office in Srinivasapura is where data is first harvested. “They first give their blood, which goes to blood biochemistry evaluation, then DNA from which we do whole genome sequencing. Then they undergo all other tests including height, weight, body mass, ECG, blood pressure, we do carotid doppler test, hearing ability, eye scan, and so on,” says Ganesh Chauhan, a CBR scientist. “And then we have a lot of socio-economic data. The genetic and MRI data form the bulk.”

There are more than 30 medical examinations in all.

“We also give them words to remember and test memory. We give them an image and ask them to draw it. [There are] verbal screening, go-no-go test to check reflexes and cognitive abilities,” Chauhan adds.

All data capture will be done through a software application developed in-house, right from the time volunteers visit homes for initial registrations to complex medical tests.

With each individual’s data running into a few Terabytes, the total storage required will be in Petabytes for sure. Currently, the server room at IISc can store data up to 400 Terabytes. It will need an upgrade very soon.

“The problem is we cannot calculate it (the exact estimate) because we have to follow them next year. Other than the genetic data, which is needed only one time, the rest have to be recaptured again, every year,” says Chauhan.

Chauhan looks up to the Framingham study for inspiration. “It’s a study that opened heart to the world, and they are now in the fourth generation of the study,” he says. In much the same way, he believes such study cohorts need to go on and on, across the next generations, for them to deliver any definitive results.

So what happens to the data lifecycle after a cohort member dies?

“Death is really not a loss of data actually because if a person had biological ending the important thing for us to understand is what led to the death. That would be much more insightful and will add to the data. If you know the cause of the death and what happened leading to it, did he or she die of a stroke, vascular risk factor and so on,” says Chauhan.“For all people, we want the end point data,” he adds.

After on-the-ground volunteers have collected data, done the medical tests, and documented datasets for every individual, Kahali, the Michigan State University Ph.D., at IISc’s CBR lab gets into action.

By sifting through Petabytes of datasets over the next few years, Kahali aims to build models and develop software algorithms, which will help predict the onset of brain-related diseases early enough.

“In these kind of studies, they are also being assessed at a higher level by neurologists who are looking for signals to detect cognitive impairments. The hardest part is getting a neurologist,” says Kahali.
Her software predictors could replace the need for a neurologist in conducting cognitive assessments.

“What if i can develop some predictors that will not take so much time? If we’re interested in daily labourers in a particular village, what if they get tested and we predict (using algorithms) what could have been predicted by a neurologist.”

For Kahali, ensuring that every single piece of data is captured and shared back with her at IISc is crucial to find answers to the dementia problem.

“One thing is that you face a challenge about is the detailed look that people like us having trained are always looking for. And it’s very hard to make others on the ground, understand that. So they are like, ‘One cell of data is missing, so what?’” she says.

A public health expert says the accuracy of data capture on a project of this scale will be crucial. "Previously, people have been looking at a smaller set of population and within a narrow set of variables. A broad range of data capturing like this hasn't happened before yet in India," says Srinath Reddy, president of the Public Health Foundation of India (PHFI), a health policy NGO.

The more data sets, the more robust the model, Kahali says. “Let’s say I am dealing with only a few individuals’ data. The biggest challenge lies in the fact that I have so few actual demented individuals, whatever model I do, no matter how many simulations I do, my simulations are getting biased,” she says. “I am trying to get hold of better and larger data sets.” Kahali is a physicist who later studied computational biology.

While incubating the research project at IISc, Kris and others realised the need to build in parallel an Indian genome database because nothing of that sort currently exists.“While doing this and while looking at genomics data, we said why don’t we also kick off a genomics project in India because we don’t have a gene pool of Indians anywhere in the world,” Kris says.So, Genome India was born — its mission to gather genetic information and to catalog the amount of genetic variation among Indians.

It’s better late than never. For instance, while India accounts for more than one-sixth of the world’s population, the DNA sequences of its people contribute a meagre 0.2% of the global genetic databases, according to the BBC.

Already, over a dozen organisations including LV Prasad Eye Hospital in Hyderabad and Mohan’s Diabetes in Chennai have decided to contribute their genomics data to the Genome India pool.

“It will create data for research purpose. That can be used for genealogy, where did the Indians come from, why are the Indians more susceptible to diabetes, is diet causing a problem...all those things can be studied once this data is available,” says Kris.

Dr Reddy, the health policy wonk, calls the project long overdue given the unique and complex traits of Indians. "Everything from distribution of body fat, dietary habits, metabolism is different for Indians," says Dr Reddy, who previously headed the cardiology department of the All India Institute of Medical Sciences.How does a genetic information catalog connect with a brain research project?

“First of all, there can be two or three broad perspectives: one is from disease point of view; second is our population, how did we emerge, where did we migrate from; and third is the moment we have a catalog, it is much easier to do any kind of genetic screening or counselling because then we will know the variations,” says Kahali.

She is also going to be relying on Genome India’s datasets for training the algorithms she is will be responsible for building to help predict the onset of ageing disorders early enough.

“On CBR part, we can say that we are aiming for at least 1,000 to 2,000 whole genome sequences in the short term, and around 20,000 from across different regions and diversity over 3-4 years,” Kahali says.

Conditions such as Alzheimer’s, diabetes, liver disorders, obesity and others have a particular genetic component to them. “If we carry some kind of variation, then we are more susceptible to developing that medical condition rather than others who are not carrying that particular genetic variation,” says Kahali.

Without its own genomic database, India has been lagging.

“If we could document the entire set of genetic variations that exist in the Indian population, we will be able to associate them with different disease conditions. Eventually, that will help us in tailoring treatments towards each individual,” explains Kahali.