Published: 20th February 2021
This Noida student's algorithm can help detect Down Syndrome using a child's photograph
Saanvi Mehra has created an algorithm that detects Down Syndrome through visual parameters such as distinct facial features
Around 30,000 children in India are born with Down Syndrome. What's more alarming though is that unlike many other countries, the fatality rate of Down Syndrome children in India is much higher. This is mostly due to lack of awareness, diagnosis and treatment.
Several diagnostic tests are available only in big cities and even if they are, they are quite expensive and not affordable for lower-income households. Saanvi Mehra, a Class X student realised that her family was among the privileged lot when her mother was pregnant with her younger sister and the doctors told them that the baby was at risk of having Down Syndrome. Although the tests results came out negative later, the incident stuck with Saanvi and she wanted to do something to help mothers and families who may not be as privileged as her.
Considering the fact that in India, penetration of phone services is much higher than basic healthcare, she decided to go the digital way with her diagnostic support. She enrolled her project at the Google India Code to Learn Project — which is basically an algorithm that detects the possibility and risks of a child having Down Syndrome across races with accuracy.
Saanvi, who studies at the Shriram Millenial School, Noida explains, "It's an AIML (Artificial Intelligence Markup Language) engine using the Google Cloud auto ML, which is Google's AIML engine. So you take photographs of the child and feed it to the system and it can determine whether the child has Down Syndrome through visual parameters, because children having Down Syndrome have distinct features like narrow eyes and flattened noses. I set some parameters, adjusted the sensitivity of each feature and then tested it on the photographs." The algorithm works best on toddlers under the age of 5, but it is also applicable for older adults as they tend to have the same features.
Saanvi came to know about the contest from her computer teacher way back in 2016. But it was during the lockdown this year that she managed to find some time to study some AI processes. She then related it to her personal experience and built this project. Recalling all the incidents, she says, "During my mother's second pregnancy, fortunately, we were living in a big city and we were able to afford the necessary tests. However, I know that that's not the case for everyone and this being very personal to me, it really struck me as something that was worth raising my voice for. Early detection and treatment of Down Syndrome could change the lives of so many children in India."
The algorithm is not available to the public now, but Saanvi plans to work with doctors and hospitals to test the clinical validity of it, and ultimately she hopes to roll out a mobile app that's accessible to everyone. In the future, Saanvi hopes to work in an intersection of both healthcare and technology. She wants to make a similar algorithm that can detect the risk of miscarriages.