Written By: Dr. Shivani Rohilla | Updated : February 5, 2026 3:52 PM IST
Medically Verified By: Dr. Shivani Rohilla
Children with autism also often appear unaware of others' feelings and avoid social interactions.
Autism Spectrum Disorder affects millions of children worldwide, but for many families, the most difficult part of the journey is not the diagnosis itself; it is the waiting.
Parents often sense that something is different long before any professional confirms it. A child who avoids eye contact. A toddler who does not respond to their name. Speech develops more slowly than expected. These signs are subtle, easy to overlook, and often easily dismissed. Families are told, "Every child develops at their own pace." So they wait. They hope. And they keep waiting.
By the time clarity finally arrives, months or even years may have passed. That lost time matters. Early support can significantly improve communication, learning, and emotional development. The delay is not only a medical concern. It becomes an emotional burden for parents, a source of confusion for families, and a systemic challenge for healthcare.
I remember speaking with a mother during my research who quietly said, "I knew something was different, but everyone kept telling me to wait." By the time she received professional clarity, her greatest regret was not trusting her instincts earlier. That conversation stayed with me long after the meeting ended.
This gap between early concern and late clarity is what shaped the direction of my work.
Over the past six years, my research in artificial intelligence and neuroscience has focused on one question: can we design technology that helps clinicians notice risk earlier, instead of confirming it only after long delays?
The intention has never been to replace doctors or psychologists. Human judgment is irreplaceable in healthcare, especially when children are involved. But there is space for supportive tools and systems that can assist clinicians, highlight patterns, and offer earlier signals when something may require closer attention.
One reality we must acknowledge is the uneven access to specialists across India. In many Tier-2 and Tier-3 cities, trained child psychologists and developmental experts are limited. Waiting periods are long. Families who are already uncertain are often left without timely guidance.
Most AI systems in healthcare today are built to deliver a final prediction: yes or no, risk or no risk. But autism does not appear suddenly. It develops gradually. Patterns evolve over time. Clinicians themselves observe repeatedly before reaching conclusions.
That understanding shaped how I approached my research.
Instead of designing static prediction models, I explored approaches based on reinforcement learning, where systems learn progressively from patterns and outcomes. The idea was to allow the model to adapt, refine its understanding of complex behavioural signals, and recognise when early warning signs begin to emerge. Not a diagnosis. Not certainty. But earlier attention.
If such systems are developed responsibly, they could encourage families to seek professional guidance sooner. They could support clinicians in monitoring more closely. And they could reduce the long periods of uncertainty that so many parents currently experience.
It goes without saying that using technology in child healthcare requires tremendous caution. Human monitoring, transparency, clinical validation, and ethical protections are crucial. In such delicate situations, AI should never make judgments on its own. However, when carefully crafted, AI can enhance human judgment rather than take its place. After years of working in this space, one belief has become very clear to me:
AI in healthcare should not be about speed or automation. It should be about timing, support, and compassion.
If we build these systems responsibly, they can help shift the experience for many families, from delayed reaction to earlier care, from confusion to clarity, and from isolation to support.
