There are justified confusions and concerns about the influence and the impact of Artificial Intelligence on human society. But AI can play a significant role in the area of healthcare and diagnosis.
Consider the case of Autism Spectrum Disorder, a lifelong condition that impacts how a person interacts with others and perceives the world. It is characterised by difficulties in communicating socially as well as a tendency towards repetitive behaviour. The condition affects how a person learns, pays attention, and processes sensory information. Since autism is a spectrum, its impact is different in different people; some people require significant support, while others have milder symptoms.
Current data estimate the prevalence of ASD in India to be approximately 1-1.5%. In West Bengal, school-based studies around Calcutta have reported lower diagnosed rates (~0.23%) although this is considered to be an underestimate since many children remain outside the school screening system. Again, the gap between the number
of children needing support and the number of professionals available is one of the
biggest healthcare challenges in West Bengal and India at large.
India has a deficit of more than 20,000 clinical psychologists. In West Bengal, the vast majority of RCI-registered licensed clinical psychologists are concentrated in Calcutta. Moreover, special educators mostly work in private special schools in cities. Rural schools, under Samagra Shiksha, often share one special educator across a cluster of institutions. The burden of screening usually falls on anganwadi and ASHA workers in rural areas. While excellent in tracking nutrition and immunisation, they generally lack rigorous training to identify the subtle, early signs of autism. This leads to children being identified only when they have severe behavioural issues or speech development delays.
Consequently, the current model in India is shifting towards ‘Parent-Mediated-Intervention’. Since there aren’t enough therapists, professionals train parents to become the primary therapists at home.
AI can become a transformative partner for the autistic community, offering tools providing accessibility, personalised support, and non-human emotional support. Moving beyond a one-size-fits-all approach, AI can help bridge the gap between neuro-diverse and neuro-typical experiences in different ways. AI tools can analyse subtle behavioural biomarkers — like gaze patterns, vocal prosody, social responses and motor movements — from simple home videos and games. Tools like ‘eye-tracking AI’ can identify social attention differences in toddlers in a very short time, allowing for early intervention. AI-based games can also leverage machine-learning to analyse behavioural patterns, which are too subtle for the human eye during a clinical visit. Moreover, these tools are designed to be engaging for children while providing objective data to clinicians. Crucially, humanoid robots can act as non-judgemental therapists interacting with children with endless patience. They can repeat social scenarios hundreds of times without getting tired, helping children practise eye contact and emotional recognition. Present-day Gen-AI models can also act as social translators to rephrase confusing instructions into clear, literal steps, thereby reducing the cognitive load for children with advanced autism.
But these tools need to be rooted in the linguistic and the socio-cultural milieus of the region and the community where the screening task will be carried out. They need to be affordable; besides, grass-root-level workers need to be trained to use them to acquire data and for analyses. A few institutes are working on these challenges; but they call for scale with appropriate start-up interventions.