Closing Bangladesh's early-intervention gap with AI
Specialists are concentrated in a few cities, while most children live far away. Here is how an AI-first, phone-first model can reach every district.
Bangladesh has made real progress in awareness of autism, Down syndrome, and developmental disabilities. Yet a stubborn gap remains: the country has only about 0.9 speech-language therapists per million people, concentrated in a handful of cities, while most families live far from them. For a child in a rural district, the nearest qualified assessment can be a day’s journey away, and 66.3% of rehabilitation spending comes out of families’ own pockets.
A gap measured in distance and time
A manual screening takes a specialist 30–60 minutes, so waitlists grow and many children, especially in rural areas, remain unassessed simply because a screening is out of reach. A national study of 38,440 toddlers found autism in 1 in 589 children, so the need is measurable at population scale. And UNICEF reports that 60% of children with disabilities aged 5–17 are out of education. The result is the same in every case: lost months during the years that matter most.
0.9
speech-language therapists per million people in Bangladesh
Why an AI-first model fits Bangladesh
Almost every family has access to a smartphone, even where they have no access to a specialist. That single fact makes a phone-first, AI-assisted model uniquely suited to Bangladesh. A guided screening in Bengali can produce a preliminary picture for a clinician to review, and a tailored, clinician-approved plan can reach the family without anyone leaving their district.
- Screening starts on a phone, in Bengali, in the family’s own home.
- A clinician reviews every result. AI is decision-support, never autonomous diagnosis.
- Home practice and progress flow back to the therapy centre between visits.
- The care model is built to extend, district by district, toward all 64.
Technology with a human purpose
AI does not replace clinicians. It extends their reach. By supporting the parts that create the longest queues (screening, planning, progress tracking), the same number of specialists can assess more children per day. Speech AI’s committed validation target is a screening-time reduction of at least 30%.
That is the promise behind Speech AI, built with clinical partner NeuroSparks and academic supervision from the University of Dhaka: a service that meets families where they are, so geography is no longer the reason a child waits for care.
