Every morning, across more than 54,000 schools in Gujarat, an algorithm is quietly doing something no teacher or administrator could do alone. It scans attendance records, academic scores, health data, family income levels, migration status and demographic details for every student from Class 1 to Class 8. It looks for patterns. It finds children who are drifting toward the door.
Then it sends an alert.
Last year, that system flagged 1,67,446 children as being at serious risk of dropping out of school. Every single one of them was retained. Gujarat’s state education department announced the results this week, crediting the AI-based Early Warning System installed at the Vidya Samiksha Kendra in Gandhinagar for what it called a significant breakthrough in its mission toward a zero dropout rate.
This year, the system has already identified 1,18,234 children at risk. Timely intervention measures are now underway for each of them.
How The System Works
The Early Warning System runs on a straightforward but powerful logic. It does not wait for a child to disappear from the register. It watches for the conditions that make disappearance likely.
The algorithm analyses student records including age, gender and disability status, alongside academic performance, attendance patterns and assessment scores. It also factors in school-level data such as infrastructure quality, classroom type and whether the institution is government-run, aided or private. Family circumstances carry equal weight. Economic condition, number of children in the household, parental attitudes toward education and migration history all feed into the system’s risk calculations.
When the data points converge into a warning pattern, the system generates an alert. Teachers and local administrators receive it. They act before the child leaves.
Once a child is flagged, targeted support follows. Preventive response strategies address the specific factors driving each child’s risk. A child whose attendance is falling due to a health condition gets different support from one whose family is preparing to migrate for seasonal work.
The Child Tracking System Behind The Numbers
The Early Warning System does not work alone. Gujarat’s Education Department built it on top of the Child Tracking System, an online platform developed with Samagra Shiksha that manages educational records for every student in the state.
CTS covers more than 54,000 schools and tracks over one crore students from Balvatika through primary education. Its objectives include enrolling every child in the state, identifying and re-enrolling those who have already dropped out, monitoring attendance and migration, and ensuring government education schemes reach the students who qualify for them.
Last year, working alongside the Early Warning System, CTS brought back 90,212 children who had already dropped out of school. Two systems, two different interventions. One catches children before they leave. The other finds them after they have gone and pulls them back.
The Gujarat government is now integrating both platforms so they share data and operate as a single, continuous loop. Flag early, intervene fast, re-enroll if needed.
Why This Matters Beyond Gujarat
India’s school dropout problem is not small. According to government data, dropout rates in upper primary school have historically hovered in ranges that translate to millions of children leaving the system each year, with rural, low-income and migrant families disproportionately affected.
Gujarat’s results are significant not just as a state-level success story but as a proof of concept. An AI system can be built on existing administrative data, deployed through an existing government platform and produce measurable, human outcomes at scale. No new hardware. No experimental technology. Just data that was already being collected, finally being used intelligently.
The real test is replication. If the model that retained 1.67 lakh children in Gujarat last year can be adopted by other states, the number of children it protects could scale by an order of magnitude.
