A passing grade is a lagging indicator
A grade is a summary that arrives after the grading period is over. A student can clear the passing line while quietly losing the foundational skills the next grade level assumes they already have. Across Philippine basic education, the share of learners rated proficient falls from 30.5% in Grade 3 to 0.47% by Grade 12 — a slow slide that rarely shows up clearly on any single report card.
By the time grades are computed, the window to act has usually closed. The evidence of a slipping student is real — a dipping quiz score, a missed competency, a reading passage that took twice as long — but it is scattered across separate quizzes and gradebooks, and Filipino teachers already work an average of 52 hours a week. Assembling an early-warning picture for every student, every week, by hand isn't realistic.
Five signals, drawn from everyday classwork
AcadiumLab compiles a weekly snapshot per student from the work teachers already do, and runs five independent detectors over it: reading-comprehension strain, a rote-versus-mastery transfer gap, early-dropout attendance patterns, missing-work spirals, and silent strugglers — learners who still look fine on the surface but are quietly sliding.
Each one looks for a different pattern that tends to precede a fall, so a student who looks fine on one measure can still be flagged on another. The point isn't a single risk score; it's making the early evidence legible, in plain language, while there is still time to respond.
Reactive, not a prediction
This is deliberately not a dropout-prediction system. The signals are reactive: they surface evidence that already exists in a student's classwork, rather than forecasting the future. The early-dropout signal, for example, is based on an observed attendance pattern — not a probability the platform invents.
That framing matters for a school. A reactive signal points to something a teacher can verify and act on this week; a prediction asks them to trust a black box about something that hasn't happened yet.
Detection is only half of it
Surfacing a struggling student only helps if it leads to action. When a teacher or school launches an intervention, AcadiumLab tracks it from a baseline score to an outcome score, so the question stops being "did we do something?" and becomes "did it work?" — and the answer feeds back into the next week's picture.
How does AcadiumLab identify students who are falling behind?
It runs five independent detectors over each student's weekly snapshot — reading comprehension, the rote-versus-mastery transfer gap, early-dropout attendance patterns, missing-work spirals, and silent strugglers who still look fine on paper. The detectors surface early evidence that already exists in everyday classwork; the dropout signal is pattern-based, not a future prediction.
Is this a dropout-prediction system?
No. The signals are reactive — they surface evidence that already exists in students' classwork, in time to act on it, rather than forecasting the future.