Predicting literacy intervention responsiveness using semi-supervised machine learning.
Verbal comprehension, visual memory, and verbal working-memory scores forecast which children with SEN will benefit most from long-term systematic phonics.
01Research in Context
What this study did
The team fed test scores into two computer models. The goal was to spot which kids would gain the most from a long phonics program.
All children had special learning needs. No labels were excluded. The models looked at memory, language, and visual skills.
What they found
Verbal comprehension, visual memory, and verbal working memory rose to the top. These three scores told the model who would grow in reading.
The tool was right seven times out of ten. That is strong for a screening test.
How this fits with other research
Nicolosi et al. (2024) show the upside: even a teen with autism and profound ID learned to follow written directions when phonics was drilled daily. The new model says start the same program, but only after you check verbal and memory scores.
Carter et al. (2013) found that semantics did not matter for high-functioning autism readers. The new study says semantics (verbal comprehension) does matter when the group is mixed. The clash is only on the surface: the 2013 paper looked at HFASD alone; the 2025 pool is wider.
Freed et al. (2015) told us to test spoken story skills in kids with pragmatic language impairment. The 2025 model keeps verbal comprehension on the list, backing that advice.
Why it matters
You now have a quick way to rank students before you start a months-long phonics plan. Give a short verbal comprehension, visual memory, and working-memory probe. Kids who score low on all three may need a different path or more support. Use the screen, save weeks of trial and error, and move the right kids into systematic phonics today.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Run a five-minute verbal comprehension and memory probe; place top scorers into your systematic phonics block and re-tool the lesson for the rest.
02At a glance
03Original abstract
BACKGROUND: There is pervasive non-responsiveness to systematic phonics interventions which have furthermore tended to focus on near-transfer outcomes related to phonology. There is a need to predict intervention responsiveness related to far transfer outcomes such as literacy-relevant word reading and spelling. Furthermore, there is potential for the use of advanced machine learning to maximize predictive power. AIMS: This study aims to longitudinally predict systematic phonics intervention using machine learning models. METHOD: The sample included children with special educational needs (M = 98.08 months, N = 838) who either received long-term intervention (average duration of 33.62 months) (labeled data) or only had baseline data without intervention (unlabeled data). We applied 12 semi-supervised learning models learned from the mix of labeled and unlabeled data to predict intervention responsiveness outcomes of word reading and spelling. Predictors were background information, domain-general cognitive abilities, and language-related achievement scores, with expanded predictors consisting of differences among these predictors. RESULTS: Amongst 12 models developed, Random Forest and Gaussian Naïve Bayes models achieved the highest F1 score of 0.7 in the test set, supported by the incorporation of unlabeled data and expanded predictors. The top predictors were related to verbal comprehension, visual memory, and verbal working memory. CONCLUSIONS: We identified important predictors of intervention responsiveness and showed the promise of machine learning models with implications on the allocation of resources, mitigation of risk of failure, and tailoring of interventions.
Research in developmental disabilities, 2025 · doi:10.1016/j.ridd.2025.105090