In the context of the increasing importance of foreign language learning, accurate diagnosis and effective intervention of dyslexia are crucial. There are still some problems such as traditional diagnostic methods are time-consuming, inaccurate and lack of personalized intervention strategies. This paper first collects a large amount of data related to learners’ reading behavior through various channels. Then, artificial intelligence algorithms such as neural networks and decision trees are used to analyze these data to achieve accurate diagnosis. In the intervention phase, data analysis revealed that some learners had problems with slow reading speed or low comprehension. The study empirically evaluates the intervention's efficacy in foreign language reading contexts. Comparative analysis reveals an 18% improvement in reading comprehension accuracy for the experimental group (versus 10% for controls), demonstrating that AI-driven personalized interventions significantly enhance learners’ ability to overcome linguistic barriers (p < 0.05).