The basis for individual students' instructional support by teachers is an individual diagnosis of one's learning advances and difficulties. Even though sophisticated diagnostic tools exist, it remains an open question how diagnosis and learning can be merged into a consistent pedagogical method to support both teachers and students with feedback about the learning process. Aspects like the usage of open format questions, social interdependencies and heterogeneity, group-forming processes and the teachers supervision issue are taken into account. We propose a model for integration of peer assessment functionality for learning into a computer-based Adaptive Diagnostic Learning Environment to solve central problems of classroom diagnostic assessment, adaptive learning and knowledge transfer between peers in a classroom environment. The research approach focused on math learning scenarios for evaluation, but is expected to be applicable for other educational content as well. In the following paragraphs we identify several problems in classroom scenarios to be addressed, describe the model of the underlying approach, show the implementation details and explain the evaluation setup. (Contains 5 figures.)