This article focuses on SESSAD-type structures (Service d'Éducation Spéciale et de Soins A Domicile) that provide individualized support to children and adolescents with disabilities in their normal living and educational environment. Our goal is to propose an efficient model to schedule the activities of Speech and Language Therapists that can be used in a real case. We thus propose a formulation of the problem as a vehicle, solved thanks to a clustering method, improved using neighborhood operators and optimization processes. Some tests on real data enable us to compare different methods and determine the most efficient one.