Educational Leadership Enhanced by Machine Learning: Proposal for a Conceptual Model for Executive Decision-Making in Public Educational Institutions in Peru

School leadership is, after teachers' classroom practice, the second within-school factor with the greatest influence on student learning outcomes. In Arequipa's public educational institutions, however, principals' decision-making still relies mainly on experience and fragmented information, since the systematic use of evidence for school management remains an exceptional practice within the Peruvian educational system. At the same time, machine learning has become a consolidated tool capable of anticipating dropout risk, characterising school performance and sustaining dashboards for educational management. Nevertheless, the literature reviewed shows a disconnection between the technical advances of ML applied to education and the pedagogical leadership frameworks currently in force in developing countries' educational systems, which are marked by digital divides, low data literacy and technology-adoption barriers. This theoretical-conceptual article proposes an integrative model termed Machine-Learning-Augmented Educational Leadership (LEA-ML, for its Spanish acronym) articulating six dimensions: baseline pedagogical leadership, data infrastructure, the ML analytical layer, augmented decision interpretation, ethical governance and organisational adoption. The model was built through an integrative documentary analysis of 54 sources indexed in Scopus, Web of Science, Redalyc, SciELO, Dialnet and ERIC, published mostly between 2015 and 2026. Its relevance for the Arequipa context is discussed considering the existing infrastructure of the Educational Institution Management Support Information System (SIAGIE) and the region's persistent connectivity gaps. It is concluded that the model offers a pertinent theoretical basis for future empirical validation aimed at strengthening evidence-based school management among public-school principals.