This study proposes a comprehensive approach to modeling and strategic management of migration flows of scientific personnel in the context of national innovation systems (NIS) of developing countries. Using the methodology of system dynamics, network analysis and a hybrid neural network based on a genetic algorithm (GA-BP), the work identifies key drivers of intellectual migration, assesses risks to knowledge development and offers tools for aligning intellectual resources with the institutional environment. The results demonstrate stable negative trends in the outflow of scientists, caused by a complex of economic, socio-political and psychological factors, and substantiate the need to adapt scientific and technical policies to mitigate NIS dysfunctions.