Abstract Systematically examining the correlation between the lifestyle habits of undergraduate students and their academic performance holds significant practical implications in advancing higher education. This study adopts an integrated perspective and analyzes a substantial dataset (3,123,840 data points) of 3499 undergraduates at a Chinese university. This study employs a Long short-term memory neural network to identify eating behavior indicators and develops a comprehensive model to evaluate the relationship between students’ lifestyle habits and academic performance; the lifestyle habits cover eating, hygiene, and studying habits. The findings challenge conventional wisdom by revealing that stringent eating schedules do not consistently correlate with superior academic performance. Instead, a higher degree of inertia in eating behavior (e.g., waking up early) correlates with better academic outcomes. Positive correlations also exist between students’ hygiene and studying habits and their academic performance. These results provide valuable insights into the relationship between students’ behavior and academic performance. This work carries implications for promoting the digitalization of higher education and enhancing education management for undergraduate students.