<p>Micro-hydropower systems have emerged as crucial renewable energy solutions in response to growing environmental concerns about fossil fuels. While screw turbines have been extensively studied through numerical and experimental methods, the application of artificial intelligence for their performance prediction remains underdeveloped. This study presents a comprehensive artificial neural network framework for efficiency forecasting of screw turbines, trained on integrated empirical, computational, and field-collected datasets. The developed artificial neural networks architecture, featuring 20 hidden layers, demonstrated exceptional predictive accuracy across training (70%), validation (20%), and testing (10%) datasets, achieving mean squared errors of 7.2 × 10⁻⁴, 4.7 × 10⁻⁴, and 2.8 × 10⁻³, respectively. The model exhibited outstanding generalization capability with a coefficient of determination of 0.96 across all test cases, indicating it captures 96% of the variability in turbine efficiency. Prediction errors remained within a tight range, with root mean square error of 0.02 efficiency points and relative errors between 0.01 and 5.08%, demonstrating consistent accuracy even at operational boundaries. These results confirm the artificial neural network ability to identify complex, nonlinear relationships between key parameters (inclination angle, diameter-to-head ratio, and specific speed) and turbine performance.</p>