A critical factor influencing the design and safe operation of offshore wind turbines (OWTs) is the dynamic properties of seabed soils under long-term cyclic loading caused by wind, waves, and currents. This study proposes the use of advanced machine learning (ML) techniques to enhance the geotechnical assessment of marine soils for OWT foundations. A key constraint in offshore wind foundation design is accurately assessing the stiffness degradation and cumulative strain of marine soils under long-term cyclic loading. Standard practice reliance on testing methods, like cyclic triaxial testing and cyclic simple shear, is time-consuming and costly process, hampered by sparse, localised sampling that cannot fully capture the inherent variability of seabed conditions. These limitations increase geotechnical uncertainty in offshore environments, where soil degradation under cyclic loads governs foundation performance. This study presents a literature-driven data framework for predicting the shear modulus and cumulative strain of marine sediments, using key soil index properties and cyclic loading parameters. A comprehensive database was compiled from published cyclic triaxial tests on marine fine-grained soils, including liquid and plastic limits, plasticity index, effective confining stress, cyclic stress ratio, number of cycles, loading frequency, and shear strain level. Building on recent studies which have demonstrated the potential of ML models, in this study we propose and validate an advanced ML-based predictive model, developed to assess dynamic properties in fine grained marine soils. The study investigates the performance, of ensemble methods such as AdaBoost, CatBoost, XGBoost, and LightGBM, as well as Random Forest, Decision Tree, Support Vector Regression (SVR) optimised with Particle Swarm Optimisation, and interpreted using SHapley Additive Explanations. The resulting models show prediction accuracy above 85% and indicate that the dynamic triaxial test parameters have a greater influence on the results. These models provide reliable and scientifically grounded references for the design of OWT in broader regions, thereby contributing to the safety and economic development of marine engineering.