Drug combinations have demonstrated considerable potential in enhancing cancer chemotherapy efficiency, reducing adverse side effects, and overcoming multidrug resistance. Over the past few years, AI-based computational models have been developed to screen the huge search space of all potential drug combinations and accurately detect those with synergistic anti-cancer effects. These methods can efficiently bypass the laborious and time-consuming experimental steps involved in-vitro drug combination testing. Here we used the NCI-ALMANAC dataset to develop cell line-specific machine learning models capable of predicting synergistic growth inhibition of 108 cancer medications, tested at multiple dosages. The average performance of these models, as indicated by an AUC of 0.75 underscores the promise of leveraging neural networks for predicting drug combination outcomes.