Deep learning provides a powerful tool for studying the interactions between biomaterials by analysing their drying droplet patterns. Traditional available techniques, such as circular dichroism (CD) spectroscopy, Fluorescence spectroscopy, and mass spectrometry are costly and time consuming causing a limitation in their capacity to do high-throughput interactions screening. Previous research has successfully utilized deep learning models to classify biomaterial interactions with high accuracy including histone-DNA complexes, protein-immunoglobulin and amyloid beta mutants interactions. This research aims to utilize the convolutional neural networks (CNNs) to predict the drug-carrier binding affinity. In this study, Human Serum Albumin (HSA) was used as a carrier along with various drugs. Drug-carrier mixtures were pipetted onto hydrophobic glass surfaces, followed by imaging their drying droplet patterns using polarized light microscopy (PLM). These images were used to train the CNN models to classify the binding affinity as validated by complementary techniques such as CD. The trained model was tested on new unseen drugs to predict their binding affinity to HSA, followed by verification of the prediction with experimental data. This method shows great potential to serve as high-throughput screening method for optimal drug-carrier combinations.