Malaria, a significant global health threat, is traditionally diagnosed through manual examination of blood smears for parasite-infected cells, a method limited by its reliance on the examiner’s expertise. To overcome these limitations, this study employs deep learning for malaria prediction, leveraging models like XCEPTION, AlexNet, DenseNet, and MobileNet, enhanced with transfer learning, to classify blood smears as parasitized or uninfected using Anaconda software. The models were trained on a dataset comprising 80% of the online medical records, with 10% used for validation and the remaining 10% for testing on 224 × 224 pixel images. DenseNet emerged as the top performer with a 97% accuracy rate, followed by MobileNet, AlexNet, and XCEPTION. The study further incorporates the LIME (Local Interpretable Model-agnostic Explanations) technique for interpretability, providing understandable explanations for predictions to bolster trust and transparency in the deep learning model’s diagnostic capabilities.