This study uses Convolutional Neural Networks (CNNs) to construct an automated parasite identification method in blood smear pictures, addressing the urgent need for accurate and easy malaria diagnosis. We use secondary data collected from several datasets, utilizing an interpretive mindset, a deductive technique, as well as a descriptive research strategy. Technical considerations such as architecture selection, data preparation, and interpretability approaches are carefully taken into account when perfecting the CNN model. The model's effectiveness has been demonstrated by results showing excellent sensitivity (95%) and specificity (92%) in parasite identification. Techniques for visualization make regions of interest clearer, increasing confidence and transparency. Testing for generalizability shows consistent results across several datasets and demographic groupings. The CNN-based methodology performs better than traditional microscopy and quick diagnostic tests when compared to current approaches, especially in situations of low parasitemia. With implications for improvements in healthcare delivery in malaria-endemic areas, this research advances automated malaria diagnosis.