This study addresses the need for a quick and automatic COVID-19 diagnosis and a faster alternative to PCR testing. Researchers Chen et al. proposed a new deep learning algorithm for automated COVID-19 diagnoses requiring only a small sample set for training. The authors used contrastive learning to train an encoder, ResNet-50, which can capture a large variety of information from publicly available lung data sets and prototypically issue diagnoses. Performance of the model was evaluated for accuracy, precision, number of true positives, and the relation between false and true positive results. When sample sizes were larger than 3, the model achieved higher performance over the currently competing alternative model for medical image analysis. Lack of annotated CT scans available as a dataset is a significant challenge to deep-learning based diagnostic methods; however, the team plans to continue to validate the model’s generalizability with further training on COVID-19 datasets. The model is also expected to be useful for further medical imaging analysis in the event of a data shortage such as a novel pandemic.