Researchers faced challenges in accessing large and unbiased data sets for training models during the early stages of the pandemic. The use of small, unknown origin data sets with lower quality images created biases in the models. Some papers aimed to aid clinicians, while others explored deep learning models as an academic question.
