Sample size determination for biomedical big data with limited labels

The era of big data has produced vast amounts of information that can be used to build machine learning models. In many cases, however, there is a point where adding more data only marginally increases model performance. This is especially important for scenarios of limited labeled data, as annotation can be expensive and time consuming. If the required sample size for accurate model performance can be determined early, then resources can be allocated appropriately to minimize time and cost. In this study, we explore sample size determination methods for four real-world biomedical datasets, spanning genomics, proteomics, electronic health records, and insurance claims data, all with millions of instances each and<2% class ratio. The methods used involve approximating a learning curve for a large amount of data using a small amount of data. We evaluate an existing method that fits an inverse power law model to a small learning curve and introduce a novel semi-supervised method that utilizes the large amount of unlabeled data for estimating a learning curve. We find that the inverse power law method is applicable to big data, while the semi-supervised method can be better at detecting convergence. To the best of our knowledge, this is the first study to apply an inverse power law curve fitting method to big data with limited labels and compare it to a semi-supervised approach.