Quantum Machine Learning for minimal omics datasets with large feature space using embeddings and feature selection techniques

Despite the amount of omics data generated in the last decade, the low data regime remains a significant challenge in healthcare, particularly in clinical trials and the study of rare diseases. In this study, we explore the application of quantum machine learning techniques to address the challenges posed by low data availability in healthcare. Leveraging the power of quantum computing, we focus on efficient selection of relevant feature and embedding, reducing the model complexity (and number of qubit needed) to enhance the performance even with limited data. In our case study, we used RNA data from epilepsy patients demonstrating how the predictive performance of QML models varies significantly depending on the feature selection approach employed, with noticeable improvements over classical counter part when domain knowledge is integrated to guide the selection process.