In the digital era, large amounts of data are collected without efficient means to analyze the information, especially in medicine. To solve this problem, a group of scientists, led by Professor Smita Krishnaswamy of Yale University, created SAUCIE (Sparse Autoencoder for Clustering, Imputation, and Embedding), an artificial neural network that evaluates single cell datasets to organize useful information and identify immune response patterns to different pathogens and diseases. In the initial research phase, the researchers used SAUCIE to analyze patients with dengue fever along with a healthy control group from the same region of India; now, SAUCIE is also being used to analyze diseases ranging from diabetes to the flu.
Neural network machine learning provides an innovative method to create narrow classifications and focus on important data at the cellular level. In addition, SAUCIE is advantageous because it can function unsupervised—it automatically evaluates data without requiring scientists’ supervision. The artificial neural network detects the difference between random and meaningful information, presenting only relevant data to scientists. “This is only the beginning of single cell analysis,” said first-author Matthew Amodio. For a long time, scientists were only able to analyze patients’ blood by taking a physical sample and measuring the average outcome of all the datasets within cells. Now, single cells can be analyzed efficiently, and SAUCIE can be used on whatever single-cell datasets researchers have to improve their analysis. In the future, SAUCIE will be extremely useful in conducting research and shortening the time needed for researchers to evaluate gathered data.