Machine Learning Methods in Clinical Flow Cytometry

Abstract

This review will explore the integration of machine learning (ML) techniques to enhance the analysis of increasingly complex and voluminous flow cytometry data, as traditional manual methods are insufficient for handling this data. We attempt to provide a comprehensive introduction to ML in flow cytometry, detailing the transition from manual gating to computational methods and emphasizing the importance of data quality. Key ML techniques are discussed, including supervised learning methods like logistic regression, support vector machines, and neural networks, which rely on labeled data to classify disease states. Unsupervised methods, such as k-means clustering, FlowSOM, UMAP, and t-SNE, are highlighted for their ability to identify novel cell populations without predefined labels. We also delve into newer semi-supervised and weakly supervised methods, which leverage partial labeling to improve model performance. Practical aspects of implementing ML in clinical settings are addressed, including regulatory considerations, data preprocessing, model training, validation, and the importance of generalizability, and we underscore the collaborative effort required among pathologists, data scientists, and laboratory professionals to ensure robust model development and deployment. Finally, we show the transformative potential of ML in flow cytometry in uncovering new biological insights through advanced computational techniques.

Keywords: acute leukemia; clinical flow cytometry; machine learning; operational efficiency.

Figure 1

A diagrammatic overview of the model development workflow for supervised machine learning, from data annotation through training and validation to production-ready inference.

Figure 2

Unsupervised learning algorithms. Labelled cells with unknown identities are analyzed by flow cytometry and partitioned in the feature space based on marker expression. Clustering methods are used to group data points based on their marker profiles, while dimensionality reduction can be applied to high-parameter datasets to simplify their representation and enhance the biological interpretability. These tools output cell annotations to identify distinct populations, enabling the characterization of disease states and facilitating novel discoveries.

Figure 3

Twin networks: one neural network model is learned on paired examples. Pairs are weakly supervised, only requiring a label for if they are the “same” or not. N.b.: this supervised signal could be developed via an unsupervised method.

Figure 4

AI model lifecycle from initial data analysis to model development and operationalization.