A User-Centric Approach to Reliable Automated Flow Cytometry Data Analysis for Biomedical Applications
Georg Popp1, Lisa Jöckel2, Michael Kläs2
1Department of Cell and Gene Therapy Development, Fraunhofer Institute for Cell Therapy and Immunology IZI, Leipzig, Germany.
Summary
This study introduces a user-centered approach for automated flow cytometry (FCM) analysis, combining machine learning and quality assurance to improve immune cell phenotyping. The goal is to make automated gating more reliable and accessible for routine use.
Area of Science:
- Immunology
- Computational Biology
- Data Science
Background:
- Flow cytometry (FCM) generates complex immune cell phenotyping data.
- Automated gating methods face adoption challenges due to user accessibility and model reliability issues.
Purpose of the Study:
- To develop a user-centered solution for automated FCM analysis.
- To enhance the routine application of automated immune cell phenotyping.
- To improve the reliability and efficiency of FCM data analysis.
Main Methods:
- Utilized supervised machine learning (SML) for event classification.
- Employed rapid application development (RAD) for software prototype generation.
- Integrated structured argumentation from assurance cases (ACs) for quality assurance.
- Incorporated uncertainty estimation to guide model operation and inform users.
Main Results:
- Developed a data-driven model for automated FCM gating.
- Created software prototypes enabling user-driven application of the model.
- Established quality analyses based on structured argumentation.
- Implemented uncertainty estimation for model confidence assessment.
Conclusions:
- The proposed user-centered approach addresses barriers to routine automated FCM gating.
- This method enhances the dependability of data-driven models in medical diagnostics.
- Further research is encouraged for SML, ACs, and uncertainty estimation in clinical FCM and pharmaceutical research.
