Detection of acute myeloid leukemia and remission states using heterogeneous flow cytometry data

Renjun Bao¹Ming Feng²Mian Wang¹Yunkai Liu³Liang Hu^2*Yonghua Yao^1*

• ¹Department of Hematology, Shidong Hospital of Shanghai Yangpu District, Shanghai, China
• ²School of Computer Science and Technology, Tongji University, Shanghai, China
• ³Laboratory Diagnosis Department, Shanghai Kingmed Center for Clinical Laboratory Co., Ltd., Shanghai, China

Acute myeloid leukemia (AML) is a hematological malignancy that requires accurate diagnosis and monitoring to guide effective treatment. Flow cytometry is widely used for AML diagnosis because it detects minimal residual disease. However, most existing methods rely on consistent marker panels across samples, overlooking the marker heterogeneity in real-world clinical settings, where different markers or staining methods are used across patients. Additionally, current approaches often fail to account for the significance of remission states, which are critical for disease management and prognosis. To address these challenges, we propose a machine learning-based classification model that integrates heterogeneous flow cytometry data to distinguish between AML, complete remission (AML-CR), and normal samples. By leveraging a diverse set of 53 markers, we constructed and evaluated predictive models using six different machine learning classifiers. Our results showed that the Random Forest model achieved the best performance, with an accuracy of 94.92%, F1 score of 94.13%, precision of 94.58%, recall of 93.74%, and an AUC of 94.83%. This study demonstrates the potential of machine learning to overcome the limitations of traditional methods and provide a more robust and accurate classification of AML and its remission states using heterogeneous flow cytometry data. The code for this study is available at https://zenodo.org/records/15110287.