Objective: Automated hematology analyzer flags may provide early warning signals for hematological malignancies. This study aimed to evaluate the diagnostic performance of Sysmex XN-1000 analyzer flags and to develop a multivariable prediction model for identifying active acute myeloid leukemia (AML) in adult patients.

Methods: A retrospective analysis of 5215 complete blood count (CBC) tests (15 cases and 5200 controls) was conducted at Gazi University Hospital. Data obtained from the hematology analyzer were evaluated. Diagnostic accuracy metrics were calculated for specific Q-flag (“Blasts/Abn Lympho?”) and for the interpretive flags of the analyzer. A logistic regression model was employed to identify independent predictors. As a result of these analyses, the best-performing model was selected.

Results: In the multivariable model, “Thrombocytopenia” (odds ratio [OR=57; 95% CI =15-218]) and “Blasts/Abn Lympho?” ≥120 (OR=33; 95% CI =8-135) were independent predictors of active AML. The AML index model demonstrated high diagnostic performance, with a sensitivity of 66.7%, a specificity of 99.8%, a positive predictive value of 56%, a negative predictive value of 99.9%, and an area under the curve (AUC) of 0.894 (95% CI=0.731-0.963) at a cut-off of >0.034. Based on the final multivariable model, an AML index was derived.

Conclusion: The Sysmex XN-1000 flag-based regression model demonstrated high diagnostic accuracy for detection of active AML disease. This method may allow identifying patients who require prompt peripheral blood smear review and further hematological evaluation by allowing for rapid classification of high-risk samples directly from the hematology analyzer. These results, from a retrospective, single-center study with a small sample size, need to be confirmed in larger prospective study designs.