International testing and refinement of AI algorithms predicting acute leukemia subtypes from routine laboratory data

A Amin T. Turki Y Yi Fan (Department of Radiation Oncology, University of Pennsylvania) A Alberto Hernández-Sánchez W Wellington Silva S Shaun Fleming K Koray Yalcin C Catharina H.M.J. Van Elssen Y Yazan Madanat M Magdalena Karasek M Mahmoud Aljurf M Matteo G. Della Porta A Alexandra Martinez-Roca L Luca Guarnera K Katarina Steffen E Evangelia Antoniou M Maria M. Rivas D Deepak K. Mishra A Ansgar T. Blum S Stephania Niry Manantsoa A Adeniyi Adiat A Amir Enshaei F Felicitas Thol M Maria Teresa Voso J Jia Chen T Tusneem Ahmed Elhassan A Anthony V. Moorman M María Belén Vidriales N Nina R. Neuendorff A Ahmet Koc P Pratyush Mishra D Dirk Strumberg R Roma S. Fourmanov L Lukas Heine J Jens Kleesiek D Daniel Munárriz G Gianluca Asti M Mridula Mokoonlall M Marisa Kometas E Eduardo Rego R Rabea Mecklenbrauck M Marta Sobas D Depei Wu F Felix Nensa M Merlin Engelke

Abstract

Abstract Despite advances for patients with acute leukemia health disparities limit access to diagnosis and treatment. Artificial Intelligence (AI) approaches may address some disparities. We retrospectively assemble a diverse, international cohort of 6206 leukemia patients from 20 centers to test an AI tool designed to support leukemia diagnosis using standard laboratory results. Executing the pretrained algorithm results in varying accuracy metrics. With confidence cutoff predictions, 2000-fold bootstrapped area under the curve (AUROC) metrics are 0.94 for acute myeloid leukemia (AML), 0.98 for the promyelocytic subtype and 0.84 for acute lymphoblastic leukemia. However, this cutoff excludes 70.8–92.5% of patients from predictions. We improve accuracy and robustness, while maintaining generalizability via an ensemble of Isolation Forest and Local Outlier Factor increasing AUROC for AML from 0.72 to 0.84 (hold-out test set, patients below confidence threshold), while excluding only 12.1% of patients. Furthermore, we retrain the algorithm for pediatric patients.

Article Details

Volume / Issue Vol. 17, Issue 1
Published March 20, 2026
ISSN 2041-1723
Publisher Nature Portfolio

Journal Info

Nature Communications

Nature Portfolio

ISSN: 2041-1723 Open Access Life Sciences

Authors (44)

A

Amin T. Turki

Y

Yi Fan

Department of Radiation Oncology, University of Pennsylvania

A

Alberto Hernández-Sánchez

W

Wellington Silva

S

Shaun Fleming

K

Koray Yalcin

C

Catharina H.M.J. Van Elssen

Y

Yazan Madanat

M

Magdalena Karasek

M

Mahmoud Aljurf

M

Matteo G. Della Porta

A

Alexandra Martinez-Roca

L

Luca Guarnera

K

Katarina Steffen

E

Evangelia Antoniou

M

Maria M. Rivas

D

Deepak K. Mishra

A

Ansgar T. Blum

S

Stephania Niry Manantsoa

A

Adeniyi Adiat

A

Amir Enshaei

F

Felicitas Thol

M

Maria Teresa Voso

J

Jia Chen

T

Tusneem Ahmed Elhassan

A

Anthony V. Moorman

M

María Belén Vidriales

N

Nina R. Neuendorff

A

Ahmet Koc

P

Pratyush Mishra

D

Dirk Strumberg

R

Roma S. Fourmanov

L

Lukas Heine

J

Jens Kleesiek

D

Daniel Munárriz

G

Gianluca Asti

M

Mridula Mokoonlall

M

Marisa Kometas

E

Eduardo Rego

R

Rabea Mecklenbrauck

M

Marta Sobas

D

Depei Wu

F

Felix Nensa

M

Merlin Engelke