Tandem Moritz Middeke & Jan-Niklas Eckardt
Artificial Intelligence in Hematology
Hematologic malignancies can present as diagnostic and therapeutic emergencies requiring rapid recognition and management, such as acute myeloid leukemia (AML). To establish a diagnosis and guide treatment, clinicians must integrate information from cytomorphology, cytogenetics, immunophenotyping, and molecular testing. With the growing volume of data available, clinical decision-making is becoming increasingly complex. Our research group combines expertise in hematology, biology, data science, and computer science to develop artificial intelligence–based models that enhance diagnosis, risk assessment, and support clinical decision-making in hematology. We train our models using multimodal data from both research and clinical routine, in cooperation with university centers and laboratories nationally and internationally.
Our current research focuses on four complementary areas:
Computer Vision: Diagnosing hematologic malignancies currently relies on expert interpretation of bone marrow smears, a process that is both time-consuming and prone to observer-related variability. To overcome this diagnostic bottleneck, our group develops computer vision models that automate the detection of malignant entities from bone marrow images, predict genotype-phenotype links (visual digital biomarkers), and forecast treatment response.
Risk Assessment and Outcome Prediction: Using large datasets, including clinical, laboratory, and genetic information, we develop data-driven models to stratify patients by risk, identify genetic alterations linked to divergent outcomes, cluster patients by disease biology, and predict treatment response.
Large Language Models (LLMs): To support personalized clinical decision-making, we evaluate and benchmark state-of-the-art LLMs and vision–language models on hematology-specific tasks and multimodal patient cases.
Synthetic data: We use generative models to create synthetic patients that closely mimic disease biology and outcome behavior of real patients. These computationally generated cohorts can boost model training, support model benchmarking, overcome data sharing restrictions, and enable novel clinical trial designs.
Research aims
- Improve the accuracy, speed, and safety of diagnosing hematological malignancies
- Gain insights into cancer biology that drives disease initiation, progress, and relapse
- Refine patient risk assessment and outcome prediction using multimodal data
- Enable personalized treatment strategies
- Create synthetic data for next-generation clinical trials in hematology
- Seamless integration of AI algorithms to improve diagnostic and clinical workflows in hematology and deliver profound results to our patients
Members
PD Dr. med. Jan Moritz Middeke, Research Group Leader MSNZ
Dr. med. Jan-Niklas Eckardt, MSc, MHBA, Research Group Leader MSNZ
Dr. rer. nat. Susann Winter, Scientific Coordinator
Franziska Schmidt, Administrative Coordinator
Dr. med. Freya Schulze, Clinician Scientist
Dr. med. Martin Schneider, Clinician Scientist
Lara Neubert, Doctoral student (medicine)
Ishan Srivastava, Computer Vision Specialist
Martina Radoynova, AI Engineer (LLM & Computer Vision)
Dr. Kumar Reddy Kakularam, ML Data Scientist
Maria Tanzmann, Student Assistant (SHK)
Claudia Dill, Medical Laboratory Assistant (MTA)
Selected publications
Eckardt JN, Srivastava I, Wang Z, et al. Synthetic bone marrow images augment real samples in developing acute myeloid leukemia microscopy classification models. NPJ Digit Med. 2025;8(1):173. doi:10.1038/s41746-025-01563-9
Eckardt JN, Hahn W, Ries RE, et al. Age-stratified machine learning identifies divergent prognostic significance of molecular alterations in AML. Hemasphere. 2025;9(5):e70132. doi:10.1002/hem3.70132
Eckardt JN, Srivastava I, Schulze F, et al. Image-based explainable artificial intelligence accurately identifies myelodysplastic neoplasms beyond conventional signs of dysplasia. NPJ Precis Oncol. 2025;10(1):26. doi:10.1038/s41698-025-01222-y
Eckardt JN, Hahn W, Prelaj A, Bornhäuser M, Middeke JM, Kather JN. Artificial intelligence-generated synthetic data for cancer research and clinical trials. Nat Rev Cancer. Published online February 20, 2026. doi:10.1038/s41568-026-00912-4
Eckardt JN, Middeke JM. Promises and challenges of artificial intelligence in haematological diagnostics. Br J Haematol. 2025;207(3):754-756. doi:10.1111/bjh.70055
Eckardt JN, Hahn W, Röllig C, et al. Mimicking clinical trials with synthetic acute myeloid leukemia patients using generative artificial intelligence. NPJ Digit Med. 2024;7(1):76. doi:10.1038/s41746-024-01076-x
Links
https://ai-in-cancer.org/about-us/