Eugene Fan: Dual-Model Explainability in Blood Cell Classification
Eugene Fan, Senior Consultant at Tan Tock Seng Hospital, shared a post on LinkedIn about a recent article he and his colleague co-authored, published in PLOS Digital Health, adding:
“Delighted to share our newly published paper in PLOS Digital Health (Impact Factor 7.7): ‘Aligning deep learning and interpretable models for blood cell classification: A dual-model framework for explainability.’
Deep-learning systems can classify blood cells with remarkable accuracy, but their ‘black-box’ nature remains a barrier to clinical trust and adoption. In this study, we developed the dream framework, pairing a high-performing YOLO classifier with an interpretable Random Forest model that explains predictions using familiar morphological features such as cell size, nuclear shape, nuclear–cytoplasmic ratio, lobulation and colour.
The YOLO model correctly classified 99.84% of cells, while the interpretable model reproduced 98.91% of its predictions. Importantly, the explanations also showed 96.9% concurrence with the reasoning of haematologists and laboratory medical technologists.
For me, the key message is that accuracy in AI algorithms alone is not enough.
Clinical AI should also be able to show us why it has reached a decision, using concepts that us clinicians already recognize and understand.
Congratulations and sincere thanks to Magdalene C., Xinran XU, Li Rong Wang, Timothy Yeo, Hemalatha Shanmugam, Shu Ping Lim, Xiuyi Fan and everyone who contributed to this work.”
Title: Aligning deep learning and interpretable models for blood cell classification: A dual-model framework for explainability
Authors: Magdalene Yeok Yu Cheong, Xinran Xu, Li Rong Wang, Timothy Xiao Jing Yeo, Hemalatha Shanmugam, Shu Ping Lim, Bingwen Eugene Fan, Xiuyi Fan
Other posts featuring Eugene Fan on Hemostasis Today.
-
Sep 23, 2026, 11:08John Cush: Do JAK Inhibitors Increase VTE Risk After Joint Replacement?
-
Sep 23, 2026, 11:05Fonkou Steve: Blood Choice Requires Access to Safe Transfusion Options
-
Sep 23, 2026, 11:00Linda Rockwood: Cultural and Ethical Change in Transfusion Medicine
-
Sep 23, 2026, 10:30Michael Angers: Advancing Transfusion Safety Through Smarter Laboratory Workflows
-
Sep 23, 2026, 07:28Isabelle Grenier-Pleau: Targeting Von Willebrand Disease to Transform Women’s Health
-
Sep 23, 2026, 07:07Wolfgang Miesbach: Mechanisms of Liver Injury After Hemophilia A Gene Therapy
-
Sep 23, 2026, 06:24Nicolas Gendron: Everolimus and Venous Thromboembolic Risk in Kidney Transplantation
-
Sep 23, 2026, 06:07Hoori Shahwar: Cerebral Venous Sinus Thrombosis Diagnosis and Management
-
Sep 23, 2026, 06:06Advancing Hemophilia Care and Patient Advocacy in Bangladesh – World Federation of Hemophilia