Application deadline
April 27, 2026
Organizers
Dr Helen Latsoudis (WG-3 Leader)
Dr Giacomo Cavalca (WG-3 Co-Leader)
Apply now!
Applicants are required to submit their application through the following link:https://neutro-narps.eu/ml-hematology-application/
no later than April 27, 2026.
The application should include:
- A brief C.V.
- A motivation letter outlining the relevance to their research and the value of participating in the Training Course
Applicants will be formally notified of the outcome of their application by May 4, 2026 at the latest and will receive a formal COST invitation via eCOST.
Participants will receive a certificate of attendance for the sessions they have successfully completed.
For further information please contact:
coordinator@neutro-narps.eu
Purpose of the Training School
The Training School on “Machine Learning Approaches & Analytical Methodologies for Hematological Disorder Diagnostics and Prognostics” is organized by the COST Action CA24124 “Network for the Advancement of Neutropenia Research and Patient Support” (Neutro-NARPS) as a virtual event.
Data analytics methods in rare diseases are crucial for overcoming challenges related to small patient populations, data fragmentation, and the “diagnostic odyssey”. These methods focus on leveraging Real World Data, core statistics and AI to improve diagnosis, understand disease progression, and accelerate drug development.
The participating scientists will have the opportunity to expand their knowledge via interactive lectures from experts on aspects related to data analytics. They will also observe and familiarize themselves with methods to study rare blood diseases, discuss, and get hands-on experience on performing specific analyses that will promote better study design, more efficient use of data, and improved interpretation of results, ultimately contributing to more informed clinical decision-making and better patient outcomes.
The Neutro-NARPS Training School will be held in 2 distinct parts that will take place on separate dates.
PART A: 27-28 May, 2026 on “Machine Learning Approaches for Hematological Disorder Diagnostics and Prognostics” will introduce participants to machine learning (ML) methods and their applications in hematological diseases, with a particular focus on improving diagnosis, prognosis, and treatment stratification. Given the increasing availability of clinical, genomic, and imaging data in hematology, ML techniques offer powerful tools to uncover complex patterns and support precision medicine. The course combines theoretical foundations with practical, hands-on experience using real-world datasets.
By the end of the training, participants will be able to:
- Understand core concepts of ML and their relevance to hematology
- Apply supervised and unsupervised learning methods to clinical and omics data
- Develop predictive models for diagnosis, prognosis, and treatment response
- Evaluate model performance and avoid common pitfalls such as overfitting
- Interpret machine learning outputs in a clinically meaningful way
PART B: 03-04 June, 2026 on “Analytical Methodologies for Hematological Disorder Diagnostics and Prognostics” will combine background information and hands on experience on core statistical foundations, especially tailored to rare and complex blood disorders, like Neutropenia.
By the end of the training, participant will be able to:
- Perform descriptive and inferential statistical analyses
- Select appropriate statistical tests based on study aim, type (continuous or categorical) & distribution (parametric or non-parametric) of data, and observations (independent or paired)
- Conduct survival analyses using methods such as Kaplan–Meier estimator and regression models like Cox proportional hazards model
- Handle small datasets
- Integrate of prior knowledge with limited new data to produce more robust and clinically meaningful inferences.
- Lead to better clinical decision-making and patient outcomes
Target audience: researchers, clinicians, biologists, data scientists, healthcare and biomedical professionals, students, non-experts in biostatistics
Training School Agendas
Agenda Part A | 27-28 May, 2026
on “Machine Learning Approaches for Hematological Disorder Diagnostics and Prognostics”
on “Machine Learning Approaches for Hematological Disorder Diagnostics and Prognostics”
DAY 1 – Afternoon Session (3Hrs)
14.00 – 15.30 | Foundations of Machine Learning in Hematology (Dr Gabriel Vignolle)
(Download the presentation)
The talk will provide substantial information on important aspects of
- Data to Model: A Machine Learning workflow
- Data preprocessing, Feature engineering & Model selection
- Tips and Tricks: Handling small sample sizes, missingness and sparse data
- Interpretation of predictive models for clinicians, translating ML into daily patient care
- Ethical considerations and bias in ML models
15.30 – 17.00 | Machine Learning for Complex Biomedical Data” (Dr Andrea Cappozzo)
(Download the presentation)
The talk will cover aspects related to
- Introduction to Machine Learning and Data Science: key concepts and taxonomy of approaches
- Supervised vs. unsupervised learning: distinctions, use cases
- Handling complexity: multicentric data with linear mixed models
- Handling complexity: high-dimensional data via penalized estimation
- Case study: DNA Methylation surrogate biomarker creation with penalized mixed-effects multitask learning
DAY 2 – Afternoon session (3Hrs)
14.00 – 17.00 | Hands-on training ( Dr George Manikis)
Trainees will practice on datasets, on their own electronic device, in real-time following their trainers’ instructions.
TRAINERS:
Dr. Gabriel Alexander Vignolle, Mag.pharm. Dr.rer.nat.Postdoctoral Scholar
– The Vatche and Tamar Manoukian Division of Digestive Diseases, UCLA, Los Angeles, CA, USA
– Department of Medicine, David Geffen School of Medicine, UCLA, Los Angeles, CA, USA
– Goodman-Luskin Microbiome Center, UCLA, Los Angeles, CA, USA
Dr. Andrea Cappozzo,
– Associate Professor of Statistics, Department of Statistical Sciences, Università Cattolica del Sacro Cuore, Milan, Italy
Dr Georgios Manikis, PhD, Electronic Engineer
– Associate Professor, Department of Electrical & Computer Engineering, Hellenic Mediterranean University, Heraklion, Crete, Greece
– Collaborating researcher, Computational Biomedicine Laboratory(CBML), Institute of Computer Sciences(ICS), Foundation for Research and Technology (FORTH), Heraklion, Crete, Greece
Agenda Part B | 03-04 June, 2026
on “Analytical Methodologies for Hematological Disorder Diagnostics and Prognostics”
on “Analytical Methodologies for Hematological Disorder Diagnostics and Prognostics”
DAY 1 – Afternoon Session (3Hrs)
14.00 – 15.30 | Survival analysis of time-to-event data: basic concepts and methods” (Prof. Evangelos Kritsotakis)
This session will cover essential concepts and methods for analyzing time-to-event data or survival times. It will explain the special features of survival data, censoring, and the basic mathematical functions central to describing survival times (the survival, hazard rate, and cumulative hazard functions) and their interrelationships. Basic methods such as the non-parametric Kaplan–Meier method to estimate survival probabilities, the log-rank and Wilcoxon-Peto-Peto tests for comparing survival time distributions between two or more groups will be briefly presented and illustrated. We will also discuss the formulation and application of the Cox proportional hazards model for regression analysis.
At the end of the session, participants will be able to:
- Describe the type of data for which survival analysis is applied.
- Explain what censoring of survival times is.
- Distinguish between incidence probability or risk, incidence rate, and hazard rate.
- Contrast between the hazard rate, survival function, cumulative event function and cumulative hazard function.
- Explain how the non-parametric Kaplan-Meier method estimates survival probabilities, survival curves and quartiles of survival time.
- Discuss the use of the log-rank test and the Wilcoxon-Peto-Peto test to compare survival experiences of two or more groups.
- Explain the concept of the hazard ratio.
- Outline the general idea of the Cox proportional hazards regression model.
15.30 – 17.00 | Application of Bayesian Methods in Hematology (Prof. Esin Avci)
(Download the presentation)
The main aim of this talk is to introduce clinicians to the practical applications of Bayesian methods in medicine. Rather than emphasizing mathematical details, the presentation will highlight how these approaches can better understand patient outcomes, incorporate prior clinical knowledge, and make more informed decisions under uncertainty. Through examples, the talk will demonstrate how Bayesian methods can improve the analysis of time-to-event data and model complex relationships between risk factors. By the end of the session, participants will gain an intuitive understanding of how prior information and different types of clinical data can be integrated to support more personalized and evidence-based medical decision-making.
The talk on the “Application of Bayesian Methods in Hematology” will cover aspects related to
Α. Core statistical concepts for a proper study design of a rare hematological disease, such as
- handling small sample sizes and sparse data
- descriptive statistics
- inference statistics
- probabilities
- handling small sample sizes and sparse data
Β. Bayesian methods
- core principles of Bayesian inference (prior, likelihood, posterior)
- Bayesian regression (linear, logistic) and survival analyses (time to event models)
- Simulations and interpretation of graphical outputs (posterior plots, probability curves)
DAY 2
Hands-on training
Speakers/Trainers:
Prof. Evangelos I. Kritsotakis, BSc(Hons) MSc PhD CStat FHEA MRSPH, Associate Professor of Biostatistics
– Division of Social Medicine, School of Medicine, University of Crete, Heraklion, Crete, Greece
Prof Esin Avci, Associate Professor of Statistics
– Department of Statistics, Faculty of Arts and Sciences, Giresun University, Giresun, Türkiye
