Deadline Extension 14 October 2021 - Medical & Biological Engineering & Computing Designing Machine Learning approaches for early-stage prediction of complications and risk stratification of COVID-19 patients
Luca Romeo
[Apologies if you receive multiple copies of this CFP]
Introduction and General Justification
During the COVID-19 emergency and lockdown phase, intensive care was at its limit, and doctors were forced to choose which of their ICU patients had the best chances of survival. This worldwide emergency highlighted the need for a predictive care model capable of providing an accurate estimate of resources and preventive medicine. The analytical capability of machine learning (ML) methods have proven to be extremely accurate, and are in some cases superior to classical statistical approaches for solving this task. However, the ability to predict complications by analyzing the medical records of patients in intensive and non-intensive care is hampered by numerous challenges such as difficulty in finding structured clinical data, missing values, and a lack of annotation with respect to a target variable that may represent the patient's own risk level. Under these conditions, predicting the risk of a particular patient to develop complications associated with COVID-19 is a relevant and topical challenge.
Recent years have witnessed an increasing amount of available Electronic Health Record (EHR) data as ML techniques continually evolved. To support clinicians during the COVID-19 pandemic, new ML and Deep Learning (DL) algorithms could be designed to discover complex patterns and set up powerful models which can be integrated into Clinical Decision Support Systems to provide (1.) risk profiles of individual patients from which a different intensity of care can be deduced, and (2.) predictions of risk of short-term complications, which will activate personalized prevention systems directly addressed to the patient.
Aims & Scope
This special issue aims to cover all research related to ML and DL methodologies in providing risk profiles of individual patients in intensive and non-intensive care units from which a different intensity of care can be deduced. All contributions related to the design and development of novel ML and DL methodologies involving the prediction of complications related to COVID-19 using ICU and non-ICU data, EHR data, medical imagery, and genomic data are particularly welcome. These contributions may be proposed to overcome relevant challenges in the context of clinical decision support systems. These challenges include but are not limited to: model interpretability, missing values, high-imbalanced setting, bias, and sparse annotations over time.
The main goal of this collection is to promote synergy between the ML and the biomedical communities by encouraging contributions that fully intersect with the mission of Medical & Biological Engineering & Computing.
Topics and Keywords
Potential topics include, but are not limited to:
● Automatic disease prediction using ML or DL
● Genomic and proteomic data analysis
● Biomedical Image Analysis using DL
● Biosignal Processing
● ML and DL for Personalized Medicine
Dataset Requirements
Suitably large sample sizes will be required for submission to this SI. This is essential in order to properly measure and ensure the generalizability of the machine learning model.
Our criteria to evaluate the quality of the dataset is reported below:
- the authors should clearly report in the manuscript how the ICU data, non-ICU data, Electronic Health Record data, medical imagery, and genomic data were chosen (i.e., center, descriptive statistic) and provide a rationale for why they are sufficient;
- the authors should report a descriptive statistic of the employed dataset for the three-dimensions (patient, features, time);
- the authors should demonstrate (from a statistical perspective) that the employed dataset (multi-center or single-center) should be representative of the entire population;
- the authors should deal with any bias factor (e.g. age, sex, pathology) present in the data that might lead to an overestimation of the predictive performance;
- the authors should evaluate the robustness of the dataset in terms of the presence of missing values and outliers;
- the authors are encouraged (especially for a dataset with moderate sample size) to perform a cross-validation procedure for checking the ML performance and a nested cross-validation procedure for optimizing the hyperparameters;
- we require authors to present a dataset with > 1,000 patients, as well as replication samples of these patients over time. High sample size will avoid a strongly-biased performance estimation while increasing the robustness of the found results and the impact of the clinical evidence.
Timeline
Paper submission deadline (extended): 14 October 2021
First review decision: 31 Oct 2021
Revised paper due: 15 Dec 2021
Final review decision: 15 Jan 2022
Final manuscript submission: 31 Jan 2022
Website
https://www.springer.com/journal/11517/updates/19039740
Submit Now
https://www.editorialmanager.com/mbec/default.aspx
Guest Editors
Luca Romeo <l.r...@univpm.it> Department of Information Engineering, Università Politecnica delle Marche, Ancona (Italy)
Michele Bernardini <m.bern...@pm.univpm.it> Department of Information Engineering, Università Politecnica delle Marche, Ancona (Italy)
Emanuele Frontoni <e.frontoni@.univpm.it> Department of Information Engineering, Università Politecnica delle Marche, Ancona (Italy)
Jonathan Montomoli <jonathan....@gmail.com> Department of Intensive Care, Hospital Infermi, Rimini Department of Intensive Care Medicine, Erasmus medical Center, Rotterdam, Netherlands
Paul WG Elbers <p.el...@amsterdamumc.nl> Intensivist at VU medical center, Amsterdam
Maggie Cheng <maggie...@iit.edu> Illinois Institute of Technology, USA
Farshad Firouzi <farshad...@duke.edu> Duke University, USA
Lars Pedersen <l...@clin.au.dk> Professor in clinical data science, Department of Clinical Medicine, Department of Clinical Epidemiology, Aarhus University
Matthias Hilty <Matthia...@usz.ch> The RISC-19-ICU registry board, Institute of Intensive Care Medicine, University of Zurich and University Hospital of Zurich, Zurich, Switzerland
Reto Schuepbach <Reto.Sc...@usz.ch> The RISC-19-ICU registry board, Institute of Intensive Care Medicine, University of Zurich and University Hospital of Zurich, Zurich, Switzerland