A new method for early mortality prediction in intensive care units (ICUs) has been published today by a team from the University of Portsmouth and Queen Alexandra Hospital.
This study highlights the main data challenges in early mortality prediction for patients in intensive care and introduces a new machine learning-based framework for Early Mortality Prediction for Intensive Care Unit patients (EMPICU).
The project leader Dr Mohamed Bader El-Den hopes that this research, published in the International Journal of Medical Informatics, will eventually help doctors recognise which patients are most at risk and which patients are most likely to benefit from intensive care treatment.
It is intended that risk prediction will enable patients and relatives to be better informed about the potential benefits of intensive care treatment, before choosing which treatments to accept.
Intensive care units are where the most seriously ill patients are treated, with interventions that are both costly and invasive. Early mortality risk prediction may help patients to understand risk more fully and doctors to advise them. It may also help doctors to manage resources more effectively.
He said: “Intensive care units are where the most seriously ill patients are treated, with interventions that are both costly and invasive. Early mortality risk prediction may help patients to understand risk more fully and doctors to advise them. It may also help doctors to manage resources more effectively.”
Also working on the project is Aya Awad from the University of Portsmouth. The project is part of her PhD study. She hopes that when putting the proposed EMPICU framework into practice it will relieve the patient symptoms, prevent complications and prolong the patient’s life.
This new study should draw the attention of the medical and data science communities to the importance and feasibility of early death prediction in intensive care units.
She said: “This new study should draw the attention of the medical and data science communities to the importance and feasibility of early death prediction in intensive care units.”
The idea of the project was motivated by Dr James McNicholas who is a consultant in intensive care medicine at Queen Alexandra and a member of the project team. He said: “The study seeks to show that modern data science may provide valuable tools to enable shared decision-making, by doctors and patients, early in critical care admission.
The study seeks to show that modern data science may provide valuable tools to enable shared decision-making, by doctors and patients, early in critical care admission. Predicting mortality in intensive care units is a significant challenge for the critical care community. In the past, most research has focused on severity of illness scoring systems or data mining models designed for risk estimation at least 24 or 48 hours after admission.
“Predicting mortality in intensive care units is a significant challenge for the critical care community. In the past, most research has focused on severity of illness scoring systems or data mining models designed for risk estimation at least 24 or 48 hours after admission.
“This study makes an early prediction of mortality within the first six hours of the patient being admitted to intensive care. We hope to refine the methodology for greater accuracy and clinical utility.”
This study makes an early prediction of mortality within the first six hours of the patient being admitted to intensive care. We hope to refine the methodology for greater accuracy and clinical utility.