Early prediction of ICU readmissions using classification algorithms

Early prediction of ICU readmissions using classification algorithms
Computers in Biology and Medicine; Mar 2020; v118, March 2020, 103636 https://doi.org/10.1016/j.compbiomed.2020.103636

• Existing work on predicting ICU readmissions relies on information available at the time of discharge, however, in order to be more useful and to prevent complications, predictions need to be made earlier. This work investigates the hypothesis that the basal characteristics and information collected at the time of the patient's admission can enable accurate predictions of ICU readmission.

Abstract:

Analysis and prediction of unplanned intensive care unit readmission

Analysis and prediction of unplanned intensive care unit readmission using recurrent neural networks with long short-term memory
PLoS ONE 14(7): e0218942. 8 July 2019

• Research using machine learning methods on comprehensive, longitudinal clinical data from the MIMIC-III to predict the ICU readmission of patients within 30 days of their discharge.
• Machine learning techniques such as Recurrent Neural Networks (RNN) with Long Short-Term Memory (LSTM) were used to incorporate the multivariate features of EHRs and capture sudden fluctuations in chart event features. These machine learning models identify ICU readmissions at a higher sensitivity rate and an improved Area Under the Curve compared with traditional methods.

Can nurse specialist working practices reduce the burdens of lung cancer?

Can nurse specialist working practices reduce the burdens of lung cancer?
University of Nottingham, 4 July 2018

• A new study has found that patients with lung cancer experience significantly better outcomes in terms of life expectancy, avoiding unnecessary hospital admissions and managing the effects of treatment when cared for by specialist lung cancer nurses.
• The research was based on anonymised patient healthcare records available from PHE and a nationwide survey of lung cancer nurse specialists (LCNS), analysing more than 100,000 people with lung cancer and more than 200 nurses across England.
• The research carried out by the University of Nottingham and London South Bank University, was presented at the National Cancer Registration and Analysis Service (NCRAS) conference on 21 June 2018.

Predicting ICU readmission with machine learning using electronic health record data

Predicting intensive care unit readmission with machine learning using electronic health record data [Full text available using NHS OpenAthens ID]
Annals of the American Thoracic Society; Jul 2018; vol. 15 (no. 7); p. 846-853

• A range of patient characteristics were extracted from the electronic health record of ICU patients transferred to wards using a gradient boosted machine model. The machine-learning-derived model had significantly better performance  than either the Stability and Workload Index for Transfer score  or Modified Early Warning Score.
• Study: observational study, USA, n=24,885 ICU transfers to wards, 2834 readmissions (11%) 

Abstract:

The utility of ICU readmission as a quality indicator and the effect of selection

The utility of ICU readmission as a quality indicator and the effect of selection [PubMed]
Critical Care Medicine; May 2018; vol. 46 (no. 5); p. 749-756

• Retrospective cohort study of adult patients admitted to 262 ICUs in the UK (n=682,975), of which 1.53% were readmitted within 2 days of discharge.
• Post-ICU admission hospital mortality and ICU readmission were poorly correlated (r = 0.130)

Abstract:

Readmission to the Intensive Care Unit Following Cardiac Surgery: A Derived and Validated Risk Prediction Model in 4,869 Patients

Readmission to the Intensive Care UnitFollowing Cardiac Surgery: A Derived and Validated Risk Prediction Model in4,869 Patients (in press)
Journal of Cardiothoracic and Vascular Anesthesia, 2018 DOI: https://doi.org/10.1053/j.jvca

• From a comprehensive perioperative dataset, the authors derived and internally validated a risk index incorporating 9 easily identifiable and routinely collected variables to predict readmission following cardiac surgery.
• Study: Retrospective nonrandomized study, analysis of 4,869 consecutive adult patients. 156 patients (3.2%) readmitted. St George's Hospital, London

Abstract:

Predicting risk of unplanned hospital readmission in survivors of critical illness

Predicting risk of unplanned hospital readmission in survivors of critical illness: a population-level cohort study
Thorax Online First: 05 April 2018. doi: 10.1136/thoraxjnl-2017-210822

• A cohort study of ICU survivors (n=55,975) found that pre-existing illness indices are better predictors of readmission than acute illness factors and identifying additional patient-centred drivers of readmission may improve risk prediction models. 
• Study: Edinburgh, UK

Abstract

Prediction of early unplanned intensive care unit readmission in a UK tertiary care hospital: A cross-sectional machine learning approach

Prediction of early unplanned intensive care unit readmission in a UK tertiary care hospital: A cross-sectional machine learning approach
Desautels T.; Das R.; Calvert J.; Trivedi M.; Summers C.; Ercole A.; Wales D.J.
BMJ Open; Sep 2017; vol. 7 (no. 9)

• Analysis of data for patients (n=2018) using a novel machine learning algorithm based on transfer learning could discriminate between patients with unplanned ICU readmission or death outcome and those without this outcome, and was superior to the purpose-built Stability and Workload Index for Transfer (SWIFT) score.
• Setting: ICU Addenbrooke’s Hospital, Cambridge, UK. General and neuroscience ICU episodes collected between October 2014 and August 2016, Patients >= 16yrs