Dataset Information

Towards a decision support tool for intensive care discharge: machine learning algorithm development using electronic healthcare data from MIMIC-III and Bristol, UK.

ABSTRACT:

Objective

The primary objective is to develop an automated method for detecting patients that are ready for discharge from intensive care.

Design

We used two datasets of routinely collected patient data to test and improve on a set of previously proposed discharge criteria.

Setting

Bristol Royal Infirmary general intensive care unit (GICU).

Patients

Two cohorts derived from historical datasets: 1870 intensive care patients from GICU in Bristol, and 7592 from Medical Information Mart for Intensive Care (MIMIC)-III.

Results

In both cohorts few successfully discharged patients met all of the discharge criteria. Both a random forest and a logistic classifier, trained using multiple-source cross-validation, demonstrated improved performance over the original criteria and generalised well between the cohorts. The classifiers showed good agreement on which features were most predictive of readiness-for-discharge, and these were generally consistent with clinical experience. By weighting the discharge criteria according to feature importance from the logistic model we showed improved performance over the original criteria, while retaining good interpretability.

Conclusions

Our findings indicate the feasibility of the proposed approach to ready-for-discharge classification, which could complement other risk models of specific adverse outcomes in a future decision support system. Avenues for improvement to produce a clinically useful tool are identified.

SUBMITTER: McWilliams CJ

PROVIDER: S-EPMC6429919 | biostudies-literature | 2019 Mar

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Towards a decision support tool for intensive care discharge: machine learning algorithm development using electronic healthcare data from MIMIC-III and Bristol, UK.

McWilliams Christopher J CJ Lawson Daniel J DJ Santos-Rodriguez Raul R Gilchrist Iain D ID Champneys Alan A Gould Timothy H TH Thomas Mathew Jc MJ Bourdeaux Christopher P CP

BMJ open 20190307 3

<h4>Objective</h4>The primary objective is to develop an automated method for detecting patients that are ready for discharge from intensive care.<h4>Design</h4>We used two datasets of routinely collected patient data to test and improve on a set of previously proposed discharge criteria.<h4>Setting</h4>Bristol Royal Infirmary general intensive care unit (GICU).<h4>Patients</h4>Two cohorts derived from historical datasets: 1870 intensive care patients from GICU in Bristol, and 7592 from Medical ...[more]

PMID: 30850412

Similar Datasets

Project description:Low heart rate is a risk factor of mortality in many cardiovascular diseases. However, the relationship of minimum heart rate (MHR) with outcomes after cardiac surgery is still unclear, and the association between optimum MHR and risk of mortality in patients receiving cardiac surgery remains unknown. In this retrospective study using the Multi-parameter Intelligent Monitoring in Intensive Care (MIMIC-III) database, 8243 adult patients who underwent cardiac surgery were included. The association between MHR and the 30-day, 90-day, 180-day, and 1-year mortality of patients undergoing cardiac surgery was analyzed using multivariate Cox proportional hazard analysis. As a continuous variable, MHR was evaluated using restricted cubic regression splines, and appropriate cut-off points were determined. Kaplan-Meier curve was used to further explore the relationship between MHR and prognosis. Subgroup analyses were performed based on age, sex, hypertension, diabetes, and ethnicity. The rates of the 30-day, 90-day, 180-day, and 1-year mortalities of patients in the low MHR group were higher than those in the high MHR group (4.1% vs. 2.9%, P < 0.05; 6.8% vs. 5.3%, P < 0.05; 8.9% vs. 7.0%, P < 0.05, and 10.9% vs. 8.8%, P < 0.05, respectively). Low MHR significantly correlated with the 30-day, 90-day, 180-day, and 1-year mortality after adjusting for confounders. A U-shaped relationship was observed between the 30-day, 90-day, 180-day, and 1-year mortality and MHR, and the mortality was lowest when the MHR was 69 bpm. Kaplan-Meier curve analysis also indicated that low MHR had poor prognosis in patients undergoing cardiac surgery. According to subgroup analyses, the effect of low MHR on post-cardiac surgery survival was restricted to patients who were < 75 years old, male, without hypertension and diabetes, and of White ethnicity. MHR (69 bpm) was associated with better 30-day, 90-day, 180-day, and 1-year survival in patients after cardiac surgery. Therefore, effective HR control strategies are required in this high-risk population.

Project description:BackgroundParoxysmal Nocturnal Haemoglobinuria (PNH) is an ultra-rare, acquired disorder that is challenging to diagnose due to varied symptoms, heterogeneous patient presentations, and lack of awareness among healthcare professionals. This leads to frequent misdiagnosis and delays in diagnosis. This study evaluated the feasibility of a machine learning model to identify undiagnosed PNH patients using structured electronic health records.MethodsThe study used data from the Optimum Patient Care Research Database, which contains electronic health records from general practitioner (GP) practices across the United Kingdom. PNH patients were identified by the presence, and control patients by the absence of a PNH diagnosis code in their records. Clinical features (symptoms, diagnoses, healthcare utilisation) from 131 patients in the PNH group and 593,838 patients in the control group, were inputted to a tree-based XGBoost machine learning model to classify patients as either "positive" or "negative" for PNH suspicion. The algorithm was finalised after additional exclusions and inclusions applied. Performance was assessed using positive predictive value (PPV), recall and specificity. As the sample used to develop the algorithm was not representative of the true population prevalence, PPV was additionally adjusted to reflect performance in the wider population.ResultsOf all the patients in the PNH group, 27% were classified as positive (recall). 99.99% of the control group were classified as negative (specificity). Of all the patients classified as positive, 60.4% had a diagnosis of PNH in their record (PPV). The PPV adjusted for the population prevalence of PNH was 19.59 suggesting nearly 1 in 5 patients flagged may warrant further PNH investigation. The key clinical features in the model were aplastic anaemia, pancytopenia, haemolytic anaemia, myelodysplastic syndrome, and Budd-Chiari syndrome.ConclusionThis is the first study to combine clinical understanding of PNH with machine learning, demonstrating the ability to discriminate between PNH and control patients in retrospective electronic health records. With further investigation and validation, this algorithm could be deployed on live health data, potentially leading to earlier diagnosis for patients who currently experience long diagnostic delays or remain undiagnosed.

Project description:The rapid urbanisation of the twentieth century, along with the spread of high-consumption urban lifestyles, has led to cities becoming the dominant drivers of global anthropogenic greenhouse gas emissions. Reducing these impacts is crucial, but production-based frameworks of carbon measurement and mitigation-which encompass only a limited part of cities' carbon footprints-are much more developed and widely applied than consumption-based approaches that consider the embedded carbon effectively imported into a city. Frequently, therefore, cities are left blind to the importance of their wider consumption-related climate impacts, while at the same time left lacking effective tools to reduce them. To explore the relevance of these issues, we implement methodologies for assessing production- and consumption-based emissions at the city-level and estimate the associated emissions trajectories for Bristol, a major UK city, from 2000 to 2035. We develop mitigation scenarios targeted at reducing the former, considering potential energy, carbon and financial savings in each case. We then compare these mitigation potentials with local government ambitions and Bristol's consumption-based emissions trajectory. Our results suggest that the city's consumption-based emissions are three times the production-based emissions, largely due to the impacts of imported food and drink. We find that low-carbon investments of circa £3 billion could reduce production-based emissions by 25% in 2035. However, we also find that this represents <10% of Bristol's forecast consumption-based emissions for 2035 and is approximately equal to the mitigation achievable by eliminating the city's current levels of food waste. Such observations suggest that incorporating consumption-based emission statistics into cities' accounting and decision-making processes could uncover largely unrecognised opportunities for mitigation that are likely to be essential for achieving deep decarbonisation.

Dataset Information

Towards a decision support tool for intensive care discharge: machine learning algorithm development using electronic healthcare data from MIMIC-III and Bristol, UK.

Objective

Design

Setting

Patients

Results

Conclusions

Publications

Towards a decision support tool for intensive care discharge: machine learning algorithm development using electronic healthcare data from MIMIC-III and Bristol, UK.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets