Evaluation of machine learning-based models for prediction of clinical deterioration: A systematic literature review

Jahandideh, Sepideh; Ozavci, Guncag; Sahle, Berhe W.; Kouzani, Abbas Z.; Magrabi, Farah; Bucknall, Tracey

doi:10.1016/j.ijmedinf.2023.105084

Cited by 11 publications

(4 citation statements)

References 50 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These algorithms make use of vital signs (such as heart rate and blood pressure), laboratory results (such as white blood cell count and lactate level), and clinical factors (such as age and comorbidities) to predict the likelihood of sepsis. High sensitivity, specificity, and area under the receiver operating characteristic curve (AUC-ROC) values have been reported for these models by researchers, demonstrating their propensity for sepsis prediction [18,[22][23][24][25][26][27][28][29]. In response to the persisting challenge of sepsis-related fatalities, Wang et al [30] sought to develop an AI algorithm for early sepsis prediction, successfully creating a random forest model utilizing 55 clinical features from ICU patient data, yielding an AUC of 0.91, 87% sensitivity, and 89% specificity, with potential wider applicability pending external validation.…”

Section: Introductionmentioning

confidence: 91%

See 1 more Smart Citation

Machine Learning-Based Early Prediction of Sepsis Using Electronic Health Records: A Systematic Review

Islam,

Prithula,

Kumar

et al. 2023

JCM

View full text Add to dashboard Cite

Background: Sepsis, a life-threatening infection-induced inflammatory condition, has significant global health impacts. Timely detection is crucial for improving patient outcomes as sepsis can rapidly progress to severe forms. The application of machine learning (ML) and deep learning (DL) to predict sepsis using electronic health records (EHRs) has gained considerable attention for timely intervention. Methods: PubMed, IEEE Xplore, Google Scholar, and Scopus were searched for relevant studies. All studies that used ML/DL to detect or early-predict the onset of sepsis in the adult population using EHRs were considered. Data were extracted and analyzed from all studies that met the criteria and were also evaluated for their quality. Results: This systematic review examined 1942 articles, selecting 42 studies while adhering to strict criteria. The chosen studies were predominantly retrospective (n = 38) and spanned diverse geographic settings, with a focus on the United States. Different datasets, sepsis definitions, and prevalence rates were employed, necessitating data augmentation. Heterogeneous parameter utilization, diverse model distribution, and varying quality assessments were observed. Longitudinal data enabled early sepsis prediction, and quality criteria fulfillment varied, with inconsistent funding–article quality correlation. Conclusions: This systematic review underscores the significance of ML/DL methods for sepsis detection and early prediction through EHR data.

show abstract

Section: Introductionmentioning

confidence: 91%

“…Jahandideh et al (2023) [25] explored the use of ML techniques for predicting patient clinical deterioration in hospitals. A total of 29 primary studies were identified, utilizing various ML models, including supervised, unsupervised, and classical techniques.…”

Section: Summaries Of Recent Systematic Reviews In Relevant Fieldsmentioning

confidence: 99%

Machine Learning-Based Early Prediction of Sepsis Using Electronic Health Records: A Systematic Review

Islam,

Prithula,

Kumar

et al. 2023

JCM

View full text Add to dashboard Cite

show abstract

“…With the introduction of electronic health record (EHR) systems in healthcare institutions, a new research paradigm of statistical, machine, and deep learning methodologies has emerged. Many researchers have used the clinical data of patients for disease inference, clinical deterioration, decision support systems, optimization, and outcome prediction [10,11]. One of the most prevalent research areas is length of stay prediction.…”

Section: Related Workmentioning

confidence: 99%

Predicting the Length of Stay of Cardiac Patients Based on Pre-Operative Variables—Bayesian Models vs. Machine Learning Models

Abdurrab,

Mahmood,

Sheikh

et al. 2024

Healthcare

View full text Add to dashboard Cite

Length of stay (LoS) prediction is deemed important for a medical institution’s operational and logistical efficiency. Sound estimates of a patient’s stay increase clinical preparedness and reduce aberrations. Various statistical methods and techniques are used to quantify and predict the LoS of a patient based on pre-operative clinical features. This study evaluates and compares the results of Bayesian (simple Bayesian regression and hierarchical Bayesian regression) models and machine learning (ML) regression models against multiple evaluation metrics for the problem of LoS prediction of cardiac patients admitted to Tabba Heart Institute, Karachi, Pakistan (THI) between 2015 and 2020. In addition, the study also presents the use of hierarchical Bayesian regression to account for data variability and skewness without homogenizing the data (by removing outliers). LoS estimates from the hierarchical Bayesian regression model resulted in a root mean squared error (RMSE) and mean absolute error (MAE) of 1.49 and 1.16, respectively. Simple Bayesian regression (without hierarchy) achieved an RMSE and MAE of 3.36 and 2.05, respectively. The average RMSE and MAE of ML models remained at 3.36 and 1.98, respectively.

show abstract

“…Thus, a recent review found that out of 29 studies using machine learning (AI) to predict clinical deterioration, 15 different approaches were used, many rendering the average clinician helpless in understanding the underlying methodology. The main conclusion was that real-world application and effectiveness is needed [20]. patients had blinded CVSM-data collection for up to 4 days and were subsequently followed for clinical complications until postoperative day 30 [3 & ] Significant vital sign deviations occurred in almost all patients with and without complications, with 96% of all patients having at least one episode of SpO 2 <92% similar to other vital sign without significant differences between patient groups.…”

Section: Key Pointsmentioning

confidence: 99%

The future of postoperative vital sign monitoring in general wards: improving patient safety through continuous artificial intelligence-enabled alert formation and reduction

Aasvang,

Meyhoff

2023

Current Opinion in Anaesthesiology

View full text Add to dashboard Cite

Purpose Monitoring of vital signs at the general ward with continuous assessments aided by artificial intelligence (AI) is increasingly being explored in the clinical setting. This review aims to describe current evidence for continuous vital sign monitoring (CVSM) with AI-based alerts − from sensor technology, through alert reduction, impact on complications, and to user-experience during implementation. Recent findings CVSM identifies significantly more vital sign deviations than manual intermittent monitoring. This results in high alert generation without AI-evaluation, both in patients with and without complications. Current AI is at the rule-based level, and this potentially reduces irrelevant alerts and identifies patients at need. AI-aided CVSM identifies complications earlier with reduced staff workload and a potential reduction of severe complications. Summary The current evidence for AI-aided CSVM suggest a significant role for the technology in reducing the constant 10–30% in-hospital risk of severe postoperative complications. However, large, randomized trials documenting the benefit for patient improvements are still sparse. And the clinical uptake of explainable AI to improve implementation needs investigation.

show abstract

Evaluation of machine learning-based models for prediction of clinical deterioration: A systematic literature review

Cited by 11 publications

References 50 publications

Machine Learning-Based Early Prediction of Sepsis Using Electronic Health Records: A Systematic Review

Machine Learning-Based Early Prediction of Sepsis Using Electronic Health Records: A Systematic Review

Predicting the Length of Stay of Cardiac Patients Based on Pre-Operative Variables—Bayesian Models vs. Machine Learning Models

The future of postoperative vital sign monitoring in general wards: improving patient safety through continuous artificial intelligence-enabled alert formation and reduction

Contact Info

Product

Resources

About