Utilizing community level factors to improve prediction of out of hospital cardiac arrest outcome using machine learning

Application of Machine Learning for Patients With Cardiac Arrest: Systematic Review and Meta-Analysis

BACKGROUND

Currently, there is a lack of effective early assessment tools for predicting the onset and development of cardiac arrest (CA). With the increasing attention of clinical researchers on machine learning (ML), some researchers have developed ML models for predicting the occurrence and prognosis of CA, with certain models appearing to outperform traditional scoring tools. However, these models still lack systematic evidence to substantiate their efficacy.

OBJECTIVE

This systematic review and meta-analysis was conducted to evaluate the prediction value of ML in CA for occurrence, good neurological prognosis, mortality, and the return of spontaneous circulation (ROSC), thereby providing evidence-based support for the development and refinement of applicable clinical tools.

METHODS

PubMed, Embase, the Cochrane Library, and Web of Science were systematically searched from their establishment until May 17, 2024. The risk of bias in all prediction models was assessed using the Prediction Model Risk of Bias Assessment Tool.

RESULTS

In total, 93 studies were selected, encompassing 5,729,721 in-hospital and out-of-hospital patients. The meta-analysis revealed that, for predicting CA, the pooled C-index, sensitivity, and specificity derived from the imbalanced validation dataset were 0.90 (95% CI 0.87-0.93), 0.83 (95% CI 0.79-0.87), and 0.93 (95% CI 0.88-0.96), respectively. On the basis of the balanced validation dataset, the pooled C-index, sensitivity, and specificity were 0.88 (95% CI 0.86-0.90), 0.72 (95% CI 0.49-0.95), and 0.79 (95% CI 0.68-0.91), respectively. For predicting the good cerebral performance category score 1 to 2, the pooled C-index, sensitivity, and specificity based on the validation dataset were 0.86 (95% CI 0.85-0.87), 0.72 (95% CI 0.61-0.81), and 0.79 (95% CI 0.66-0.88), respectively. For predicting CA mortality, the pooled C-index, sensitivity, and specificity based on the validation dataset were 0.85 (95% CI 0.82-0.87), 0.83 (95% CI 0.79-0.87), and 0.79 (95% CI 0.74-0.83), respectively. For predicting ROSC, the pooled C-index, sensitivity, and specificity based on the validation dataset were 0.77 (95% CI 0.74-0.80), 0.53 (95% CI 0.31-0.74), and 0.88 (95% CI 0.71-0.96), respectively. In predicting CA, the most significant modeling variables were respiratory rate, blood pressure, age, and temperature. In predicting a good cerebral performance category score 1 to 2, the most significant modeling variables in the in-hospital CA group were rhythm (shockable or nonshockable), age, medication use, and gender; the most significant modeling variables in the out-of-hospital CA group were age, rhythm (shockable or nonshockable), medication use, and ROSC.

CONCLUSIONS

ML represents a currently promising approach for predicting the occurrence and outcomes of CA. Therefore, in future research on CA, we may attempt to systematically update traditional scoring tools based on the superior performance of ML in specific outcomes, achieving artificial intelligence-driven enhancements.

TRIAL REGISTRATION

PROSPERO International Prospective Register of Systematic Reviews CRD42024518949; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=518949.

Post-Cardiac arrest outcome prediction using machine learning: A systematic review and meta-analysis

BACKGROUND

Early and reliable prognostication in post-cardiac arrest patients remains challenging, with various factors linked to return of spontaneous circulation (ROSC), survival, and neurological results. Machine learning and deep learning models show promise in improving these predictions. This systematic review and meta-analysis evaluates how effective these approaches are in predicting clinical outcomes at different time points using structured data.

METHODS

This study followed PRISMA guidelines, involving a comprehensive search across PubMed, Scopus, and Web of Science databases until March 2024. Studies aimed at predicting ROSC, survival (or mortality), and neurological outcomes after cardiac arrest through the application of machine learning or deep learning techniques with structured data were included. Data extraction followed the guidelines of the CHARMS checklist, and the bias risk was evaluated using PROBAST tool. Models reporting the AUC metric with 95 % confidence intervals were incorporated into the quantitative synthesis and meta-analysis.

RESULTS

After extracting 2,753 initial records, 41 studies met the inclusion criteria, yielding 97 machine learning and 16 deep learning models. The pooled AUC for predicting favorable neurological outcomes (CPC 1 or 2) at hospital discharge was 0.871 (95 % CI: 0.813 - 0.928) for machine learning models and 0.877 (95 % CI: 0.831-0.924) across deep learning algorithms. For survival prediction, this value was found to be 0.837 (95 % CI: 0.757-0.916). Considerable heterogeneity and high risk of bias were observed, mainly attributable to inadequate management of missing data and the absence of calibration plots. Most studies focused on pre-hospital factors, with age, sex, and initial arrest rhythm being the most frequent features.

CONCLUSION

Predictive models utilizing AI-based approaches, including machine and deep learning models exhibit enhanced effectiveness compared to previous regression algorithms, but significant heterogeneity and high risk of bias limit their dependability. Evaluating state-of-the-art deep learning models tailored for tabular data and their clinical generalizability can enhance outcome prediction after cardiac arrest.

Application of multi-feature-based machine learning models to predict neurological outcomes of cardiac arrest

Cardiac arrest (CA) is a major disease burden worldwide and has a poor prognosis. Early prediction of CA outcomes helps optimize the therapeutic regimen and improve patients' neurological function. As the current guidelines recommend, many factors can be used to evaluate the neurological outcomes of CA patients. Machine learning (ML) has strong analytical abilities and fast computing speed; thus, it plays an irreplaceable role in prediction model development. An increasing number of researchers are using ML algorithms to incorporate demographics, arrest characteristics, clinical variables, biomarkers, physical examination findings, electroencephalograms, imaging, and other factors with predictive value to construct multi-feature prediction models for neurological outcomes of CA survivors. In this review, we explore the current application of ML models using multiple features to predict the neurological outcomes of CA patients. Although the outcome prediction model is still in development, it has strong potential to become a powerful tool in clinical practice.

Joint Modeling of Social Determinants and Clinical Factors to Define Subphenotypes in Out-of-Hospital Cardiac Arrest Survival: Cluster Analysis

Background

Machine learning clustering offers an unbiased approach to better understand the interactions of complex social and clinical variables via integrative subphenotypes, an approach not studied in out-of-hospital cardiac arrest (OHCA).

Objective

We conducted a cluster analysis for a cohort of OHCA survivors to examine the association of clinical and social factors for mortality at 1 year.

Methods

We used a retrospective observational OHCA cohort identified from Medicare claims data, including area-level social determinants of health (SDOH) features and hospital-level data sets. We applied k-means clustering algorithms to identify subphenotypes of beneficiaries who had survived an OHCA and examined associations of outcomes by subphenotype.

Results

We identified 27,028 unique beneficiaries who survived to discharge after OHCA. We derived 4 distinct subphenotypes. Subphenotype 1 included a distribution of more urban, female, and Black beneficiaries with the least robust area-level SDOH measures and the highest 1-year mortality (2375/4417, 53.8%). Subphenotype 2 was characterized by a greater distribution of male, White beneficiaries and had the strongest zip code-level SDOH measures, with 1-year mortality at 49.9% (4577/9165). Subphenotype 3 had the highest rates of cardiac catheterization at 34.7% (1342/3866) and the greatest distribution with a driving distance to the index OHCA hospital from their primary residence >16.1 km at 85.4% (8179/9580); more were also discharged to a skilled nursing facility after index hospitalization. Subphenotype 4 had moderate median household income at US $51,659.50 (IQR US $41,295 to $67,081) and moderate to high median unemployment at 5.5% (IQR 4.2%-7.1%), with the lowest 1-year mortality (1207/3866, 31.2%). Joint modeling of these features demonstrated an increased hazard of death for subphenotypes 1 to 3 but not for subphenotype 4 when compared to reference.

Conclusions

We identified 4 distinct subphenotypes with differences in outcomes by clinical and area-level SDOH features for OHCA. Further work is needed to determine if individual or other SDOH domains are specifically tied to long-term survival after OHCA.

Clinical Validation of Explainable Deep Learning Model for Predicting the Mortality of In-Hospital Cardiac Arrest Using Diagnosis Codes of Electronic Health Records

Background

Using deep learning for disease outcome prediction is an approach that has made large advances in recent years. Notwithstanding its excellent performance, clinicians are also interested in learning how input affects prediction. Clinical validation of explainable deep learning models is also as yet unexplored. This study aims to evaluate the performance of Deep SHapley Additive exPlanations (D-SHAP) model in accurately identifying the diagnosis code associated with the highest mortality risk.

Methods

Incidences of at least one in-hospital cardiac arrest (IHCA) for 168,693 patients as well as 1,569,478 clinical records were extracted from Taiwan's National Health Insurance Research Database. We propose a D-SHAP model to provide insights into deep learning model predictions. We trained a deep learning model to predict the 30-day mortality likelihoods of IHCA patients and used D-SHAP to see how the diagnosis codes affected the model's predictions. Physicians were asked to annotate a cardiac arrest dataset and provide expert opinions, which we used to validate our proposed method. A 1-to-4-point annotation of each record (current decision) along with four previous records (historical decision) was used to validate the current and historical D-SHAP values.

Results

A subset consisting of 402 patients with at least one cardiac arrest record was randomly selected from the IHCA cohort. The median age was 72 years, with mean and standard deviation of 69 ± 17 years. Results indicated that D-SHAP can identify the cause of mortality based on the diagnosis codes. The top five most important diagnosis codes, namely respiratory failure, sepsis, pneumonia, shock, and acute kidney injury were consistent with the physician's opinion. Some diagnoses, such as urinary tract infection, showed a discrepancy between D-SHAP and clinical judgment due to the lower frequency of the disease and its occurrence in combination with other comorbidities.

Conclusions

The D-SHAP framework was found to be an effective tool to explain deep neural networks and identify most of the important diagnoses for predicting patients' 30-day mortality. However, physicians should always carefully consider the structure of the original database and underlying pathophysiology.

Factors associated with the arrival of smartphone-activated first responders before the emergency medical services in Out-of-Hospital cardiac arrest dispatch

BACKGROUND

First responder programs were developed to speed up access to cardiopulmonary resuscitation and defibrillation for out-of-hospital cardiac arrest (OHCA) victims. Little is known about the factors influencing the efficiency of the first responders arriving before the EMS and, therefore, effectively contributing to the chain of survival.

OBJECTIVES

The primary objective of this retrospective observational study was to identify the factors associated with first responders' arrival before EMS in the context of a regional first responder program arranged to deliver automated external defibrillators on suspected OHCA scenes.

METHODS

Eight hundred ninety-six dispatches where FRs intervened were collected from 2018 to 2022. A robust Poisson regression was performed to estimate the role of the time of day, the immediate availability of a defibrillator, the type of first responder, distances between the responder, the event and the dispatched vehicle, and the nearest available defibrillator on the probability of responder arriving before EMS. Moreover, a geospatial logistic regression model was built.

RESULTS

Responders arrived before EMS in 13.4% of dispatches and delivered a shock in 0.9%. The immediate availability of a defibrillator for the responder (OR = 3.24) and special categories such as taxi drivers and police (OR = 1.74) were factors significantly associated with the responder arriving before EMS. Moreover, a geospatial effect suggested that first responder programs may have a greater impact in rural areas.

CONCLUSIONS

When dispatched to OHCA scenes, responders already carrying defibrillators could more probably reach the scene before EMS. Special first responder categories are more competitive and should be further investigated.