Manifold ranking based scoring system with its application to cardiac arrest prediction: A retrospective study in emergency department patients

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.

Diversity in Machine Learning: A Systematic Review of Text-Based Diagnostic Applications

OBJECTIVE

 As the storage of clinical data has transitioned into electronic formats, medical informatics has become increasingly relevant in providing diagnostic aid. The purpose of this review is to evaluate machine learning models that use text data for diagnosis and to assess the diversity of the included study populations.

METHODS

 We conducted a systematic literature review on three public databases. Two authors reviewed every abstract for inclusion. Articles were included if they used or developed machine learning algorithms to aid in diagnosis. Articles focusing on imaging informatics were excluded.

RESULTS

 From 2,260 identified papers, we included 78. Of the machine learning models used, neural networks were relied upon most frequently (44.9%). Studies had a median population of 661.5 patients, and diseases and disorders of 10 different body systems were studied. Of the 35.9% (N = 28) of papers that included race data, 57.1% (N = 16) of study populations were majority White, 14.3% were majority Asian, and 7.1% were majority Black. In 75% (N = 21) of papers, White was the largest racial group represented. Of the papers included, 43.6% (N = 34) included the sex ratio of the patient population.

DISCUSSION

 With the power to build robust algorithms supported by massive quantities of clinical data, machine learning is shaping the future of diagnostics. Limitations of the underlying data create potential biases, especially if patient demographics are unknown or not included in the training.

CONCLUSION

 As the movement toward clinical reliance on machine learning accelerates, both recording demographic information and using diverse training sets should be emphasized. Extrapolating algorithms to demographics beyond the original study population leaves large gaps for potential biases.

Decision tree model for predicting in-hospital cardiac arrest among patients admitted with acute coronary syndrome

BACKGROUND

In-hospital cardiac arrest (IHCA) may be preventable, with patients often showing signs of physiological deterioration before an event. Our objective was to develop and validate a simple clinical prediction model to identify the IHCA risk among cardiac arrest (CA) patients hospitalized with acute coronary syndrome (ACS).

HYPOTHESIS

A predicting model could help to identify the risk of IHCA among patients admitted with ACS.

METHODS

We conducted a case-control study and analyzed 21 337 adult ACS patients, of whom 164 had experienced CA. Vital signs, demographic, and laboratory data were extracted from the electronic health record. Decision tree analysis was applied with 10-fold cross-validation to predict the risk of IHCA.

RESULTS

The decision tree analysis detected seven explanatory variables, and the variables' importance is as follows: VitalPAC Early Warning Score (ViEWS), fatal arrhythmia, Killip class, cardiac troponin I, blood urea nitrogen, age, and diabetes. The development decision tree model demonstrated a sensitivity of 0.762, a specificity of 0.882, and an area under the receiver operating characteristic curve (AUC) of 0.844 (95% CI, 0.805 to 0.849). A 10-fold cross-validated risk estimate was 0.198, while the optimism-corrected AUC was 0.823 (95% CI, 0.786 to 0.860).

CONCLUSIONS

We have developed and internally validated a good discrimination decision tree model to predict the risk of IHCA. This simple prediction model may provide healthcare workers with a practical bedside tool and could positively impact decision-making with regard to deteriorating patients with ACS.

SVR ensemble-based continuous blood pressure prediction using multi-channel photoplethysmogram

In this paper, a continuous non-occluding blood pressure (BP) prediction method is proposed using multiple photoplethysmogram (PPG) signals. In the new method, BP is predicted by a committee machine or ensemble learning framework comprising multiple support vector regression (SVR) machines. The existing methods for continuous BP prediction rely on a single calibration model obtained from a single arterial segment. Our ensemble framework is the first BP estimation method which uses multiple SVR models for calibration from multiple arterial segments. This permits reducing of the mean prediction error and the risk of overfitting associated with a single model. Each SVR in the ensemble is trained on a comprehensive feature set that is constructed from a distinct PPG segment. The feature set includes pulse morphological parameters such as systolic pulse amplitude and area under the curve, heart rate variability (HRV) frequency, time domain parameters and the pulse wave velocity (PWV). Empirical evaluation using 40 volunteers with no serious health conditions shows that the proposed method is more reliable for estimating both the systolic and diastolic BP than similar methods employing a single calibration model under identical settings. Moreover, the combined output is found to be more stable than the output of any of the constituent models in the ensemble for both the systolic and diastolic cases.

Electrocardiogram Sampling Frequency Range Acceptable for Heart Rate Variability Analysis

Objectives

Heart rate variability (HRV) has gained recognition as a noninvasive marker of autonomic activity. HRV is considered a promising tool in various clinical scenarios. The optimal electrocardiogram (ECG) sampling frequency required to ensure sufficient precision of R–R intervals for HRV analysis has not yet been determined. Here, we aimed to determine the acceptable ECG sampling frequency range by analyzing ECG signals from patients who visited an emergency department with the chief complaint of acute intoxication or overdose.

Methods

The study included 83 adult patients who visited an emergency department with the chief complaint of acute poisoning. The original 1,000-Hz ECG signals were down-sampled to 500-, 250-, 100-, and 50-Hz sampling frequencies with linear interpolation. R–R interval data were analyzed for time-domain, frequency-domain, and nonlinear HRV parameters. Parameters derived from the data on down-sampled frequencies were compared with those derived from the data on 1,000-Hz signals, and Lin's concordance correlation coefficients were calculated.

Results

Down-sampling to 500 or 250 Hz resulted in excellent concordance. Signals down-sampled to 100 Hz produced acceptable results for time-domain analysis and Poincaré plots, but not for frequency-domain analysis. Down-sampling to 50 Hz proved to be unacceptable for both time- and frequency-domain analyses. At 50 Hz, the root-mean-squared successive differences and the power of high frequency tended to have high values and random errors.

Conclusions

A 250-Hz sampling frequency would be acceptable for HRV analysis. When frequency-domain analysis is not required, a 100-Hz sampling frequency would also be acceptable.

Artificial intelligence and machine learning in emergency medicine

Interest in artificial intelligence (AI) research has grown rapidly over the past few years, in part thanks to the numerous successes of modern machine learning techniques such as deep learning, the availability of large datasets and improvements in computing power. AI is proving to be increasingly applicable to healthcare and there is a growing list of tasks where algorithms have matched or surpassed physician performance. Despite the successes there remain significant concerns and challenges surrounding algorithm opacity, trust and patient data security. Notwithstanding these challenges, AI technologies will likely become increasingly integrated into emergency medicine in the coming years. This perspective presents an overview of current AI research relevant to emergency medicine.