Different Stroke Scales; Which Scale or Scales Should Be Used?

Predicting stroke severity of patients using interpretable machine learning algorithms

BACKGROUND

Stroke is a significant global health concern, ranking as the second leading cause of death and placing a substantial financial burden on healthcare systems, particularly in low- and middle-income countries. Timely evaluation of stroke severity is crucial for predicting clinical outcomes, with standard assessment tools being the Rapid Arterial Occlusion Evaluation (RACE) and the National Institutes of Health Stroke Scale (NIHSS). This study aims to utilize Machine Learning (ML) algorithms to predict stroke severity using these two distinct scales.

METHODS

We conducted this study using two datasets collected from hospitals in Urmia, Iran, corresponding to stroke severity assessments based on RACE and NIHSS. Seven ML algorithms were applied, including K-Nearest Neighbor (KNN), Decision Tree (DT), Random Forest (RF), Adaptive Boosting (AdaBoost), Extreme Gradient Boosting (XGBoost), Support Vector Machine (SVM), and Artificial Neural Network (ANN). Hyperparameter tuning was performed using grid search to optimize model performance, and SHapley Additive Explanations (SHAP) were used to interpret the contribution of individual features.

RESULTS

Among the models, the RF achieved the highest performance, with accuracies of 92.68% for the RACE dataset and 91.19% for the NIHSS dataset. The Area Under the Curve (AUC) was 92.02% and 97.86% for the RACE and NIHSS datasets, respectively. The SHAP analysis identified triglyceride levels, length of hospital stay, and age as critical predictors of stroke severity.

CONCLUSIONS

This study is the first to apply ML models to the RACE and NIHSS scales for predicting stroke severity. The use of SHAP enhances the interpretability of the models, increasing clinicians' trust in these ML algorithms. The best-performing ML model can be a valuable tool for assisting medical professionals in predicting stroke severity in clinical settings.

Mediatory role of the serum mineral level and discharge disability of stroke survivors

Background

Possible association between minerals contributing and mortality related to stroke were seen, but prospective data on the relation of vitamin D, magnesium and calcium serum levels with stroke were inconsistent. Consideration about the potential health effects of minerals and nutrients, the current study was conducted.

Methods

This analytical cross-sectional study was conducted on 216 stroke survivors who were referred to the Ayatollah Rouhani Hospital of Babol, Iran. Demographic characteristics, clinical variables, and serum mineral levels were completed in the checklist. Admit score and discharge scale of these patients were determined according to the National Institute of Health Stroke Scale. A path model was constructed to explore the interrelationship between variables and to verify the relationship between variables and disability discharges.

Results

Of 216 stroke patients, 185 (85.6%) cases were ischemic. The discharge status of 29 (12.9%) cases were severe or expired. The patients with moderate and severe admit scores, hemorrhagic stroke type, diabetes mellitus, hypertension and live in the village significantly had a poor discharge disability scale (all of p<0.05). Of all direct paths, Mg (β=-2.85), and among indirect paths, calcium(β=-3.59) had the highest effect on the discharge scale. Only mg had affected the discharge scale through direct and indirect (β=-2.45) paths and had the greatest reverse effect on the discharge scale (β=-5.30; totally).

Conclusion

Hypomagnesemia and hypocalcemia play a mediatory role in poor outcomes. Especially, hypomagnesemia was the direct parameter for poor outcomes. The independent role of each mineral in this issue is difficult to define and suggested for future study.

A Machine Learning Approach to Support Urgent Stroke Triage Using Administrative Data and Social Determinants of Health at Hospital Presentation: Retrospective Study

BACKGROUND

The key to effective stroke management is timely diagnosis and triage. Machine learning (ML) methods developed to assist in detecting stroke have focused on interpreting detailed clinical data such as clinical notes and diagnostic imaging results. However, such information may not be readily available when patients are initially triaged, particularly in rural and underserved communities.

OBJECTIVE

This study aimed to develop an ML stroke prediction algorithm based on data widely available at the time of patients' hospital presentations and assess the added value of social determinants of health (SDoH) in stroke prediction.

METHODS

We conducted a retrospective study of the emergency department and hospitalization records from 2012 to 2014 from all the acute care hospitals in the state of Florida, merged with the SDoH data from the American Community Survey. A case-control design was adopted to construct stroke and stroke mimic cohorts. We compared the algorithm performance and feature importance measures of the ML models (ie, gradient boosting machine and random forest) with those of the logistic regression model based on 3 sets of predictors. To provide insights into the prediction and ultimately assist care providers in decision-making, we used TreeSHAP for tree-based ML models to explain the stroke prediction.

RESULTS

Our analysis included 143,203 hospital visits of unique patients, and it was confirmed based on the principal diagnosis at discharge that 73% (n=104,662) of these patients had a stroke. The approach proposed in this study has high sensitivity and is particularly effective at reducing the misdiagnosis of dangerous stroke chameleons (false-negative rate <4%). ML classifiers consistently outperformed the benchmark logistic regression in all 3 input combinations. We found significant consistency across the models in the features that explain their performance. The most important features are age, the number of chronic conditions on admission, and primary payer (eg, Medicare or private insurance). Although both the individual- and community-level SDoH features helped improve the predictive performance of the models, the inclusion of the individual-level SDoH features led to a much larger improvement (area under the receiver operating characteristic curve increased from 0.694 to 0.823) than the inclusion of the community-level SDoH features (area under the receiver operating characteristic curve increased from 0.823 to 0.829).

CONCLUSIONS

Using data widely available at the time of patients' hospital presentations, we developed a stroke prediction model with high sensitivity and reasonable specificity. The prediction algorithm uses variables that are routinely collected by providers and payers and might be useful in underresourced hospitals with limited availability of sensitive diagnostic tools or incomplete data-gathering capabilities.

The prognostic value of the Barthel Index for mortality in patients with COVID-19: A cross-sectional study

Objective

This study aimed to analyze the association between the activity of daily living (ADL), coronavirus disease (COVID-19), and the value of the Barthel Index in predicting the prognosis of patients.

Methods

This study included 398 patients with COVID-19, whose ADL at admission to hospital were assessed with the Barthel Index. The relationship between the index and the mortality risk of the patients was analyzed. Several regression models and a decision tree were established to evaluate the prognostic value of the index in COVID-19 patients.

Results

The Barthel Index scores of deceased patients were significantly lower than that of discharged patients (median: 65 vs. 90, P < 0.001), and its decrease indicated an increased risk of mortality in patients (P < 0.001). After adjusting models for age, gender, temperature, pulse, respiratory rate, mean arterial pressure, oxygen saturation, etc., the Barthel Index could still independently predict prognosis (OR = 0.809; 95% CI: 0.750-0.872). The decision tree showed that patients with a Barthel Index of below 70 had a higher mortality rate (33.3-40.0%), while those above 90 were usually discharged (mortality: 2.7-7.2%).

Conclusion

The Barthel Index is of prognostic value for mortality in COVID-19 patients. According to their Barthel Index, COVID-19 patients can be divided into emergency, observation, and normal groups (0-70; 70-90; 90-100), with different treatment strategies.

Two-Stage Clinical Model for Screening the Suspected Cases of Acute Ischemic Stroke in Need of Imaging in Emergency Department; a Cross-sectional Study

Introduction

Just as failure to diagnose an acute ischemic stroke (AIS) in a timely manner affects the patient's outcome; an inaccurate and misplaced impression of the AIS diagnosis is not without its drawbacks. Here, we introduce a two-stage clinical tool to aid in the screening of AIS cases in need of imaging in the emergency department (ED).

Methods

This was a multicenter cross-sectional study, in which suspected AIS patients who underwent a brain magnetic resonance imaging (MRI) were included. The 18 variables from nine existing AIS screening tools were extracted and a two-stage screening tool was developed based on expert opinion (stage-one or rule in stage) and multivariate logistic regression analysis (stage-two or rule out stage). Then, the screening performance characteristics of the two-stage mode was evaluated.

Results

Data from 803 patients with suspected AIS were analyzed. Among them, 57.4 % were male, and their overall mean age was 66.9 ± 13.9 years. There were 561 (69.9%) cases with a final confirmed diagnosis of AIS. The total sensitivity and specificity of the two-stage screening model were 99.11% (95% CI: 98.33 to 99.89) and 35.95% (95% CI: 29.90 to 42.0), respectively. Also, the positive and negative predictive values of two-stage screening model were 78.20% (95% CI: 75.17 to 81.24) and 94.57% (95% CI: 89.93 to 81.24), respectively. The area under the receiver operating characteristic (ROC) curve of the two-stage screening model for AIS was 67.53% (95% CI: 64.48 to 70.58). Overall, using the two-stage screening model presented in this study, more than 11% of suspected AIS patients were not referred for MRI, and the error of this model is about 5%.

Conclusion

Here, we proposed a 2-step model for approaching suspected AIS patients in ED for an attempt to safely exclude patients with the least probability of having an AIS as a diagnosis. However, further surveys are required to assess its accuracy and it may even need some modifications.

Caspian Nursing Process: Impactions on New-Onset Constipations in Admission, Discharge, and Follow-up of Acute Stroke Patients

Background

Structural planning is essential for the management of constipation in stroke patients. The current study aims to determine the impact of a care plan on the frequency of new-onset constipation following stroke. Materials and Methods.

Materials and Methods

This clinical trial was conducted on 132 stroke patients (two groups of 66) in three phases (pre-intervention, during discharge, 1 month after discharge). Clients were randomly assigned to blocks based on gender, type of stroke, and age. The care plan according to the nursing process was conducted. Data collection tools included a demographic-clinical information questionnaire, Rome IV criteria (diagnosis of constipation), and Bristol scale (consistency of stool). Data were analyzed using the Chi-square, McNemar, Wilcoxon, Analysis of Variance (ANOVA), and a general estimated model.

Results

The prevalence of new-onset constipation following stroke in the control group decreased from 66 (100%) at admission to 39 (67.20%) at discharge and in the intervention group from 66 cases (100%) to 18 cases (34%) (p = 0.001), but it was not significant at follow-up (p = 0.16). The trend of frequency of constipation from admission to follow-up was generally significant in the intervention group (p = 0.03) vs the control group (p = 0.21). The difference in the mean number of cases of constipation was statistically significant (2.89) 2.10) control group vs 1.58 (1.65) intervention group, p < 0.001).

Conclusions

A significant impact of the care plan was observed from admission to discharge, but further follow-up was required with more client-side collaboration. Therefore, the present care plan is recommended in the hospital and home care.

Use of stroke scales in clinical practice: Current concepts

With stroke being the leading global cause of disability in adults, the use of clinical rating scales in stroke patients is important not only for diagnostic and therapeutic purposes but also for prognostic and care implications. Scales that quantify neurological disability can be particularly useful for assessing and guiding decisions in acute management and rehabilitative treatment. We analyzed and discussed some of the main rating scales most used in stroke in clinical practice, which measure both acute neurological deficit and functional outcome in stroke. In acute stroke, it is that in most cases, the scales evaluate a neurological deficit attributable to an alteration of the anterior and not posterior circulation and most of them assess a moderate stroke rather than a mild or severe one. In a rehabilitation treatment, they are sometimes too simplified; thus, the patient can reach a near-normal score and can have significant cognitive deficits that can affect both the possibility of communication and the reliability of responses. A patient with autonomy in the activities of daily living may not be completely autonomous. In future, the use of composite rating scales could be useful for a detailed measurement of neurological deficits in acute stroke and better assess the efficacy of a treatment and functional outcome.