Simple variables predict miserable outcome after intravenous thrombolysis

Artificial Intelligence for Clinical Decision Support in Acute Ischemic Stroke: A Systematic Review

BACKGROUND

Established randomized trial-based parameters for acute ischemic stroke group patients into generic treatment groups, leading to attempts using various artificial intelligence (AI) methods to directly correlate patient characteristics to outcomes and thereby provide decision support to stroke clinicians. We review AI-based clinical decision support systems in the development stage, specifically regarding methodological robustness and constraints for clinical implementation.

METHODS

Our systematic review included full-text English language publications proposing a clinical decision support system using AI techniques for direct decision support in acute ischemic stroke cases in adult patients. We (1) describe data and outcomes used in those systems, (2) estimate the systems' benefits compared with traditional stroke diagnosis and treatment, and (3) reported concordance with reporting standards for AI in healthcare.

RESULTS

One hundred twenty-one studies met our inclusion criteria. Sixty-five were included for full extraction. In our sample, utilized data sources, methods, and reporting practices were highly heterogeneous.

CONCLUSIONS

Our results suggest significant validity threats, dissonance in reporting practices, and challenges to clinical translation. We outline practical recommendations for the successful implementation of AI research in acute ischemic stroke treatment and diagnosis.

Using a Multiclass Machine Learning Model to Predict the Outcome of Acute Ischemic Stroke Requiring Reperfusion Therapy

Prediction of functional outcome in ischemic stroke patients is useful for clinical decisions. Previous studies mostly elaborate on the prediction of favorable outcomes. Miserable outcomes, which are usually defined as modified Rankin Scale (mRS) 5–6, should be considered as well before further invasive intervention. By using a machine learning algorithm, we aimed to develop a multiclass classification model for outcome prediction in acute ischemic stroke patients requiring reperfusion therapy. This was a retrospective study performed at a stroke medical center in Taiwan. Patients with acute ischemic stroke who visited between January 2016 and December 2019 and who were candidates for reperfusion therapy were included. Clinical outcomes were classified as favorable outcome, intermediate outcome, and miserable outcome. We developed four different multiclass machine learning models (Logistic Regression, Supportive Vector Machine, Random Forest, and Extreme Gradient Boosting) to predict clinical outcomes and compared their performance to the DRAGON score. A sample of 590 patients was included in this study. Of them, 180 (30.5%) had favorable outcomes and 152 (25.8%) had miserable outcomes. All selected machine learning models outperformed the DRAGON score on accuracy of outcome prediction (Logistic Regression: 0.70, Supportive Vector Machine: 0.67, Random Forest: 0.69, and Extreme Gradient Boosting: 0.67, vs. DRAGON: 0.51, p < 0.001). Among all selected models, Logistic Regression also had a better performance than the DRAGON score on positive predictive value, sensitivity, and specificity. Compared with the DRAGON score, the multiclass machine learning approach showed better performance on the prediction of the 3-month functional outcome of acute ischemic stroke patients requiring reperfusion therapy.

Prediction of Functional Outcome After Acute Ischemic Stroke: Comparison of the CT-DRAGON Score and a Reduced Features Set

Background and Purpose: The CT-DRAGON score was developed to predict long-term functional outcome after acute stroke in the anterior circulation treated by thrombolysis. Its implementation in clinical practice may be hampered by its plethora of variables. The current study was designed to develop and evaluate an alternative score, as a reduced set of features, derived from the original CT-DRAGON score.

Methods: This single-center retrospective study included 564 patients treated for stroke, in the anterior and the posterior circulation. At 90 days, favorable [modified Rankin Scale score (mRS) of 0–2] and miserable outcome (mRS of 5–6) were predicted by the CT-DRAGON in 427 patients. Bootstrap forests selected the most relevant parameters of the CT-DRAGON, in order to develop a reduced set of features. Discrimination, calibration and misclassification of both models were tested.

Results: The area under the receiver operating characteristic curve (AUROC) for the CT-DRAGON was 0.78 (95% CI 0.74–0.81) for favorable and 0.78 (95% CI 0.72-0.83) for miserable outcome. Misclassification was 29% for favorable and 13.5% for miserable outcome, with a 100% specificity for the latter. National Institutes of Health Stroke Scale (NIHSS), pre-stroke mRS and age were identified as the strongest contributors to favorable and miserable outcome and named the reduced features set. While CT-DRAGON was only available in 323 patients (57%), the reduced features set could be calculated in 515 patients (91%) (p < 0.001). Misclassification was 25.8% for favorable and 14.4% for miserable outcome, with a 97% specificity for miserable outcome. The reduced features set had better discriminative power than CT-DRAGON for both outcomes (both p < 0.005), with an AUROC of 0.82 (95% CI 0.79–0.86) and 0.83 (95% CI 0.77–0.87) for favorable and miserable outcome, respectively.

Conclusions: The CT-DRAGON score revealed acceptable discrimination in our cohort of both anterior and posterior circulation strokes, receiving all treatment modalities. The reduced features set could be measured in a larger cohort and with better discrimination. However, the reduced features set needs further validation in a prospective, multicentre study.

Clinical Trial Registration: http://www.clinicaltrials.gov. Identifiers: NCT03355690, NCT04092543.

Simple prediction scores predict good and devastating outcomes after stroke more accurately than physicians

Introduction

physicians are often asked to prognosticate soon after a patient presents with stroke. This study aimed to compare two outcome prediction scores (Five Simple Variables [FSV] score and the PLAN [Preadmission comorbidities, Level of consciousness, Age, and focal Neurologic deficit]) with informal prediction by physicians.

Methods

demographic and clinical variables were prospectively collected from consecutive patients hospitalised with acute ischaemic or haemorrhagic stroke (2012-13). In-person or telephone follow-up at 6 months established vital and functional status (modified Rankin score [mRS]). Area under the receiver operating curves (AUC) was used to establish prediction score performance.

Results

five hundred and seventy-five patients were included; 46% female, median age 76 years, 88% ischaemic stroke. Six months after stroke, 47% of patients had a good outcome (alive and independent, mRS 0-2) and 26% a devastating outcome (dead or severely dependent, mRS 5-6). The FSV and PLAN scores were superior to physician prediction (AUCs of 0.823-0.863 versus 0.773-0.805, P < 0.0001) for good and devastating outcomes. The FSV score was superior to the PLAN score for predicting good outcomes and vice versa for devastating outcomes (P < 0.001). Outcome prediction was more accurate for those with later presentations (>24 hours from onset).

Conclusion

the FSV and PLAN scores are validated in this population for outcome prediction after both ischaemic and haemorrhagic stroke. The FSV score is the least complex of all developed scores and can assist outcome prediction by physicians.

Simple prediction model for unfavorable outcome in ischemic stroke after intravenous thrombolytic therapy

Objective

We aimed to develop a model to predict unfavorable outcome in patients with acute ischemic stroke treated with intravenous thrombolytic therapy (IVT), based on simple variables present on admission.

Methods

Retrospective analysis of acute ischemic stroke patients treated with IVT in a hospital in Rio de Janeiro. Clinical and radiographic variables were selected for analysis. Multivariate logistic regression was used to develop a predictive model.

Results

We analyzed a total of 82 patients. Median National Institutes of Health Stroke Scale (NIHSS) on admission was 9 (3-22), 40.2% presented with a hyperdense artery sign (HAS), 62% had identifiable early parenchymal changes and 61.6% experienced a favorable outcome. An NIHSS score of > 12 on arrival, age > 70 and the presence of HAS were associated with the outcome, even after correction in a logistic regression model.

Conclusion

An NIHSS > 12 on arrival, presence of HAS and age > 70 years were predictors of unfavorable outcome at three months in patients with acute ischemic stroke treated with IVT.

ASTRAL, DRAGON and SEDAN scores predict stroke outcome more accurately than physicians

BACKGROUND AND PURPOSE

ASTRAL, SEDAN and DRAGON scores are three well-validated scores for stroke outcome prediction. Whether these scores predict stroke outcome more accurately compared with physicians interested in stroke was investigated.

METHODS

Physicians interested in stroke were invited to an online anonymous survey to provide outcome estimates in randomly allocated structured scenarios of recent real-life stroke patients. Their estimates were compared to scores' predictions in the same scenarios. An estimate was considered accurate if it was within 95% confidence intervals of actual outcome.

RESULTS

In all, 244 participants from 32 different countries responded assessing 720 real scenarios and 2636 outcomes. The majority of physicians' estimates were inaccurate (1422/2636, 53.9%). 400 (56.8%) of physicians' estimates about the percentage probability of 3-month modified Rankin score (mRS) > 2 were accurate compared with 609 (86.5%) of ASTRAL score estimates (P < 0.0001). 394 (61.2%) of physicians' estimates about the percentage probability of post-thrombolysis symptomatic intracranial haemorrhage were accurate compared with 583 (90.5%) of SEDAN score estimates (P < 0.0001). 160 (24.8%) of physicians' estimates about post-thrombolysis 3-month percentage probability of mRS 0-2 were accurate compared with 240 (37.3%) DRAGON score estimates (P < 0.0001). 260 (40.4%) of physicians' estimates about the percentage probability of post-thrombolysis mRS 5-6 were accurate compared with 518 (80.4%) DRAGON score estimates (P < 0.0001).

CONCLUSIONS

ASTRAL, DRAGON and SEDAN scores predict outcome of acute ischaemic stroke patients with higher accuracy compared to physicians interested in stroke.

Validation of the DRAGON Score in a Chinese Population to Predict Functional Outcome of Intravenous Thrombolysis-Treated Stroke Patients

BACKGROUND

The DRAGON score predicts functional outcome of ischemic stroke patients treated with intravenous thrombolysis. Our aim was to evaluate its utility in a Chinese stroke population.

METHODS

Patients with acute ischemic stroke treated with intravenous thrombolysis were prospectively registered in the Thrombolysis Implementation and Monitor of acute ischemic Stroke in China. We excluded patients with basilar artery occlusion and missing data, leaving 970 eligible patients. We calculated the DRAGON score, and the clinical outcome was measured by the modified Rankin Scale at 3 months. Model discrimination was quantified by calculating the C statistic. Calibration was assessed using Pearson correlation coefficient.

RESULTS

The C statistic was .73 (.70-.76) for good outcome and .75 (.70-.79) for miserable outcome. Proportions of patients with good outcome were 94%, 83%, 70%, and 0% for 0 to 1, 2, 3, and 8 to 10 score points, respectively. Proportions of patients with miserable outcome were 0%, 3%, 9%, and 50% for 0 to 1, 2, 3, and 8 to 10 points, respectively. There was high correlation between predicted and observed probability of 3-month favorable and miserable outcome in the external validation cohort (Pearson correlation coefficient, .98 and .98, respectively, both P < .0001).

CONCLUSIONS

The DRAGON score showed good performance to predict functional outcome after tissue-type plasminogen activator treatment in the Chinese population. This study demonstrated the accuracy and usability of the DRAGON score in the Chinese population in daily practice.