A clinical prediction model based on interpretable machine learning algorithms for prolonged hospital stay in acute ischemic stroke patients: a real-world study.

Objective: Acute ischemic stroke (AIS) brings an increasingly heavier economic burden nowadays. Prolonged length of stay (LOS) is a vital factor in healthcare expenditures. The aim of this study was to predict prolonged LOS in AIS patients based on an interpretable machine learning algorithm. Methods: We enrolled AIS patients in our hospital from August 2017 to July 2019, and divided them into the "prolonged LOS" group and the "no prolonged LOS" group. Prolonged LOS was defined as hospitalization for more than 7 days. The least absolute shrinkage and selection operator (LASSO) regression was applied to reduce the dimensionality of the data. We compared the predictive capacity of extended LOS in eight different machine learning algorithms. SHapley Additive exPlanations (SHAP) values were used to interpret the outcome, and the most optimal model was assessed by discrimination, calibration, and clinical utility. Results: Prolonged LOS developed in 149 (22.0%) of the 677 eligible patients. In eight machine learning algorithms, prolonged LOS was best predicted by the Gaussian naive Bayes (GNB) model, which had a striking area under the curve (AUC) of 0.878 ± 0.007 in the training set and 0.857 ± 0.039 in the validation set. The variables sorted by the gap values showed that the strongest predictors were pneumonia, dysphagia, thrombectomy, and stroke severity. High net benefits were observed at 0%-76% threshold probabilities, while good agreement was found between the observed and predicted probabilities. Conclusions: The model using the GNB algorithm proved excellent for predicting prolonged LOS in AIS patients. This simple model of prolonged hospitalization could help adjust policies and better utilize resources.

Prognostic significance of blood pressure parameters after mechanical thrombectomy according to collateral status.

BACKGROUND: Mechanical thrombectomy (MT) has been proven as an effective and safe therapy for patients with acute ischemic stroke from large vessel occlusion. However, there is still a controversial topic about post-procedural management including blood pressure (BP). METHODS: A total of 294 patients who received MT in Second Affiliated Hospital of Soochow University from April 2017 to September 2021 were included consecutively. The association of blood pressure parameters (BPV and hypotension time) with poor functional outcome was evaluated using logistic regression models. Meanwhile, the effects of BP parameters on mortality was analyzed using cox proportional hazards regression models. Furthermore, the corresponding multiplicative term was added to the above models to study the interaction between BP parameters and CS. RESULTS: Two hundred ninety four patients were included finally. The mean age was 65.5 years. At the 3-month follow-up, 187(61.5%) had poor functional outcome and 70(23.0%) died. Regardless of the CS, BP CV is positively associated with poor outcome. Hypotension time was negatively associated with poor outcome. We conducted a subgroup analysis according to CS. BPV was significantly associated with mortality at 3-month and displayed a trend toward poor outcome for patients with poor CS only. The interaction between SBP CV and CS with respect to mortality after adjusting for confounding factors was statistically significant (P for interaction = 0.025) and the interaction between MAP CV and CS with respect to mortality after multivariate adjustment was also statistically significant (P for interaction = 0.005). CONCLUSION: In MT-treated stroke patients, higher BPV in the first 72 h is significantly associated with poor functional outcome and mortality at 3-month regardless of CS. This association was also found for hypotension time. Further analysis showed CS modified the association between BPV and clinical prognosis. BPV displayed a trend toward poor outcome for patients with poor CS.

Value of angiographic regional circulation signs in predicting hemorrhagic transformation after endovascular thrombectomy.

OBJECTIVE: The aim of this study was to evaluate the predictive value of the early venous filling (EVF) sign, the basal ganglia blush (BGB) sign and both the EVF and BGB signs for the hemorrhagic transformation (HT) and parenchymal hematoma (PH) in patients after endovascular thrombectomy. METHODS: This study included patients with anterior circulation large vessel occlusive stroke treated with endovascular thrombectomy from May 2017 to December 2021. The predictive value of regional circulation signs for HT and PH were assessed using logistic regression models adjusted for confounders, and further a multiplicative interaction term was added to investigate the effect of different stroke severity on its predictive value. RESULTS: Among the 350 patients included and after adjusting for confounders, those with the EVF sign (adjusted OR=3.934, 95% CI:2.326-6.655), the BGB sign (adjusted OR=3.776, 95% CI:2.341-6.089), and both the EVF and BGB signs (adjusted OR=3.250, 95% CI: 1.886-5.600) were more likely to have HT. The EVF sign (adjusted OR=3.545, 95% CI:2.036-6.170), the BGB sign (adjusted OR=3.742, 95% CI:2.110-6.639), and both the EVF and BGB signs (adjusted OR=3.139, 95% CI: 1.776-5.549) were also significantly correlated with PH. When stratified according to stroke severity, we further found there were significant interactions between regional circulation signs and stroke severity on postoperative HT and PH (all P for interaction < 0.001). CONCLUSIONS: Regional circulation signs were independently associated with HT and PH after endovascular thrombectomy and had a higher predictive value in patients with severe stroke compared with mild to moderate stroke.

Predictors and Dynamic Nomogram to Determine the Individual Risk of Malignant Brain Edema After Endovascular Thrombectomy in Acute Ischemic Stroke.

BACKGROUND AND PURPOSE: This study aimed to construct an optimal dynamic nomogram for predicting malignant brain edema (MBE) in acute ischemic stroke (AIS) patients after endovascular thrombectomy (ET). METHODS: We enrolled AIS patients after ET from May 2017 to April 2021. MBE was defined as a midline shift of >5 mm at the septum pellucidum or pineal gland based on follow-up computed tomography within 5 days after ET. Multivariate logistic regression and LASSO (least absolute shrinkage and selection operator) regression were used to construct the nomogram. The area under the receiver operating characteristic curve (AUC) and decisioncurve analysis were used to compare our nomogram with two previous risk models for predicting brain edema after ET. RESULTS: MBE developed in 72 (21.9%) of the 329 eligible patients. Our dynamic web-based nomogram (https://successful.shinyapps.io/DynNomapp/) consisted of five parameters: basal cistern effacement, postoperative National Institutes of Health Stroke Scale (NIHSS) score, brain atrophy, hypoattenuation area, and stroke etiology. The nomogram showed good discrimination ability, with a C-index (Harrell's concordance index) of 0.925 (95% confidence interval=0.890-0.961), and good calibration (Hosmer-Lemeshow test, p=0.386). All variables had variance inflation factors of <1.5 and tolerances of >0.7, suggesting no significant collinearity among them. The AUC of our nomogram (0.925) was superior to those of Xiang-liang Chen and colleagues (0.843) and Ming-yang Du and colleagues (0.728). CONCLUSIONS: Our web-based dynamic nomogram reliably predicted the risk of MBE in AIS patients after ET, and hence is worthy of further evaluation.

Mechanical thrombectomy versus medical care alone in large ischemic core: An up-to-date meta-analysis.

OBJECTIVE: We compared outcomes and adverse events of thrombectomy versus medical management in acute ischemic stroke (AIS) patients with baseline large infarct core. METHODS: We searched Ovid MEDLINE(R) ALL, Cochrane Library Clinical Controlled Trials and EMBASE from inception to January 2021 for studies comparing thrombectomy and medical management alone in AIS patients who had ASPECTS <=7 or ischemic core volume >=50 ml. Imaging modalities to valuate ASPECTS and core volume were without restriction. The functional outcome was measured by mRS (modified Rankin Scale) score 0-2 at 90 days or discharge. The safety end point included the rates of mortality and sICH (symptomatic intracranial hemorrhage) or PH2 (parenchymal hematoma type 2). RESULTS: Fourteen studies with a total of 2547 patients (thrombectomy n = [1197]; medical care alone [n = 1350]) fulfilled our criteria. As for patients with low ASPECTS, pooled results indicated a higher odds of good functional outcome (OR = 3.47; 95% CI 1.99 to 6.07; P < 0.0001, I2=66%) and a lower risk of mortality (OR = 0.62; 95% CI 0.46 to 0.83; P = 0.001, I2=32%) in thrombectomy group compared with no thrombectomy group, but the risk of sICH or PH2 did not differ between two groups. As for patients with large core volume, both functional outcome and safety end point between two groups showed no statistically significant difference. CONCLUSION: Thrombectomy remained safe and effective by careful selection in patients with low ASPECTS. More studies were warranted to explore contraindications for mechanical thrombectomy in AIS patients, especially in patients with large core volume.

Clinical Significance of Hyperdense Area after Endovascular Therapy in Patients with Acute Ischemic Stroke: A Systematic Review and Meta-Analysis.

OBJECTIVE: We performed a systematic review and meta-analysis to investigate the clinical significance of hyperdense area after thrombectomy in patients with acute ischemic stroke (AIS). METHODS: We searched Ovid MEDLINE(R) and Epub Ahead of Print, In-Process and other Non-Indexed, Cochrane Library Clinical Controlled Trials and Embase from inception to September 2020 and collected the cohort and case-control studies about the clinical significance of hyperdense area on different types of computed tomography (CT) after thrombectomy in patients with AIS. Outcomes were poor functional outcome (modified Rankin Scale [mRS] Score 3-6 at discharge or 90-day), mortality and subtypes of hemorrhage according to the European Cooperative Acute Stroke Study (ECASS). RESULTS: 1,999 patients from 16 studies were included in this meta-analysis. Pooled results indicated higher risk of symptomatic intracerebral hemorrhage (odds ratio [OR] = 3.02; 95% confidence interval [CI] 1.84-4.95; p < 0.0001, I2 = 0%) in patients with hyperdense area, and the subtype of parenchymal hematoma as well. There was also higher odds of poor functional outcome based on the mRS 3-6 at discharge or 90-day (OR = 1.92; 95% CI 1.35-2.73; p = 0.0003, I2 = 31%) and mortality (OR = 2.06; 95% CI 1.41-3.02; p = 0.0002, I2 = 0%) in patients with hyperdense area after thrombectomy compared with those without hyperdense area. CONCLUSIONS: Our results indicated that the presence of hyperdense area on CT after thrombectomy was associated with high risk of symptomatic intracerebral hemorrhage, poor functional outcome, as well as mortality in patients with AIS. However, further studies were needed to confirm these results. The meta-analysis was conducted in adherence with the PRISMA Statement and was registered at the International Prospective Register of Systematic Reviews (CRD42020164165). To the best of our knowledge, this study is the first meta-analysis investigating the effect of hyperdense area after endovascular therapy in patients with AIS.