Development of a Simple Clinical Score to Predict Early Neurological Improvement after Mechanical Thrombectomy: A Single-Centre Cohort Study

Development of a Simple Clinical Score to Predict Early Neurological Improvement after Mechanical Thrombectomy: A Single-Centre Cohort Study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Development of a Simple Clinical Score to Predict Early Neurological Improvement after Mechanical Thrombectomy: A Single-Centre Cohort Study Chenyang Zhao, Xihua Li, Yi Han, Tao Zhou, Xuefei Ren, Yaxuan Sun This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8536054/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 10 You are reading this latest preprint version Abstract Background Early neurological improvement (ENI) after endovascular thrombectomy (EVT) is strongly associated with long-term functional outcome, but simple bedside tools to predict ENI are limited. We aimed to develop and internally evaluate a pragmatic clinical model and simplified risk score for ENI after EVT. Methods We conducted a single-centre retrospective cohort study including consecutive patients with acute ischemic stroke who underwent emergency mechanical thrombectomy at Shanxi Provincial People’s Hospital between January 2020 and December 2022. ENI was defined as a decrease in NIHSS ≥ 8 points or NIHSS ≤ 1 at day 7. Candidate predictors included age, sex, baseline NIHSS, diabetes, cardioembolic etiology, prior cerebrovascular disease and door-to-puncture time (DPT, including DPT > 12 h). Multivariable logistic regression was used to build a full prediction model. A simplified bedside score based on NIHSS strata, diabetes and cardioembolic etiology was derived from model coefficients. Calibration, Brier score, and decision curve analysis (DCA) were assessed, and internal validation was performed using bootstrap resampling (1,000 repetitions) to obtain optimism-corrected performance estimates. Results A total of 185 patients were included, of whom 53 (28.6%) achieved early neurological improvement (ENI). In multivariable logistic regression, baseline NIHSS was the dominant predictor of ENI, whereas the incremental effects of other candidate predictors were modest. The full model (Model 2) showed an apparent AUC of 0.706 and an optimism-corrected AUC of 0.657 after 1,000 bootstrap resamples; the Brier score was 0.181 (apparent) and 0.197 (optimism-corrected). The simplified bedside score demonstrated an apparent and optimism-corrected AUC of 0.677, while the NIHSS-only model yielded an apparent AUC of 0.673 and an optimism-corrected AUC of 0.674. Bootstrap-corrected calibration of the full model suggested some overfitting (intercept − 0.335, slope 0.591), whereas the simplified score showed acceptable calibration (intercept − 0.108, slope 0.893). Decision curve analysis indicated that the full model and simplified score provided net benefit over treat-all and treat-none across clinically relevant threshold probabilities. Conclusions A parsimonious clinical model and a simplified bedside score based on baseline NIHSS category and cardioembolic aetiology provided moderate discrimination for ENI after EVT; external validation is warranted.External validation in larger multicentre cohorts is warranted. stroke thrombectomy early neurological improvement prediction model risk score nomogram Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Endovascular thrombectomy (EVT) has become the standard of care for selected patients with acute ischemic stroke due to large-vessel occlusion, after a series of landmark randomized controlled trials demonstrated substantial improvements in functional outcomes compared with best medical therapy alone.[ 1 – 4 ] These trials have reshaped acute stroke workflows and underscored the importance of rapid reperfusion. Nevertheless, outcomes after EVT remain heterogeneous, and a considerable proportion of patients fail to achieve functional independence despite technically successful recanalization.[ 1 – 4 ] In this context, early neurological improvement (ENI) has emerged as a pragmatic intermediate endpoint that reflects early treatment response and correlates with long-term prognosis. ENI is typically defined according to changes in the National Institutes of Health Stroke Scale (NIHSS), most often as an improvement of ≥ 8–10 points or as achieving a NIHSS score of 0–1 within the first 24 hours or by day 7.[ 5 – 8 ] Several studies have shown that ENI after intravenous thrombolysis or EVT is strongly associated with favourable 90-day modified Rankin Scale (mRS) [ 9 ] scores and reduced mortality, and may serve as a surrogate endpoint for long-term outcomes in clinical research.[ 5 – 8 , 10 ] For example, Wang et al. reported that ENI at day 1 following mechanical thrombectomy for distal medium-vessel occlusion stroke independently predicted good to excellent 3-month outcomes.[ 10 ] Similarly, large registry and cohort studies have highlighted that early neurological trajectories—encompassing both ENI and early neurological deterioration (END)—are key determinants of functional recovery after EVT.[ 11 , 12 ] Collectively, these data underscore the clinical value of identifying, soon after thrombectomy, patients who are likely or unlikely to experience ENI. Despite the growing interest in ENI, evidence on its prediction after EVT remains limited. Most prior work has focused on a restricted set of factors, such as baseline NIHSS, age, infarct core size, collateral status and degree of reperfusion, often examined in heterogeneous populations or within single vascular territories.[ 11 , 12 ] A few groups have proposed multivariable models or nomograms for ENI or closely related early outcomes. Zhang et al. developed a nomogram to predict ENI in ischemic stroke patients treated with EVT, identifying baseline NIHSS and imaging markers as key predictors.[ 13 ] Wu et al. later constructed a nomogram to jointly predict 3-month unfavourable outcomes and ENI, further emphasizing the prognostic relevance of early neurological response.[ 14 ] More recently, machine learning–based nomograms have been reported for ENI after intravenous thrombolysis.[ 11 , 15 ] However, many of these models are relatively complex, are not easily translated into simple bedside tools, and often lack comprehensive evaluation of calibration and clinical utility. In parallel, there has been increasing attention on END and unexplained END after EVT, which are consistently associated with poor outcomes.[ 11 , 12 , 16 , 17 ] Several nomograms have been proposed to predict END after mechanical thrombectomy, mainly in anterior circulation large-vessel occlusion.[ 17 , 18 ] These studies highlight that early post-procedural neurological change—whether improvement or deterioration—reflects a complex interplay among baseline clinical status, vascular pathology and reperfusion quality. Nevertheless, ENI-specific prediction models after EVT remain relatively scarce, and even fewer have been developed and evaluated in contemporary real-world cohorts from Asian populations, where stroke etiologies, risk factor profiles and treatment workflows may differ from those in Western trials.[ 11 , 13 – 15 ] Time metrics from hospital arrival to reperfusion represent another crucial dimension that may influence ENI. Workflow analyses from EVT trials and registries have shown that shorter door-to-puncture (DPT) or door-to-arterial puncture times are associated with higher rates of functional independence and better quality-of-life outcomes.[ 19 – 22 ] Quality improvement initiatives targeting DPT < 60 minutes have demonstrated that systematic process optimization can translate into improved clinical results.[ 19 , 21 , 22 ] However, the specific relationship between DPT and ENI after EVT is less well characterized, particularly in the ultra-late window (> 12 hours from last known well), where patient selection and tissue viability are more heterogeneous. Existing data suggest that the impact of in-hospital delays may not be strictly linear across the entire time range, raising the possibility that non-linear modelling approaches, such as restricted cubic splines, could provide additional insight.[ 11 , 19 – 22 ] From a methodological perspective, contemporary standards for clinical prediction research emphasize not only discrimination but also calibration and clinical usefulness. Decision curve analysis (DCA) provides a formal framework for quantifying the net benefit of prediction models across a range of threshold probabilities, and for comparing them with default “treat-all” and “treat-none” strategies.[ 23 , 24 ] However, DCA has rarely been applied to ENI prediction after EVT, and few studies have integrated traditional regression modelling, simplified bedside scoring systems, nomograms and DCA within a single coherent framework. Against this background, we sought to address several knowledge gaps using a real-world cohort of patients undergoing EVT in a high-volume centre. Our primary objective was to develop and internally evaluate a multivariable logistic regression model for predicting ENI after thrombectomy based on routinely available clinical variables. Building on this model, we aimed to derive a simplified bedside risk score and a corresponding nomogram to facilitate rapid risk stratification in everyday practice. A key secondary objective was to explore the independent and potentially non-linear association between DPT and ENI, with particular attention to patients treated in an ultra-late time window. To provide a comprehensive assessment, we examined model discrimination, Brier score, calibration and net benefit using receiver operating characteristic (ROC) analysis, calibration plots and DCA. Our study therefore seeks to deliver a pragmatic, clinically interpretable tool for early risk assessment after EVT, while also clarifying the extent to which in-hospital treatment delays influence early neurological recovery. Methods 3.1 Study design and population For the purposes of this study, the data were accessed for research purposes on 01/12/2025. We conducted a single-centre retrospective cohort study at Shanxi Provincial People’s Hospital, including consecutive patients who underwent emergency endovascular thrombectomy (EVT) for acute ischaemic stroke between 01/01/2020 and 31/12/2022. Consecutive adult patients with acute ischemic stroke who underwent emergency mechanical thrombectomy (endovascular thrombectomy, EVT) during this period were screened for eligibility. Inclusion criteria were: (1) age ≥ 18 years; (2) acute ischemic stroke treated with EVT involving the anterior and/or posterior circulation; and (3) availability of both a baseline National Institutes of Health Stroke Scale (NIHSS) score on admission and a follow-up NIHSS assessment at approximately 7 days after EVT (or at discharge, if earlier). Exclusion criteria were: (1) missing data on any component required to define early neurological improvement (ENI); (2) missing key clinical covariates required for model development (e.g. baseline NIHSS, history of diabetes or cardioembolic disease, door-to-puncture time); and (3) loss to in-hospital follow-up before completion of the 7-day neurological assessment. After application of these criteria, 185 patients were included in the final analysis; among them, 53 (28.6%) achieved ENI and 132 did not, as shown in the patient selection flow chart (Fig. 1 ). The study protocol was approved by the Ethics Committee of Shanxi Provincial People’s Hospital (approval No. 891). Owing to the retrospective design and use of de-identified data, the requirement for written informed consent was waived. All procedures complied with national regulations and the Declaration of Helsinki. Data for this retrospective analysis were extracted from the hospital electronic medical record system for patients treated between 01/01/2020 and 31/12/2022.For the purposes of this study, all data used for analysis were accessed from the hospital’s electronic medical record system between 01/01/2020 and 31/12/2022. 3.2 Endpoints and definitions The primary endpoint was early neurological improvement (ENI) after EVT. ENI was defined a priori as either a reduction in NIHSS score of ≥ 8 points from baseline to day 7, or an absolute NIHSS score ≤ 1 at day 7. Patients who died before the day-7 assessment were classified as not having ENI. Baseline NIHSS was defined as the first NIHSS score recorded on admission before EVT.The National Institutes of Health Stroke Scale (NIHSS) is an established and previously published measure of neurological deficit. [ 25 ] Hypertension was defined as a documented history of hypertension, current use of antihypertensive medication, or blood pressure ≥ 140/90 mmHg on repeated measurements during admission. Diabetes mellitus was defined as a previous diagnosis of diabetes, use of glucose-lowering therapy, or HbA1c ≥ 6.5% when available. Cardioembolic aetiology was considered present when atrial fibrillation or another major cardiac source of embolism (e.g. recent myocardial infarction, left ventricular thrombus, prosthetic valve) was documented by the treating team on the basis of standard investigations. Prior cerebrovascular disease (CVD) was defined as any history of ischaemic stroke, intracerebral haemorrhage or transient ischaemic attack. Door-to-puncture time (DPT) was defined as the interval, in hours, between hospital arrival (“door”) and arterial puncture for EVT. For secondary analyses, we additionally created a dichotomous variable (DPT > 12 h vs ≤ 12 h) to reflect ultra-late treatment. Where used, wake-up stroke was defined as stroke symptoms first recognised on awakening from sleep, with the exact time of onset unknown. 3.3 Candidate predictors Candidate predictors for ENI were selected a priori based on clinical relevance and previous literature on early neurological trajectories after reperfusion therapy. These included: Demographics: age, sex; Clinical history hypertension, diabetes, cardioembolic etiology (primarily atrial fibrillation), prior cerebrovascular disease; Baseline status baseline NIHSS score, vascular territory (anterior vs posterior circulation); Time metrics door-to-puncture time (continuous, per hour). All candidate variables were first entered into univariable logistic regression models with ENI (yes/no) as the dependent variable. Baseline characteristics of patients with and without ENI are summarized in Table 1 . Table 1 Baseline clinical characteristics according to early neurological improvement status. Values are presented as mean ± standard deviation, median (interquartile range), or number (percentage), as appropriate. ENI indicates early neurological improvement. Analytic cohort includes 185 patients (ENI, n = 53; no ENI, n = 132). Variable All patients (n = 185) ENI (n = 53) No ENI (n = 132) p-value Age, years 61.9 ± 13.2 62.3 ± 13.8 61.7 ± 13.0 0.785 Male sex, n (%) 125 (67.6%) 30 (56.6%) 95 (72.0%) 0.065 Pre-stroke mRS 0.0 (0.0–0.0) 0.0 (0.0–0.0) 0.0 (0.0–0.0) 0.382 Baseline NIHSS 14.0 (11.0–19.0) 18.0 (12.0–20.0) 13.0 (11.0–17.0) < 0.001 Wake-up stroke, n (%) 48 (25.9%) 13 (24.5%) 35 (26.5%) 0.926 Hypertension, n (%) 101 (54.6%) 32 (60.4%) 69 (52.3%) 0.402 Diabetes, n (%) 27 (14.6%) 9 (17.0%) 18 (13.6%) 0.725 Prior cerebrovascular disease, n (%) 34 (18.4%) 10 (18.9%) 24 (18.2%) 1.000 Cardioembolic aetiology, n (%) 64 (34.6%) 22 (41.5%) 42 (31.8%) 0.279 Admission glucose, mmol/L 6.9 (6.0–8.4) 7.0 (6.2–8.3) 6.9 (6.0–8.4) 0.685 LDL cholesterol, mmol/L 2.7 (2.1–3.3) 2.6 (2.0–3.3) 2.8 (2.2–3.3) 0.389 HDL cholesterol, mmol/L 1.1 (0.9–1.3) 1.1 (0.9–1.2) 1.1 (0.9–1.3) 0.302 Heart rate on admission, bpm 83.2 ± 21.4 80.2 ± 18.6 84.4 ± 22.4 0.191 Door-to-puncture time, h 8.0 (5.0–11.0) 7.0 (4.5–10.0) 8.0 (5.0–11.0) 0.166 3.4 Statistical analysis 3.4.1 Descriptive and univariable analyses Continuous variables were inspected for distribution and are presented as mean ± standard deviation (SD) or median with interquartile range (IQR), as appropriate. Categorical variables are summarised as counts and percentages. Baseline characteristics were compared between patients with ENI and those without ENI using the Student’s t-test or the Mann–Whitney U test for continuous variables, and the χ² test or Fisher’s exact test for categorical variables, as appropriate. Univariable associations between each candidate predictor and ENI were evaluated using logistic regression. Results are reported as crude odds ratios (ORs) with 95% confidence intervals (CIs) and corresponding p-values (Table 2 ). Table 2 Univariable associations between candidate predictors and early neurological improvement. Crude odds ratios (ORs) and 95% confidence intervals (CIs) from univariable logistic regression models with early neurological improvement (ENI) as the dependent variable. ENI indicates early neurological improvement; DPT, door-to-puncture time. Predictor Crude OR (95% CI) p-value Age (per 10-year increase) 1.04 (0.81–1.32) 0.778 Male sex 0.51 (0.26–0.99) 0.045 Baseline NIHSS (per 1-point increase) 1.13 (1.06–1.21) < 0.001 Diabetes 1.30 (0.54–3.10) 0.561 Cardioembolic disease 1.52 (0.79–2.94) 0.212 Prior cerebrovascular disease 1.05 (0.46–2.37) 0.913 Hypertension 1.39 (0.73–2.66) 0.318 Wake-up stroke 0.90 (0.43–1.88) 0.780 DPT (per 1-hour increase) 0.93 (0.85–1.02) 0.125 DPT > 12 h vs ≤ 12 h 0.41 (0.11–1.45) 0.165 3.4.2 Multivariable model development We then fitted multivariable logistic regression models with ENI as the dependent variable. To preserve an adequate number of events per variable and reduce the risk of overfitting, we pre-specified a parsimonious set of clinically important predictors rather than relying solely on data-driven variable selection. The core model (Model 1) included age (per 10-year increase), sex (male vs female), baseline NIHSS (per 1-point increase), diabetes (yes/no), cardioembolic aetiology (yes/no) and prior cerebrovascular disease (yes/no). A second model (Model 2) additionally incorporated DPT > 12 h (yes/no) to assess the incremental contribution of ultra-late treatment beyond the clinical core model. For each model, adjusted odds ratios (ORs) with 95% confidence intervals (CIs) and p-values were reported. Multicollinearity was evaluated using variance inflation factors, and overall model fit was examined using standard goodness-of-fit statistics. The multivariable results are summarised in Table 3 . Table 3 Multivariable logistic regression models for early neurological improvement (Model 1: clinical core; Model 2: core + DPT > 12 h). Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) from multivariable logistic regression models. Model 1 includes age, sex, baseline NIHSS, diabetes, cardioembolic disease, and prior cerebrovascular disease. Model 2 additionally includes door-to-puncture time (DPT) > 12 hours. “—” indicates that the predictor was not included in the corresponding model. ENI indicates early neurological improvement; OR, odds ratio; CI, confidence interval; DPT, door-to-puncture time. Predictor Adjusted OR (95% CI) – Model 1 p-value Adjusted OR (95% CI) – Model 2 p-value Age (per 10-year increase) 0.89 (0.68–1.16) 0.394 0.90 (0.68–1.18) 0.428 Male sex 0.52 (0.25–1.07) 0.077 0.51 (0.25–1.06) 0.072 Baseline NIHSS (per 1-point increase) 1.13 (1.05–1.21) < 0.001 1.13 (1.05–1.21) < 0.001 Diabetes 1.39 (0.54–3.55) 0.494 1.36 (0.53–3.49) 0.528 Cardioembolic disease 1.32 (0.63–2.75) 0.457 1.21 (0.57–2.55) 0.620 Prior cerebrovascular disease 1.02 (0.43–2.43) 0.960 1.08 (0.45–2.58) 0.863 DPT > 12 h - - 0.44 (0.12–1.66) 0.226 3.4.3 Simplified bedside score and nomogram To enhance clinical usability, we derived a simplified bedside risk score consistent with Table 4 . Baseline NIHSS was categorised into three strata (≤ 10, 11–15, and ≥ 16) and assigned 0, 2, and 4 points, respectively. Cardioembolic aetiology contributed 1 point when present (0 when absent). The total score therefore ranged from 0 to 5 points and was further grouped into low (0–2), intermediate (3–4), and high (5) strata. The simplified score was then entered as a continuous predictor in a logistic regression model to estimate the predicted probability of ENI. In addition, a nomogram was constructed from the full multivariable model to provide an intuitive graphical tool for bedside estimation of ENI risk (Fig. 4 ). Table 4 Simplified bedside risk score for early neurological improvement. Simplified point-based score derived from the multivariable model. Higher total scores indicate a higher predicted probability of early neurological improvement (ENI). The score uses only baseline NIHSS category and the presence of cardioembolic disease to facilitate bedside use. Predictor Category Points Baseline NIHSS ≤ 10 0 11–15 2 ≥ 16 4 Cardioembolic disease Absent 0 Present 1 Total score ranges and observed ENI rates in the derivation cohort Total score range Risk category Observed ENI rate 0–2 Low 15/92 (16.3%) 3–4 Intermediate 22/63 (34.9%) 5 High 16/30 (53.3%) 3.4.4 Model performance and internal validation Model discrimination was quantified using the area under the receiver operating characteristic curve (AUC) with 95% confidence intervals (CIs). Overall prediction error was assessed using the Brier score, a proper scoring rule for probabilistic predictions. [ 26 , 27 ] Calibration was evaluated by plotting observed versus predicted probabilities of ENI across deciles of predicted risk and by estimating the calibration intercept and slope from a logistic regression of the observed outcome on the logit of the predicted risk. We compared the performance of three models: (1) an NIHSS-only model; (2) the full multivariable model (Model 2); and (3) the simplified bedside score. Receiver operating characteristic curves for these models are shown in Fig. 2 . Calibration plots for the full model and the simplified score are presented in Fig. 3 . Apparent performance metrics (AUC, Brier score, calibration intercept and slope) are summarised in Supplementary Table S2. Internal validation was performed using bootstrap resampling (1,000 repetitions) to quantify optimism and obtain optimism-corrected estimates of model performance, including AUC, Brier score, calibration intercept and calibration slope. External validation was not performed. Internal validation was performed using bootstrap resampling (1,000 repetitions) to estimate optimism and to obtain optimism-corrected AUC, Brier score, calibration intercept, and calibration slope. Both apparent and optimism-corrected performance metrics are reported in Supplementary Table S2. External validation was not performed. 3.4.5 Additional analyses To explore potential non-linear associations between DPT and ENI, we fitted logistic regression models with DPT entered as a continuous predictor using restricted cubic splines with pre-specified degrees of freedom. Odds ratios for ENI across the observed DPT range were plotted using 6 hours as the reference value, and 95% confidence intervals (CIs) were displayed to illustrate estimation uncertainty, particularly in the ultra-late window (> 12 h) (Figure S2). Sensitivity analyses comprised: (1) restricting the cohort to anterior circulation strokes; and (2) applying an alternative ENI definition based solely on a ≥ 8-point NIHSS reduction, without the additional absolute NIHSS ≤ 1 criterion. For each sensitivity scenario, the multivariable model was refitted and the key coefficients were compared with those from the primary analysis (Table S1). We also generated a forest plot of adjusted odds ratios from the main multivariable model to visually summarise the relative contribution of each predictor (Figure S1). Finally, decision curve analysis was performed to compare the net benefit of the full model, the simplified bedside score, a “treat-all” strategy and a “treat-none” strategy across a range of clinically relevant threshold probabilities (Figure S3). All statistical analyses were conducted using R (R Foundation for Statistical Computing, Vienna, Austria) and Python (Python Software Foundation), with packages including statsmodels, scikit-learn and rms. A two-sided p-value < 0.05 was considered statistically significant. Results 4.1 Study population and baseline characteristics Between January 2020 and December 2022, a total of 185 patients met the inclusion criteria and were included in the final analysis. Among them, 53 patients (28.6%) achieved early neurological improvement (ENI) and 132 did not. The detailed flow of patient selection is shown in Fig. 1 . Baseline clinical and imaging characteristics according to ENI status are summarised in Table 1 . Patients with ENI generally presented with more severe neurological deficits at baseline, as reflected by higher NIHSS scores, compared with those without ENI. The distributions of age and vascular territory (anterior vs posterior circulation) were broadly similar between groups. The prevalences of hypertension, diabetes and prior cerebrovascular disease were also comparable, whereas cardioembolic aetiology appeared somewhat more frequent among patients who achieved ENI, although this trend did not clearly reach statistical significance. At the descriptive level, door-to-puncture time did not differ appreciably between ENI and non-ENI groups. 4.2 Univariable associations In univariable logistic regression analyses (Table 2 ), higher baseline NIHSS was strongly associated with a greater likelihood of ENI (crude OR per 1-point increase 1.13; 95% CI 1.06–1.21; p < 0.001). Male sex was associated with a lower probability of ENI (OR 0.51; 95% CI 0.26–0.99; p = 0.045). Age (per 10-year increase) showed no significant association with ENI. Diabetes, cardioembolic aetiology and prior cerebrovascular disease all had crude ORs > 1, indicating numerically higher ENI rates among patients with these conditions; however, none of these associations reached statistical significance in univariable analyses (all p > 0.20). In contrast, a door-to-puncture time > 12 h tended to be associated with lower odds of ENI (OR 0.41; 95% CI 0.11–1.45; p = 0.17), but this finding was not statistically significant and is compatible with only a modest and imprecisely estimated effect. 4.3 Multivariable prediction model In the multivariable logistic regression model including age (per 10-year increase), sex, baseline NIHSS, diabetes, cardioembolic aetiology, prior cerebrovascular disease and DPT > 12 h (Model 2), baseline NIHSS remained the only independent predictor of ENI with a statistically robust effect. Each 1-point increase in baseline NIHSS was associated with approximately 13% higher odds of ENI (adjusted OR 1.13; 95% CI 1.05–1.21; p = 0.001). Male sex showed a tendency towards reduced ENI (adjusted OR 0.51; 95% CI 0.25–1.06; p = 0.07), although this did not reach conventional statistical significance after adjustment. Diabetes (adjusted OR 1.36; 95% CI 0.53–3.49; p = 0.53), cardioembolic aetiology (adjusted OR 1.21; 95% CI 0.57–2.55; p = 0.62) and prior cerebrovascular disease (adjusted OR 1.08; 95% CI 0.45–2.58; p = 0.86) were not independently associated with ENI. DPT > 12 h was associated with lower odds of ENI (adjusted OR 0.44; 95% CI 0.12–1.66; p = 0.23), but the wide confidence interval was compatible with no clear effect. The overall pattern of adjusted associations is illustrated in the forest plot (Supplementary Figure S1), which visually highlights the dominant contribution of baseline NIHSS and the comparatively modest and imprecise effects of the other variables. 4.4 Simplified bedside score and nomogram On the basis of the multivariable model, a simplified bedside risk score was constructed using baseline NIHSS category and cardioembolic aetiology (Table 4 ). Baseline NIHSS was categorised as ≤ 10, 11–15, and ≥ 16, assigned 0, 2, and 4 points, respectively, and cardioembolic aetiology contributed 1 point when present. The resulting total score ranged from 0 to 5. When grouped by total score, ENI rates increased across strata: 15/92 (16.3%) in the low-score group (0–2), 22/63 (34.9%) in the intermediate group (3–4), and 16/30 (53.3%) in the high-score group (5). These findings support pragmatic bedside risk stratification, while external validation is warranted. The corresponding nomogram derived from the full multivariable model (Fig. 4 ) provides a graphical representation of the contribution of each predictor and allows clinicians to estimate an individual patient’s ENI probability by summing points across variables. 4.5 Model performance The full multivariable model (Model 2) demonstrated moderate discrimination, with an AUC of 0.71 and a Brier score of 0.18 in the derivation cohort. An NIHSS-only model achieved an AUC of 0.67 (Brier score 0.19), indicating that baseline NIHSS alone already captured a substantial proportion of the predictive information. The simplified bedside score model yielded an AUC of 0.66 and a Brier score of 0.19, retaining most of the discriminative ability of the full model while relying on a more compact set of predictors. Bootstrap internal validation (1,000 resamples) was performed to estimate optimism and to obtain optimism-corrected performance. The full model (Model 2) achieved an apparent AUC of 0.706 and an optimism-corrected AUC of 0.657, with a corresponding Brier score of 0.181 (apparent) and 0.197 (optimism-corrected). Optimism-corrected calibration for the full model showed an intercept of − 0.335 and a slope of 0.591, suggesting some overfitting. The simplified score showed an apparent and optimism-corrected AUC of 0.677, while the NIHSS-only model showed an apparent AUC of 0.673 and an optimism-corrected AUC of 0.674. Apparent and optimism-corrected performance metrics for all models are summarised in Supplementary Table S2. In decision curve analysis (Supplementary Figure S3), both the full multivariable model and the simplified bedside score provided greater net benefit than the default “treat-all” or “treat-none” strategies across a clinically relevant range of threshold probabilities (approximately 0.20–0.50). The full model offered only modest incremental net benefit over the simplified score within this range, suggesting that the simpler tool may be sufficient for routine clinical use in many settings. 4.6 Additional and sensitivity analyses Restricted cubic spline analysis of door-to-puncture time showed no strong overall association between DPT and ENI across the observed time range (Supplementary Figure S2). Odds ratios for ENI remained close to 1 for most DPT values, and estimates beyond 12 hours were accompanied by wide confidence intervals, reflecting the small number of patients treated in this ultra-late window and substantial uncertainty regarding the true effect. In sensitivity analyses (Supplementary Table S1), restricting the cohort to anterior circulation strokes and applying an alternative ENI definition based solely on a ≥ 8-point NIHSS reduction yielded results broadly consistent with the main analysis. Baseline NIHSS remained the principal independent predictor of ENI, and effect estimates for other covariates, including DPT > 12 h, were directionally similar but imprecise. None of the sensitivity analyses materially altered the overall conclusions of the study. Discussion 5.1 Main findings In this single-centre, real-world EVT cohort, approximately one third of patients achieved early neurological improvement (ENI), and ENI once again emerged as an early marker closely linked to downstream functional outcomes, consistent with previous reports. [ 7 , 10 ] ENI was strongly and independently associated with baseline NIHSS, whereas other routinely collected clinical variables—diabetes, cardioembolic aetiology, prior cerebrovascular disease and door-to-puncture time (DPT)—showed only modest and statistically imprecise associations. We developed a multivariable logistic regression model that demonstrated moderate discrimination (AUC ≈ 0.71) and acceptable calibration, and translated this model into a simplified bedside score based on baseline NIHSS category and cardioembolic aetiology, which enabled pragmatic risk stratification with observed ENI rates of 16.3% (low), 34.9% (intermediate), and 53.3% (high) in the derivation cohort.Decision curve analysis indicated that both the full model and the simplified score provided greater net benefit than “treat-all” or “treat-none” strategies across a clinically relevant range of threshold probabilities, with only a modest incremental gain for the full model.[ 23 , 24 , 28 , 29 ] Restricted cubic spline analysis suggested that DPT—including the ultra-late window beyond 12 hours—had at most a limited and imprecisely estimated independent effect on ENI once baseline NIHSS and other factors were accounted for, which is consistent with the notion that ENI is driven primarily by underlying tissue status rather than by relatively small differences in in-hospital time metrics.[ 2 , 10 , 18 , 30 ] 5.2 Comparison with previous studies Our results are consistent with previous studies showing that ENI, or related early neurological trajectories, are strong surrogates for 90-day functional outcome and mortality after reperfusion therapies.[ 2 , 7 , 10 ] For example, Kobeissi et al. and others reported that broadly defined ENI after EVT is associated with higher rates of functional independence and lower rates of death and symptomatic intracranial haemorrhage.[ 7 ] Desai et al. and Katano et al. further demonstrated that very early or 24-hour NIHSS thresholds are powerful predictors of long-term outcome, underscoring the clinical relevance of early neurological trajectories.[ 2 ] Several groups have proposed ENI prediction tools in EVT-treated populations. Zhang et al. developed a multicentre nomogram incorporating age, glucose, ASPECTS, recanalisation status and sICH, achieving AUCs of approximately 0.80 in the derivation cohort and 0.75 in external validation.[ 31 ] Li et al. reported that CTP-ASPECTS may outperform non-contrast CT in predicting ENI after successful reperfusion, highlighting the added value of advanced imaging. [ 6 ] More recently, Lai et al. investigated “ultra-early” neurological improvement immediately after reperfusion and again found that baseline NIHSS and imaging markers are key determinants of early trajectories.[ 6 ] Wu and colleagues integrated ENI and 3-month outcomes into a broader prognostic framework using nomograms.[ 8 ] In parallel, a substantial body of literature has focused on early neurological deterioration (END) after EVT. Nomograms and machine-learning models for END typically identify baseline severity, infarct burden and procedure-related factors as major predictors. [ 11 , 13 , 32 ] Dai et al. specifically examined unexplained END in posterior circulation disease, further highlighting the complex interplay between baseline status, vascular territory and procedural variables. [ 11 ] Our study complements this work by focusing on ENI rather than END and by providing a simple clinical score to identify patients most likely to improve, which may be particularly useful for clinical counselling and for enriching clinical trials with high-benefit candidates. With respect to workflow metrics, multiple trials and registries have demonstrated that shorter onset-to-reperfusion or CT-to-reperfusion times are associated with better functional outcomes after EVT.[ 2 , 18 , 30 ] The ESCAPE trial showed that every 30-minute delay in CT-to-reperfusion time reduces the probability of functional independence,[ 30 ] while more recent work by Joundi et al. linked faster DPT with better health-related quality of life.[ 18 ] Our analysis differs in that we specifically examined ENI as the outcome in a modestly sized, single-centre cohort in which most patients were treated within institutional time targets. In this context, DPT did not exhibit a strong independent association with ENI after adjustment, and restricted cubic spline analysis revealed wide confidence intervals in the ultra-late window because of the small number of patients. Therefore, our findings should not be interpreted as contradicting the “time is brain” paradigm, but rather as reflecting limited statistical power and the possibility that ENI depends more on tissue-level factors (e.g. infarct core, collateral status, infarct growth rate) than on relatively small in-hospital time variations.[ 6 , 10 , 33 ] Finally, compared with our earlier single-centre work in which admission NIHSS and diabetes were identified as independent predictors of in-hospital neurological improvement, the present analysis represents a methodological and translational extension. We move from a two-factor association model to a more comprehensive prediction framework: we report discrimination, Brier score and calibration; derive a simplified bedside score and nomogram; and formally evaluate net clinical benefit using decision curve analysis, in line with contemporary recommendations for prediction model development and reporting. [ 16 , 23 , 24 , 28 , 29 , 34 , 35 ] 5.3 Clinical implications From a practical perspective, our findings suggest that a simple three-component bedside score based on NIHSS strata, diabetes and cardioembolic aetiology can provide an easily interpretable estimate of ENI probability after EVT. Patients with low scores (0–2 points) have roughly a one-in-five chance of ENI, whereas those with scores of 3–4 points approach a one-in-two probability. Very high scores appear to identify patients with a high likelihood of early improvement, although this estimate is derived from a small number of individuals and should be interpreted cautiously. Such risk stratification may help clinicians and families form more realistic expectations in the first days after thrombectomy, before 90-day outcomes are known, and can be integrated into post-procedural workflows to support early risk assessment, ICU and ward resource allocation, and shared decision-making. For example, patients with a low predicted probability of ENI may be flagged for closer monitoring of complications and more intensive supportive care, whereas those with higher predicted probability could be prioritised for early rehabilitation planning and discharge optimisation. In addition, by providing a quantitative and reproducible estimate of early improvement that does not rely on advanced imaging or specialised software, the score and its corresponding nomogram could be used as stratification or enrichment tools in clinical trials of neuroprotective agents, blood pressure management strategies or post-EVT care bundles, helping to target interventions to patients most likely to demonstrate measurable early benefit. [ 2 , 6 , 7 , 10 , 31 ] At the same time, the only moderate AUC and modest net-benefit gain over a NIHSS-only strategy indicate that this tool should be regarded as an adjunct rather than a gatekeeper and should not be used to deny reperfusion therapy or to justify withholding guideline-based standard care. 5.4 Strengths and limitations This study has several strengths. First, it is based on a consecutive, real-world EVT cohort from Shanxi Provincial People’s Hospital, a high-volume comprehensive stroke centre in northern China, which enhances the clinical relevance and applicability of the findings. Second, we adhered to key principles of prediction model development and reporting, including the use of multiple performance metrics (AUC, Brier score and calibration indices) and decision curve analysis to quantify clinical utility, in line with TRIPOD and related guidance. Third, we explicitly translated the statistical model into a simplified bedside score and nomogram, thereby moving beyond pure association towards a pragmatic tool that can be implemented without advanced imaging or specialised software. Several important limitations must also be acknowledged. The single-centre retrospective design raises concerns about selection bias and limits the generalisability of the results. The sample size is modest, yielding an events-per-variable ratio close to the lower recommended boundary; as a consequence, some covariate effects—particularly those related to DPT > 12 h—are estimated imprecisely. Advanced imaging predictors such as infarct core volume, infarct growth rate, collateral scores or CTP-ASPECTS, which have been shown to be relevant for ENI in other studies, were not systematically available in this cohort. Although bootstrap internal validation was used to quantify optimism and we report optimism-corrected discrimination and calibration, external validation in independent multicentre cohorts is still required before routine clinical implementation.Finally, our primary endpoint was ENI rather than the 90-day modified Rankin Scale; although ENI is a validated surrogate marker, it does not fully capture delayed complications, the impact of rehabilitation or long-term quality of life. Future work should therefore prioritise external validation and potential updating of this scoring system in larger multicentre cohorts, incorporate advanced imaging variables where available, and assess whether embedding the score and nomogram into clinical workflows can improve both ENI and longer-term functional outcomes in prospective impact studies.[ 2 , 18 , 30 , 33 ] Conclusions In this single-centre, real-world EVT cohort, we developed and evaluated an ENI prediction model based entirely on simple, routinely available clinical variables. Baseline NIHSS emerged as the dominant independent predictor of early neurological improvement, and a parsimonious multivariable logistic regression model demonstrated moderate discrimination with acceptable calibration. From this model, we derived a three-component simplified bedside score (NIHSS strata, diabetes and cardioembolic aetiology) and a corresponding nomogram, both of which provided intuitive risk stratification and yielded meaningful net clinical benefit over “treat-all” and “treat-none” strategies across a clinically relevant range of threshold probabilities. These findings indicate that the proposed model and simplified score may serve as practical tools for early post-thrombectomy risk stratification, supporting clinical decision-making, resource allocation and communication with patients and their families. However, the modest sample size, single-centre retrospective design, use of ENI rather than long-term functional outcome, and absence of external validation mean that the results should be interpreted with caution. Before routine clinical implementation, the score and nomogram should undergo independent validation and, if necessary, updating in larger multicentre cohorts, and should ideally be tested for their impact on both ENI and 90-day functional outcomes. Declarations Ethical Approval This study was approved by the Ethics Committee of Shanxi Provincial People's Hospital (approval No. 891). Owing to the retrospective design and use of de-identified clinical data, the committee waived the requirement for written informed consent. Competing Interests The authors declare that there are no potential conflicts of interest related to the publication of this article. Author Contribution CZ and YS conceived and designed the study. CZ, XL, YH, TZ, and XR were responsible for patient screening and data collection. CZ performed the statistical analyses and drafted the manuscript. XL and YH contributed to data interpretation and critical revision of the manuscript. TZ and XR contributed to data curation/verification and helped revise the manuscript. YS supervised the study and critically revised the manuscript for important intellectual content. All authors read and approved the final manuscript. Data Availability The de-identified patient-level dataset contains sensitive clinical information and cannot be publicly shared under institutional and legal requirements. Data may be made available from the Ethics Committee of Shanxi Provincial People’s Hospital (approval No. 891) upon reasonable request and with appropriate approvals and a data-use agreement. References Berkhemer OA, et al. A randomized trial of intraarterial treatment for acute ischemic stroke. N Engl J Med. 2015;372(1):11–20. Goyal M, et al. Randomized assessment of rapid endovascular treatment of ischemic stroke. N Engl J Med. 2015;372(11):1019–30. Saver JL, et al. Stent-retriever thrombectomy after intravenous t-PA vs. t-PA alone in stroke. N Engl J Med. 2015;372(24):2285–95. Jovin TG, et al. Thrombectomy within 8 hours after symptom onset in ischemic stroke. N Engl J Med. 2015;372(24):2296–306. Ong CT, et al. Early neurological improvement after intravenous tissue plasminogen activator infusion in patients with ischemic stroke aged 80 years or older. J Chin Med Assoc. 2014;77(4):179–83. Kurmann CC, et al. Association of the 24-Hour National Institutes of Health Stroke Scale After Mechanical Thrombectomy With Early and Long‐Term Survival. Volume 2. Stroke: Vascular and Interventional Neurology; 2022. p. e000244. 4. Kobeissi H, et al. Early neurological improvement as a predictor of outcomes after endovascular thrombectomy for stroke: a systematic review and meta-analysis. J Neurointerv Surg. 2023;15(6):547–51. Lai Y, et al. Identifying the predictors of ultra early neurological improvement and its role in functional outcome after endovascular thrombectomy in acute ischemic stroke. Front Neurol. 2025;16:1492013. van Swieten JC, et al. Interobserver agreement for the assessment of handicap in stroke patients. Stroke. 1988;19(5):604–7. Wang M, et al. Early Neurological Improvement Predicts Clinical Outcome After Thrombectomy for Distal Medium Vessel Occlusions. Front Neurol. 2022;13:809066. Huang Z et al. Infarct Growth Rate Predicts Early Neurological Improvement in Ischemic Stroke After Endovascular Thrombectomy. Brain Sci, 2025. 15(3). Dai Z, et al. Predictors of unexplained early neurological deterioration after thrombectomy for posterior circulation infarction: a reanalysis of the BASILAR study. J Neurosurg. 2024;141(6):1705–11. Zhang X, et al. Nomogram predicting early neurological improvement in ischaemic stroke patients treated with endovascular thrombectomy. Eur J Neurol. 2021;28(1):152–60. Wu Y, et al. Nomogram-Based Prediction of 3-Month Unfavorable Outcome and Early Neurological Deterioration After Endovascular Thrombectomy in Acute Ischemic Stroke. Ther Clin Risk Manag. 2025;21:239–56. Lv B-H et al. A machine learning-based predictive nomogram for early neurological improvement after thrombolysis in acute ischemic stroke. Front Neurol, 2025. Volume 16–2025. Wu K, et al. A nomogram predicts early neurological deterioration after mechanical thrombectomy in patients with ischemic stroke. Front Neurol. 2023;14:1255476. Luo B, et al. A novel nomogram predicting early neurological deterioration after intravenous thrombolysis for acute ischemic stroke. Heliyon. 2024;10(1):e23341. Menon BK, et al. Analysis of Workflow and Time to Treatment on Thrombectomy Outcome in the Endovascular Treatment for Small Core and Proximal Occlusion Ischemic Stroke (ESCAPE) Randomized, Controlled Trial. Circulation. 2016;133(23):2279–86. Joundi RA, et al. Time From Hospital Arrival Until Endovascular Thrombectomy and Patient-Reported Outcomes in Acute Ischemic Stroke. JAMA Neurol. 2024;81(7):752–61. Chung HI, et al. Delayed door to puncture time during off-duty hours is associated with unfavorable outcomes after mechanical thrombectomy in the early window of acute ischemic stroke. BMC Neurol. 2024;24(1):357. Gilbert F, et al. Door-to-puncture time in ischemic stroke with large vessel occlusion in France: Patient and hospital factors. Rev Neurol (Paris). 2025;181(6):556–62. Liu Z, et al. Reducing Door-to-Puncture Times for Mechanical Thrombectomy in a Large Tertiary Hospital. Neurol Clin Pract. 2024;14(5):e200325. Vickers AJ, Elkin EB. Decision curve analysis: a novel method for evaluating prediction models. Med Decis Mak. 2006;26(6):565–74. Vickers AJ, van Calster B, Steyerberg EW. A simple, step-by-step guide to interpreting decision curve analysis. Diagn Progn Res. 2019;3:18. Brott T, et al. Measurements of acute cerebral infarction: a clinical examination scale. Stroke. 1989;20(7):864–70. Kattan M. Encyclopedia of Medical Decision Making . Steyerberg EW, et al. Assessing the performance of prediction models: a framework for traditional and novel measures. Epidemiology. 2010;21(1):128–38. Collins GS, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): the TRIPOD statement. Ann Intern Med. 2015;162(1):55–63. Moons KG, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med. 2015;162(1):W1–73. Weyland CS, et al. Predictors for Failure of Early Neurological Improvement After Successful Thrombectomy in the Anterior Circulation. Stroke. 2021;52(4):1291–8. Lai Y, et al. Predictors of failure of early neurological improvement in early time window following endovascular thrombectomy: a multi-center study. Front Neurol. 2023;14:1227825. Li Y, et al. Predictors of Early Neurological Improvement in Patients with Anterior Large Vessel Occlusion and Successful Reperfusion Following Endovascular Thrombectomy-Does CT Perfusion Imaging Matter? Clin Neuroradiol. 2022;32(3):839–47. Desai S, et al. P-027Ultra-early functional improvement after stroke thrombectomy – predictors and implications. J NeuroInterventional Surg. 2021;13(Sup1):1. Zhang Z, et al. Decision curve analysis: a technical note. Ann Transl Med. 2018;6(15):308. Zhou J, et al. Development and Validation of a Nomogram for Predicting the 6-Year Risk of Cognitive Impairment Among Chinese Older Adults. J Am Med Dir Assoc. 2020;21(6):864–e8716. Additional Declarations No competing interests reported. Supplementary Files TableS1.docx Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 13 Apr, 2026 Reviews received at journal 09 Apr, 2026 Reviewers agreed at journal 29 Mar, 2026 Reviews received at journal 07 Feb, 2026 Reviewers agreed at journal 05 Feb, 2026 Reviewers invited by journal 28 Jan, 2026 Editor assigned by journal 28 Jan, 2026 Editor invited by journal 20 Jan, 2026 Submission checks completed at journal 20 Jan, 2026 First submitted to journal 20 Jan, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8536054","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":582711101,"identity":"f374bf40-2dc4-4ef1-8a40-5901329f67d9","order_by":0,"name":"Chenyang Zhao","email":"","orcid":"","institution":"Shanxi Medical University","correspondingAuthor":false,"prefix":"","firstName":"Chenyang","middleName":"","lastName":"Zhao","suffix":""},{"id":582711102,"identity":"65159246-fb61-4c22-87e3-567567f388b0","order_by":1,"name":"Xihua Li","email":"","orcid":"","institution":"Shanxi Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xihua","middleName":"","lastName":"Li","suffix":""},{"id":582711103,"identity":"bb9016e6-0c13-449d-b87c-0a1bfe6f9bb8","order_by":2,"name":"Yi Han","email":"","orcid":"","institution":"Shanxi Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yi","middleName":"","lastName":"Han","suffix":""},{"id":582711104,"identity":"3da49293-eb5d-424f-a594-49bdcae1ba8c","order_by":3,"name":"Tao Zhou","email":"","orcid":"","institution":"Shanxi Medical University","correspondingAuthor":false,"prefix":"","firstName":"Tao","middleName":"","lastName":"Zhou","suffix":""},{"id":582711105,"identity":"2ab17005-9ec6-479f-b36e-9042f107a9ee","order_by":4,"name":"Xuefei Ren","email":"","orcid":"","institution":"Shanxi Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xuefei","middleName":"","lastName":"Ren","suffix":""},{"id":582711106,"identity":"e47cc8af-8131-444c-9adc-6b7b70d8b48a","order_by":5,"name":"Yaxuan Sun","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA00lEQVRIiWNgGAWjYBACgwMQmpmxgfnAgQ8/SNPClnhwZg8RWiQb4Ewe48McbERo4Wc/Yybxc0cdO/PsMx8OM/AwyPOLHcCvhY0nLU2y98xhZsa+3A2HCywYDGfOTiCghSH5mARv2wFmxh7eDYdn8DAkGNwmpIX/YZvk37Y6oBaeB4d52IjRIpF8TJq3jRmkhYFYLc+SrWXbgH7pYTMABrIEEX7hzzG8+batLtmwh/nxhw8/bOT5pQlogYFkwwYwLUGcchCwkyde7SgYBaNgFIw0AABRxUAnAqWEMQAAAABJRU5ErkJggg==","orcid":"","institution":"Shanxi Medical University","correspondingAuthor":true,"prefix":"","firstName":"Yaxuan","middleName":"","lastName":"Sun","suffix":""}],"badges":[],"createdAt":"2026-01-07 02:53:09","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8536054/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8536054/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":101752812,"identity":"fa1d740c-5a39-44ef-873b-97fbf92d851d","added_by":"auto","created_at":"2026-02-03 10:32:49","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":22369,"visible":true,"origin":"","legend":"\u003cp\u003eFlow chart of patient selection for the ENI prediction study among patients undergoing emergency mechanical thrombectomy at Shanxi Provincial People’s Hospital, January 2020–December 2022.\u003c/p\u003e","description":"","filename":"Onlinefloatimage116.png","url":"https://assets-eu.researchsquare.com/files/rs-8536054/v1/561d52527320c495c69b7fbc.png"},{"id":101752977,"identity":"c68eadf7-4dcc-4a87-851c-d69caa967879","added_by":"auto","created_at":"2026-02-03 10:38:36","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":26860,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8536054/v1/b2a3b9349f13b78a93d94e04.png"},{"id":101752603,"identity":"67c6133c-74c5-40f6-8865-86d6b4d465fb","added_by":"auto","created_at":"2026-02-03 10:28:25","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":31589,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8536054/v1/bb58ee16a7b6726e346562a3.png"},{"id":101753028,"identity":"cef99d72-8c1e-4bfc-8368-133d74b87847","added_by":"auto","created_at":"2026-02-03 10:38:55","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":21728,"visible":true,"origin":"","legend":"\u003cp\u003eSee image above for figure legend\u003c/p\u003e","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8536054/v1/95645625f3d11e18ad8f9c1a.png"},{"id":101755620,"identity":"56ec571e-b8dd-46ff-b80d-c80e85c77911","added_by":"auto","created_at":"2026-02-03 10:53:22","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1144027,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8536054/v1/2deb0c63-5a07-48cc-970a-c376818c8591.pdf"},{"id":101752941,"identity":"a928e071-8a3d-4bf9-b727-9a81c88dc9bb","added_by":"auto","created_at":"2026-02-03 10:38:17","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":12458,"visible":true,"origin":"","legend":"","description":"","filename":"TableS1.docx","url":"https://assets-eu.researchsquare.com/files/rs-8536054/v1/c79622273e50e0a2be2c8304.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Development of a Simple Clinical Score to Predict Early Neurological Improvement after Mechanical Thrombectomy: A Single-Centre Cohort Study","fulltext":[{"header":"Introduction","content":"\u003cp\u003eEndovascular thrombectomy (EVT) has become the standard of care for selected patients with acute ischemic stroke due to large-vessel occlusion, after a series of landmark randomized controlled trials demonstrated substantial improvements in functional outcomes compared with best medical therapy alone.[\u003cspan additionalcitationids=\"CR2 CR3\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e] These trials have reshaped acute stroke workflows and underscored the importance of rapid reperfusion. Nevertheless, outcomes after EVT remain heterogeneous, and a considerable proportion of patients fail to achieve functional independence despite technically successful recanalization.[\u003cspan additionalcitationids=\"CR2 CR3\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e] In this context, early neurological improvement (ENI) has emerged as a pragmatic intermediate endpoint that reflects early treatment response and correlates with long-term prognosis.\u003c/p\u003e \u003cp\u003eENI is typically defined according to changes in the National Institutes of Health Stroke Scale (NIHSS), most often as an improvement of \u0026ge;\u0026thinsp;8\u0026ndash;10 points or as achieving a NIHSS score of 0\u0026ndash;1 within the first 24 hours or by day 7.[\u003cspan additionalcitationids=\"CR6 CR7\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e] Several studies have shown that ENI after intravenous thrombolysis or EVT is strongly associated with favourable 90-day modified Rankin Scale (mRS) [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e] scores and reduced mortality, and may serve as a surrogate endpoint for long-term outcomes in clinical research.[\u003cspan additionalcitationids=\"CR6 CR7\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e] For example, Wang et al. reported that ENI at day 1 following mechanical thrombectomy for distal medium-vessel occlusion stroke independently predicted good to excellent 3-month outcomes.[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e] Similarly, large registry and cohort studies have highlighted that early neurological trajectories\u0026mdash;encompassing both ENI and early neurological deterioration (END)\u0026mdash;are key determinants of functional recovery after EVT.[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e] Collectively, these data underscore the clinical value of identifying, soon after thrombectomy, patients who are likely or unlikely to experience ENI.\u003c/p\u003e \u003cp\u003eDespite the growing interest in ENI, evidence on its prediction after EVT remains limited. Most prior work has focused on a restricted set of factors, such as baseline NIHSS, age, infarct core size, collateral status and degree of reperfusion, often examined in heterogeneous populations or within single vascular territories.[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e] A few groups have proposed multivariable models or nomograms for ENI or closely related early outcomes. Zhang et al. developed a nomogram to predict ENI in ischemic stroke patients treated with EVT, identifying baseline NIHSS and imaging markers as key predictors.[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e] Wu et al. later constructed a nomogram to jointly predict 3-month unfavourable outcomes and ENI, further emphasizing the prognostic relevance of early neurological response.[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e] More recently, machine learning\u0026ndash;based nomograms have been reported for ENI after intravenous thrombolysis.[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e] However, many of these models are relatively complex, are not easily translated into simple bedside tools, and often lack comprehensive evaluation of calibration and clinical utility.\u003c/p\u003e \u003cp\u003eIn parallel, there has been increasing attention on END and unexplained END after EVT, which are consistently associated with poor outcomes.[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e] Several nomograms have been proposed to predict END after mechanical thrombectomy, mainly in anterior circulation large-vessel occlusion.[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e] These studies highlight that early post-procedural neurological change\u0026mdash;whether improvement or deterioration\u0026mdash;reflects a complex interplay among baseline clinical status, vascular pathology and reperfusion quality. Nevertheless, ENI-specific prediction models after EVT remain relatively scarce, and even fewer have been developed and evaluated in contemporary real-world cohorts from Asian populations, where stroke etiologies, risk factor profiles and treatment workflows may differ from those in Western trials.[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan additionalcitationids=\"CR14\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e] Time metrics from hospital arrival to reperfusion represent another crucial dimension that may influence ENI. Workflow analyses from EVT trials and registries have shown that shorter door-to-puncture (DPT) or door-to-arterial puncture times are associated with higher rates of functional independence and better quality-of-life outcomes.[\u003cspan additionalcitationids=\"CR20 CR21\" citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e] Quality improvement initiatives targeting DPT\u0026thinsp;\u0026lt;\u0026thinsp;60 minutes have demonstrated that systematic process optimization can translate into improved clinical results.[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e] However, the specific relationship between DPT and ENI after EVT is less well characterized, particularly in the ultra-late window (\u0026gt;\u0026thinsp;12 hours from last known well), where patient selection and tissue viability are more heterogeneous. Existing data suggest that the impact of in-hospital delays may not be strictly linear across the entire time range, raising the possibility that non-linear modelling approaches, such as restricted cubic splines, could provide additional insight.[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan additionalcitationids=\"CR20 CR21\" citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]\u003c/p\u003e \u003cp\u003eFrom a methodological perspective, contemporary standards for clinical prediction research emphasize not only discrimination but also calibration and clinical usefulness. Decision curve analysis (DCA) provides a formal framework for quantifying the net benefit of prediction models across a range of threshold probabilities, and for comparing them with default \u0026ldquo;treat-all\u0026rdquo; and \u0026ldquo;treat-none\u0026rdquo; strategies.[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e] However, DCA has rarely been applied to ENI prediction after EVT, and few studies have integrated traditional regression modelling, simplified bedside scoring systems, nomograms and DCA within a single coherent framework.\u003c/p\u003e \u003cp\u003eAgainst this background, we sought to address several knowledge gaps using a real-world cohort of patients undergoing EVT in a high-volume centre. Our primary objective was to develop and internally evaluate a multivariable logistic regression model for predicting ENI after thrombectomy based on routinely available clinical variables. Building on this model, we aimed to derive a simplified bedside risk score and a corresponding nomogram to facilitate rapid risk stratification in everyday practice. A key secondary objective was to explore the independent and potentially non-linear association between DPT and ENI, with particular attention to patients treated in an ultra-late time window. To provide a comprehensive assessment, we examined model discrimination, Brier score, calibration and net benefit using receiver operating characteristic (ROC) analysis, calibration plots and DCA. Our study therefore seeks to deliver a pragmatic, clinically interpretable tool for early risk assessment after EVT, while also clarifying the extent to which in-hospital treatment delays influence early neurological recovery.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\n \u003ch2\u003e3.1 Study design and population\u003c/h2\u003e\n \u003cp\u003eFor the purposes of this study, the data were accessed for research purposes on 01/12/2025. We conducted a single-centre retrospective cohort study at Shanxi Provincial People\u0026rsquo;s Hospital, including consecutive patients who underwent emergency endovascular thrombectomy (EVT) for acute ischaemic stroke between 01/01/2020 and 31/12/2022. Consecutive adult patients with acute ischemic stroke who underwent emergency mechanical thrombectomy (endovascular thrombectomy, EVT) during this period were screened for eligibility.\u003c/p\u003e\n \u003cp\u003eInclusion criteria were: (1) age\u0026thinsp;\u0026ge;\u0026thinsp;18 years; (2) acute ischemic stroke treated with EVT involving the anterior and/or posterior circulation; and (3) availability of both a baseline National Institutes of Health Stroke Scale (NIHSS) score on admission and a follow-up NIHSS assessment at approximately 7 days after EVT (or at discharge, if earlier). Exclusion criteria were: (1) missing data on any component required to define early neurological improvement (ENI); (2) missing key clinical covariates required for model development (e.g. baseline NIHSS, history of diabetes or cardioembolic disease, door-to-puncture time); and (3) loss to in-hospital follow-up before completion of the 7-day neurological assessment.\u003c/p\u003e\n \u003cp\u003eAfter application of these criteria, 185 patients were included in the final analysis; among them, 53 (28.6%) achieved ENI and 132 did not, as shown in the patient selection flow chart (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e). The study protocol was approved by the Ethics Committee of Shanxi Provincial People\u0026rsquo;s Hospital (approval No. 891). Owing to the retrospective design and use of de-identified data, the requirement for written informed consent was waived. All procedures complied with national regulations and the Declaration of Helsinki.\u003c/p\u003e\n \u003cp\u003eData for this retrospective analysis were extracted from the hospital electronic medical record system for patients treated between 01/01/2020 and 31/12/2022.For the purposes of this study, all data used for analysis were accessed from the hospital\u0026rsquo;s electronic medical record system between 01/01/2020 and 31/12/2022.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\n \u003ch2\u003e3.2 Endpoints and definitions\u003c/h2\u003e\n \u003cp\u003eThe primary endpoint was early neurological improvement (ENI) after EVT. ENI was defined a priori as either a reduction in NIHSS score of \u0026ge;\u0026thinsp;8 points from baseline to day 7, or an absolute NIHSS score\u0026thinsp;\u0026le;\u0026thinsp;1 at day 7. Patients who died before the day-7 assessment were classified as not having ENI. Baseline NIHSS was defined as the first NIHSS score recorded on admission before EVT.The National Institutes of Health Stroke Scale (NIHSS) is an established and previously published measure of neurological deficit. [\u003cspan class=\"CitationRef\"\u003e25\u003c/span\u003e]\u003c/p\u003e\n \u003cp\u003eHypertension was defined as a documented history of hypertension, current use of antihypertensive medication, or blood pressure\u0026thinsp;\u0026ge;\u0026thinsp;140/90 mmHg on repeated measurements during admission. Diabetes mellitus was defined as a previous diagnosis of diabetes, use of glucose-lowering therapy, or HbA1c\u0026thinsp;\u0026ge;\u0026thinsp;6.5% when available. Cardioembolic aetiology was considered present when atrial fibrillation or another major cardiac source of embolism (e.g. recent myocardial infarction, left ventricular thrombus, prosthetic valve) was documented by the treating team on the basis of standard investigations. Prior cerebrovascular disease (CVD) was defined as any history of ischaemic stroke, intracerebral haemorrhage or transient ischaemic attack.\u003c/p\u003e\n \u003cp\u003eDoor-to-puncture time (DPT) was defined as the interval, in hours, between hospital arrival (\u0026ldquo;door\u0026rdquo;) and arterial puncture for EVT. For secondary analyses, we additionally created a dichotomous variable (DPT\u0026thinsp;\u0026gt;\u0026thinsp;12 h vs\u0026thinsp;\u0026le;\u0026thinsp;12 h) to reflect ultra-late treatment. Where used, wake-up stroke was defined as stroke symptoms first recognised on awakening from sleep, with the exact time of onset unknown.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\n \u003ch2\u003e3.3 Candidate predictors\u003c/h2\u003e\n \u003cp\u003eCandidate predictors for ENI were selected a priori based on clinical relevance and previous literature on early neurological trajectories after reperfusion therapy. These included:\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eDemographics:\u0026nbsp;\u003c/strong\u003eage, sex;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eClinical history\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003ehypertension, diabetes, cardioembolic etiology (primarily atrial fibrillation), prior cerebrovascular disease;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eBaseline status\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003ebaseline NIHSS score, vascular territory (anterior vs posterior circulation);\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eTime metrics\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003edoor-to-puncture time (continuous, per hour).\u003c/p\u003e\n \u003cp\u003eAll candidate variables were first entered into univariable logistic regression models with ENI (yes/no) as the dependent variable. Baseline characteristics of patients with and without ENI are summarized in Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e.\u0026nbsp;\u003c/p\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eBaseline clinical characteristics according to early neurological improvement status. Values are presented as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation, median (interquartile range), or number (percentage), as appropriate. ENI indicates early neurological improvement. Analytic cohort includes 185 patients (ENI, n\u0026thinsp;=\u0026thinsp;53; no ENI, n\u0026thinsp;=\u0026thinsp;132).\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eVariable\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eAll patients (n\u0026thinsp;=\u0026thinsp;185)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eENI (n\u0026thinsp;=\u0026thinsp;53)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eNo ENI (n\u0026thinsp;=\u0026thinsp;132)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ep-value\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge, years\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e61.9\u0026thinsp;\u0026plusmn;\u0026thinsp;13.2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e62.3\u0026thinsp;\u0026plusmn;\u0026thinsp;13.8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e61.7\u0026thinsp;\u0026plusmn;\u0026thinsp;13.0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.785\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMale sex, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e125 (67.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e30 (56.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e95 (72.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.065\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePre-stroke mRS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.0 (0.0\u0026ndash;0.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.0 (0.0\u0026ndash;0.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.0 (0.0\u0026ndash;0.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.382\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBaseline NIHSS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e14.0 (11.0\u0026ndash;19.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e18.0 (12.0\u0026ndash;20.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e13.0 (11.0\u0026ndash;17.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWake-up stroke, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e48 (25.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e13 (24.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e35 (26.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.926\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHypertension, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e101 (54.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e32 (60.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e69 (52.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.402\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDiabetes, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27 (14.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e9 (17.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e18 (13.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.725\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePrior cerebrovascular disease, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e34 (18.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e10 (18.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e24 (18.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.000\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCardioembolic aetiology, n (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e64 (34.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e22 (41.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e42 (31.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.279\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAdmission glucose, mmol/L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6.9 (6.0\u0026ndash;8.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7.0 (6.2\u0026ndash;8.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6.9 (6.0\u0026ndash;8.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.685\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLDL cholesterol, mmol/L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.7 (2.1\u0026ndash;3.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.6 (2.0\u0026ndash;3.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.8 (2.2\u0026ndash;3.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.389\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHDL cholesterol, mmol/L\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.1 (0.9\u0026ndash;1.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.1 (0.9\u0026ndash;1.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.1 (0.9\u0026ndash;1.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.302\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHeart rate on admission, bpm\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e83.2\u0026thinsp;\u0026plusmn;\u0026thinsp;21.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e80.2\u0026thinsp;\u0026plusmn;\u0026thinsp;18.6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e84.4\u0026thinsp;\u0026plusmn;\u0026thinsp;22.4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.191\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDoor-to-puncture time, h\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.0 (5.0\u0026ndash;11.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7.0 (4.5\u0026ndash;10.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.0 (5.0\u0026ndash;11.0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.166\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003ch3\u003e3.4 Statistical analysis\u003c/h3\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\n \u003cdiv id=\"Sec7\" class=\"Section3\"\u003e\n \u003ch2\u003e3.4.1 Descriptive and univariable analyses\u003c/h2\u003e\n \u003cp\u003eContinuous variables were inspected for distribution and are presented as mean\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviation (SD) or median with interquartile range (IQR), as appropriate. Categorical variables are summarised as counts and percentages. Baseline characteristics were compared between patients with ENI and those without ENI using the Student\u0026rsquo;s t-test or the Mann\u0026ndash;Whitney U test for continuous variables, and the \u0026chi;\u0026sup2; test or Fisher\u0026rsquo;s exact test for categorical variables, as appropriate.\u003c/p\u003e\n \u003cp\u003eUnivariable associations between each candidate predictor and ENI were evaluated using logistic regression. Results are reported as crude odds ratios (ORs) with 95% confidence intervals (CIs) and corresponding p-values (Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e\n \u003ctable id=\"Tab2\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eUnivariable associations between candidate predictors and early neurological improvement. Crude odds ratios (ORs) and 95% confidence intervals (CIs) from univariable logistic regression models with early neurological improvement (ENI) as the dependent variable. ENI indicates early neurological improvement; DPT, door-to-puncture time.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePredictor\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCrude OR (95% CI)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ep-value\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge (per 10-year increase)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.04 (0.81\u0026ndash;1.32)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.778\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMale sex\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.51 (0.26\u0026ndash;0.99)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.045\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBaseline NIHSS (per 1-point increase)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.13 (1.06\u0026ndash;1.21)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDiabetes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.30 (0.54\u0026ndash;3.10)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.561\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCardioembolic disease\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.52 (0.79\u0026ndash;2.94)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.212\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePrior cerebrovascular disease\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.05 (0.46\u0026ndash;2.37)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.913\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHypertension\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.39 (0.73\u0026ndash;2.66)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.318\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWake-up stroke\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.90 (0.43\u0026ndash;1.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.780\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDPT (per 1-hour increase)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.93 (0.85\u0026ndash;1.02)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.125\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDPT\u0026thinsp;\u0026gt;\u0026thinsp;12 h vs\u0026thinsp;\u0026le;\u0026thinsp;12 h\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.41 (0.11\u0026ndash;1.45)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.165\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003ch3\u003e3.4.2 Multivariable model development\u003c/h3\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec8\" class=\"Section3\"\u003e\n \u003cp\u003eWe then fitted multivariable logistic regression models with ENI as the dependent variable. To preserve an adequate number of events per variable and reduce the risk of overfitting, we pre-specified a parsimonious set of clinically important predictors rather than relying solely on data-driven variable selection. The core model (Model 1) included age (per 10-year increase), sex (male vs female), baseline NIHSS (per 1-point increase), diabetes (yes/no), cardioembolic aetiology (yes/no) and prior cerebrovascular disease (yes/no). A second model (Model 2) additionally incorporated DPT\u0026thinsp;\u0026gt;\u0026thinsp;12 h (yes/no) to assess the incremental contribution of ultra-late treatment beyond the clinical core model.\u003c/p\u003e\n \u003cp\u003eFor each model, adjusted odds ratios (ORs) with 95% confidence intervals (CIs) and p-values were reported. Multicollinearity was evaluated using variance inflation factors, and overall model fit was examined using standard goodness-of-fit statistics. The multivariable results are summarised in Table \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e.\u0026nbsp;\u003c/p\u003e\n \u003ctable id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eMultivariable logistic regression models for early neurological improvement (Model 1: clinical core; Model 2: core\u0026thinsp;+\u0026thinsp;DPT\u0026thinsp;\u0026gt;\u0026thinsp;12 h). Adjusted odds ratios (ORs) and 95% confidence intervals (CIs) from multivariable logistic regression models. Model 1 includes age, sex, baseline NIHSS, diabetes, cardioembolic disease, and prior cerebrovascular disease. Model 2 additionally includes door-to-puncture time (DPT)\u0026thinsp;\u0026gt;\u0026thinsp;12 hours. \u0026ldquo;\u0026mdash;\u0026rdquo; indicates that the predictor was not included in the corresponding model. ENI indicates early neurological improvement; OR, odds ratio; CI, confidence interval; DPT, door-to-puncture time.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePredictor\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eAdjusted OR (95% CI) \u0026ndash; Model 1\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ep-value\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eAdjusted OR (95% CI) \u0026ndash; Model 2\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ep-value\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge (per 10-year increase)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.89 (0.68\u0026ndash;1.16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.394\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.90 (0.68\u0026ndash;1.18)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.428\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMale sex\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.52 (0.25\u0026ndash;1.07)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.077\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.51 (0.25\u0026ndash;1.06)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.072\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBaseline NIHSS (per 1-point increase)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.13 (1.05\u0026ndash;1.21)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.13 (1.05\u0026ndash;1.21)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDiabetes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.39 (0.54\u0026ndash;3.55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.494\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.36 (0.53\u0026ndash;3.49)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.528\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCardioembolic disease\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.32 (0.63\u0026ndash;2.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.457\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.21 (0.57\u0026ndash;2.55)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.620\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePrior cerebrovascular disease\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.02 (0.43\u0026ndash;2.43)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.960\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1.08 (0.45\u0026ndash;2.58)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.863\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDPT\u0026thinsp;\u0026gt;\u0026thinsp;12 h\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.44 (0.12\u0026ndash;1.66)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.226\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003ch3\u003e3.4.3 Simplified bedside score and nomogram\u003c/h3\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec9\" class=\"Section3\"\u003e\n \u003cp\u003eTo enhance clinical usability, we derived a simplified bedside risk score consistent with Table \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e. Baseline NIHSS was categorised into three strata (\u0026le;\u0026thinsp;10, 11\u0026ndash;15, and \u0026ge;\u0026thinsp;16) and assigned 0, 2, and 4 points, respectively. Cardioembolic aetiology contributed 1 point when present (0 when absent). The total score therefore ranged from 0 to 5 points and was further grouped into low (0\u0026ndash;2), intermediate (3\u0026ndash;4), and high (5) strata.\u003c/p\u003e\n \u003cp\u003eThe simplified score was then entered as a continuous predictor in a logistic regression model to estimate the predicted probability of ENI. In addition, a nomogram was constructed from the full multivariable model to provide an intuitive graphical tool for bedside estimation of ENI risk (Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\n \u003ctable id=\"Tab4\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eSimplified bedside risk score for early neurological improvement. Simplified point-based score derived from the multivariable model. Higher total scores indicate a higher predicted probability of early neurological improvement (ENI). The score uses only baseline NIHSS category and the presence of cardioembolic disease to facilitate bedside use.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePredictor\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCategory\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePoints\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBaseline NIHSS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026le;\u0026thinsp;10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e11\u0026ndash;15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026ge;\u0026thinsp;16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCardioembolic disease\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAbsent\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePresent\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003cp\u003eTotal score ranges and observed ENI rates in the derivation cohort\u0026nbsp;\u003c/p\u003e\n \u003ctable id=\"Taba\" border=\"1\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eTotal score range\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eRisk category\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eObserved ENI rate\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0\u0026ndash;2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLow\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e15/92 (16.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3\u0026ndash;4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIntermediate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e22/63 (34.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHigh\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e16/30 (53.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003ch3\u003e3.4.4 Model performance and internal validation\u003c/h3\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec10\" class=\"Section3\"\u003e\n \u003cp\u003eModel discrimination was quantified using the area under the receiver operating characteristic curve (AUC) with 95% confidence intervals (CIs). Overall prediction error was assessed using the Brier score, a proper scoring rule for probabilistic predictions. [\u003cspan class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e27\u003c/span\u003e] Calibration was evaluated by plotting observed versus predicted probabilities of ENI across deciles of predicted risk and by estimating the calibration intercept and slope from a logistic regression of the observed outcome on the logit of the predicted risk.\u003c/p\u003e\n \u003cp\u003eWe compared the performance of three models: (1) an NIHSS-only model; (2) the full multivariable model (Model 2); and (3) the simplified bedside score. Receiver operating characteristic curves for these models are shown in Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e. Calibration plots for the full model and the simplified score are presented in Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e. Apparent performance metrics (AUC, Brier score, calibration intercept and slope) are summarised in Supplementary Table S2. Internal validation was performed using bootstrap resampling (1,000 repetitions) to quantify optimism and obtain optimism-corrected estimates of model performance, including AUC, Brier score, calibration intercept and calibration slope. External validation was not performed.\u003c/p\u003e\n \u003cp\u003eInternal validation was performed using bootstrap resampling (1,000 repetitions) to estimate optimism and to obtain optimism-corrected AUC, Brier score, calibration intercept, and calibration slope. Both apparent and optimism-corrected performance metrics are reported in Supplementary Table S2. External validation was not performed.\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec11\" class=\"Section3\"\u003e\n \u003ch2\u003e3.4.5 Additional analyses\u003c/h2\u003e\n \u003cp\u003eTo explore potential non-linear associations between DPT and ENI, we fitted logistic regression models with DPT entered as a continuous predictor using restricted cubic splines with pre-specified degrees of freedom. Odds ratios for ENI across the observed DPT range were plotted using 6 hours as the reference value, and 95% confidence intervals (CIs) were displayed to illustrate estimation uncertainty, particularly in the ultra-late window (\u0026gt;\u0026thinsp;12 h) (Figure S2).\u003c/p\u003e\n \u003cp\u003eSensitivity analyses comprised: (1) restricting the cohort to anterior circulation strokes; and (2) applying an alternative ENI definition based solely on a\u0026thinsp;\u0026ge;\u0026thinsp;8-point NIHSS reduction, without the additional absolute NIHSS\u0026thinsp;\u0026le;\u0026thinsp;1 criterion. For each sensitivity scenario, the multivariable model was refitted and the key coefficients were compared with those from the primary analysis (Table S1).\u003c/p\u003e\n \u003cp\u003eWe also generated a forest plot of adjusted odds ratios from the main multivariable model to visually summarise the relative contribution of each predictor (Figure S1). Finally, decision curve analysis was performed to compare the net benefit of the full model, the simplified bedside score, a \u0026ldquo;treat-all\u0026rdquo; strategy and a \u0026ldquo;treat-none\u0026rdquo; strategy across a range of clinically relevant threshold probabilities (Figure S3).\u003c/p\u003e\n \u003cp\u003eAll statistical analyses were conducted using R (R Foundation for Statistical Computing, Vienna, Austria) and Python (Python Software Foundation), with packages including statsmodels, scikit-learn and rms. A two-sided p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistically significant.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e4.1 Study population and baseline characteristics\u003c/h2\u003e \u003cp\u003eBetween January 2020 and December 2022, a total of 185 patients met the inclusion criteria and were included in the final analysis. Among them, 53 patients (28.6%) achieved early neurological improvement (ENI) and 132 did not. The detailed flow of patient selection is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003eBaseline clinical and imaging characteristics according to ENI status are summarised in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Patients with ENI generally presented with more severe neurological deficits at baseline, as reflected by higher NIHSS scores, compared with those without ENI. The distributions of age and vascular territory (anterior vs posterior circulation) were broadly similar between groups. The prevalences of hypertension, diabetes and prior cerebrovascular disease were also comparable, whereas cardioembolic aetiology appeared somewhat more frequent among patients who achieved ENI, although this trend did not clearly reach statistical significance. At the descriptive level, door-to-puncture time did not differ appreciably between ENI and non-ENI groups.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e4.2 Univariable associations\u003c/h2\u003e \u003cp\u003eIn univariable logistic regression analyses (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e), higher baseline NIHSS was strongly associated with a greater likelihood of ENI (crude OR per 1-point increase 1.13; 95% CI 1.06\u0026ndash;1.21; p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Male sex was associated with a lower probability of ENI (OR 0.51; 95% CI 0.26\u0026ndash;0.99; p\u0026thinsp;=\u0026thinsp;0.045). Age (per 10-year increase) showed no significant association with ENI.\u003c/p\u003e \u003cp\u003eDiabetes, cardioembolic aetiology and prior cerebrovascular disease all had crude ORs\u0026thinsp;\u0026gt;\u0026thinsp;1, indicating numerically higher ENI rates among patients with these conditions; however, none of these associations reached statistical significance in univariable analyses (all p\u0026thinsp;\u0026gt;\u0026thinsp;0.20). In contrast, a door-to-puncture time\u0026thinsp;\u0026gt;\u0026thinsp;12 h tended to be associated with lower odds of ENI (OR 0.41; 95% CI 0.11\u0026ndash;1.45; p\u0026thinsp;=\u0026thinsp;0.17), but this finding was not statistically significant and is compatible with only a modest and imprecisely estimated effect.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e4.3 Multivariable prediction model\u003c/h2\u003e \u003cp\u003eIn the multivariable logistic regression model including age (per 10-year increase), sex, baseline NIHSS, diabetes, cardioembolic aetiology, prior cerebrovascular disease and DPT\u0026thinsp;\u0026gt;\u0026thinsp;12 h (Model 2), baseline NIHSS remained the only independent predictor of ENI with a statistically robust effect. Each 1-point increase in baseline NIHSS was associated with approximately 13% higher odds of ENI (adjusted OR 1.13; 95% CI 1.05\u0026ndash;1.21; p\u0026thinsp;=\u0026thinsp;0.001).\u003c/p\u003e \u003cp\u003eMale sex showed a tendency towards reduced ENI (adjusted OR 0.51; 95% CI 0.25\u0026ndash;1.06; p\u0026thinsp;=\u0026thinsp;0.07), although this did not reach conventional statistical significance after adjustment. Diabetes (adjusted OR 1.36; 95% CI 0.53\u0026ndash;3.49; p\u0026thinsp;=\u0026thinsp;0.53), cardioembolic aetiology (adjusted OR 1.21; 95% CI 0.57\u0026ndash;2.55; p\u0026thinsp;=\u0026thinsp;0.62) and prior cerebrovascular disease (adjusted OR 1.08; 95% CI 0.45\u0026ndash;2.58; p\u0026thinsp;=\u0026thinsp;0.86) were not independently associated with ENI. DPT\u0026thinsp;\u0026gt;\u0026thinsp;12 h was associated with lower odds of ENI (adjusted OR 0.44; 95% CI 0.12\u0026ndash;1.66; p\u0026thinsp;=\u0026thinsp;0.23), but the wide confidence interval was compatible with no clear effect.\u003c/p\u003e \u003cp\u003eThe overall pattern of adjusted associations is illustrated in the forest plot (Supplementary Figure S1), which visually highlights the dominant contribution of baseline NIHSS and the comparatively modest and imprecise effects of the other variables.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e4.4 Simplified bedside score and nomogram\u003c/h2\u003e \u003cp\u003eOn the basis of the multivariable model, a simplified bedside risk score was constructed using baseline NIHSS category and cardioembolic aetiology (Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). Baseline NIHSS was categorised as \u0026le;\u0026thinsp;10, 11\u0026ndash;15, and \u0026ge;\u0026thinsp;16, assigned 0, 2, and 4 points, respectively, and cardioembolic aetiology contributed 1 point when present. The resulting total score ranged from 0 to 5.\u003c/p\u003e \u003cp\u003eWhen grouped by total score, ENI rates increased across strata: 15/92 (16.3%) in the low-score group (0\u0026ndash;2), 22/63 (34.9%) in the intermediate group (3\u0026ndash;4), and 16/30 (53.3%) in the high-score group (5). These findings support pragmatic bedside risk stratification, while external validation is warranted.\u003c/p\u003e \u003cp\u003eThe corresponding nomogram derived from the full multivariable model (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e4\u003c/span\u003e) provides a graphical representation of the contribution of each predictor and allows clinicians to estimate an individual patient\u0026rsquo;s ENI probability by summing points across variables.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003e4.5 Model performance\u003c/h2\u003e \u003cp\u003eThe full multivariable model (Model 2) demonstrated moderate discrimination, with an AUC of 0.71 and a Brier score of 0.18 in the derivation cohort. An NIHSS-only model achieved an AUC of 0.67 (Brier score 0.19), indicating that baseline NIHSS alone already captured a substantial proportion of the predictive information. The simplified bedside score model yielded an AUC of 0.66 and a Brier score of 0.19, retaining most of the discriminative ability of the full model while relying on a more compact set of predictors.\u003c/p\u003e \u003cp\u003eBootstrap internal validation (1,000 resamples) was performed to estimate optimism and to obtain optimism-corrected performance. The full model (Model 2) achieved an apparent AUC of 0.706 and an optimism-corrected AUC of 0.657, with a corresponding Brier score of 0.181 (apparent) and 0.197 (optimism-corrected). Optimism-corrected calibration for the full model showed an intercept of \u0026minus;\u0026thinsp;0.335 and a slope of 0.591, suggesting some overfitting. The simplified score showed an apparent and optimism-corrected AUC of 0.677, while the NIHSS-only model showed an apparent AUC of 0.673 and an optimism-corrected AUC of 0.674. Apparent and optimism-corrected performance metrics for all models are summarised in Supplementary Table S2.\u003c/p\u003e \u003cp\u003eIn decision curve analysis (Supplementary Figure S3), both the full multivariable model and the simplified bedside score provided greater net benefit than the default \u0026ldquo;treat-all\u0026rdquo; or \u0026ldquo;treat-none\u0026rdquo; strategies across a clinically relevant range of threshold probabilities (approximately 0.20\u0026ndash;0.50). The full model offered only modest incremental net benefit over the simplified score within this range, suggesting that the simpler tool may be sufficient for routine clinical use in many settings.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003e4.6 Additional and sensitivity analyses\u003c/h2\u003e \u003cp\u003eRestricted cubic spline analysis of door-to-puncture time showed no strong overall association between DPT and ENI across the observed time range (Supplementary Figure S2). Odds ratios for ENI remained close to 1 for most DPT values, and estimates beyond 12 hours were accompanied by wide confidence intervals, reflecting the small number of patients treated in this ultra-late window and substantial uncertainty regarding the true effect.\u003c/p\u003e \u003cp\u003eIn sensitivity analyses (Supplementary Table S1), restricting the cohort to anterior circulation strokes and applying an alternative ENI definition based solely on a\u0026thinsp;\u0026ge;\u0026thinsp;8-point NIHSS reduction yielded results broadly consistent with the main analysis. Baseline NIHSS remained the principal independent predictor of ENI, and effect estimates for other covariates, including DPT\u0026thinsp;\u0026gt;\u0026thinsp;12 h, were directionally similar but imprecise. None of the sensitivity analyses materially altered the overall conclusions of the study.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003e5.1 Main findings\u003c/h2\u003e \u003cp\u003eIn this single-centre, real-world EVT cohort, approximately one third of patients achieved early neurological improvement (ENI), and ENI once again emerged as an early marker closely linked to downstream functional outcomes, consistent with previous reports. [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e] ENI was strongly and independently associated with baseline NIHSS, whereas other routinely collected clinical variables\u0026mdash;diabetes, cardioembolic aetiology, prior cerebrovascular disease and door-to-puncture time (DPT)\u0026mdash;showed only modest and statistically imprecise associations.\u003c/p\u003e \u003cp\u003eWe developed a multivariable logistic regression model that demonstrated moderate discrimination (AUC\u0026thinsp;\u0026asymp;\u0026thinsp;0.71) and acceptable calibration, and translated this model into a simplified bedside score based on baseline NIHSS category and cardioembolic aetiology, which enabled pragmatic risk stratification with observed ENI rates of 16.3% (low), 34.9% (intermediate), and 53.3% (high) in the derivation cohort.Decision curve analysis indicated that both the full model and the simplified score provided greater net benefit than \u0026ldquo;treat-all\u0026rdquo; or \u0026ldquo;treat-none\u0026rdquo; strategies across a clinically relevant range of threshold probabilities, with only a modest incremental gain for the full model.[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e] Restricted cubic spline analysis suggested that DPT\u0026mdash;including the ultra-late window beyond 12 hours\u0026mdash;had at most a limited and imprecisely estimated independent effect on ENI once baseline NIHSS and other factors were accounted for, which is consistent with the notion that ENI is driven primarily by underlying tissue status rather than by relatively small differences in in-hospital time metrics.[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e5.2 Comparison with previous studies\u003c/h2\u003e \u003cp\u003eOur results are consistent with previous studies showing that ENI, or related early neurological trajectories, are strong surrogates for 90-day functional outcome and mortality after reperfusion therapies.[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e] For example, Kobeissi et al. and others reported that broadly defined ENI after EVT is associated with higher rates of functional independence and lower rates of death and symptomatic intracranial haemorrhage.[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e] Desai et al. and Katano et al. further demonstrated that very early or 24-hour NIHSS thresholds are powerful predictors of long-term outcome, underscoring the clinical relevance of early neurological trajectories.[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]\u003c/p\u003e \u003cp\u003eSeveral groups have proposed ENI prediction tools in EVT-treated populations. Zhang et al. developed a multicentre nomogram incorporating age, glucose, ASPECTS, recanalisation status and sICH, achieving AUCs of approximately 0.80 in the derivation cohort and 0.75 in external validation.[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e] Li et al. reported that CTP-ASPECTS may outperform non-contrast CT in predicting ENI after successful reperfusion, highlighting the added value of advanced imaging. [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e] More recently, Lai et al. investigated \u0026ldquo;ultra-early\u0026rdquo; neurological improvement immediately after reperfusion and again found that baseline NIHSS and imaging markers are key determinants of early trajectories.[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e] Wu and colleagues integrated ENI and 3-month outcomes into a broader prognostic framework using nomograms.[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]\u003c/p\u003e \u003cp\u003eIn parallel, a substantial body of literature has focused on early neurological deterioration (END) after EVT. Nomograms and machine-learning models for END typically identify baseline severity, infarct burden and procedure-related factors as major predictors. [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e] Dai et al. specifically examined unexplained END in posterior circulation disease, further highlighting the complex interplay between baseline status, vascular territory and procedural variables. [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e] Our study complements this work by focusing on ENI rather than END and by providing a simple clinical score to identify patients most likely to improve, which may be particularly useful for clinical counselling and for enriching clinical trials with high-benefit candidates.\u003c/p\u003e \u003cp\u003eWith respect to workflow metrics, multiple trials and registries have demonstrated that shorter onset-to-reperfusion or CT-to-reperfusion times are associated with better functional outcomes after EVT.[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e] The ESCAPE trial showed that every 30-minute delay in CT-to-reperfusion time reduces the probability of functional independence,[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e] while more recent work by Joundi et al. linked faster DPT with better health-related quality of life.[\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e] Our analysis differs in that we specifically examined ENI as the outcome in a modestly sized, single-centre cohort in which most patients were treated within institutional time targets. In this context, DPT did not exhibit a strong independent association with ENI after adjustment, and restricted cubic spline analysis revealed wide confidence intervals in the ultra-late window because of the small number of patients. Therefore, our findings should not be interpreted as contradicting the \u0026ldquo;time is brain\u0026rdquo; paradigm, but rather as reflecting limited statistical power and the possibility that ENI depends more on tissue-level factors (e.g. infarct core, collateral status, infarct growth rate) than on relatively small in-hospital time variations.[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]\u003c/p\u003e \u003cp\u003eFinally, compared with our earlier single-centre work in which admission NIHSS and diabetes were identified as independent predictors of in-hospital neurological improvement, the present analysis represents a methodological and translational extension. We move from a two-factor association model to a more comprehensive prediction framework: we report discrimination, Brier score and calibration; derive a simplified bedside score and nomogram; and formally evaluate net clinical benefit using decision curve analysis, in line with contemporary recommendations for prediction model development and reporting. [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e5.3 Clinical implications\u003c/h2\u003e \u003cp\u003eFrom a practical perspective, our findings suggest that a simple three-component bedside score based on NIHSS strata, diabetes and cardioembolic aetiology can provide an easily interpretable estimate of ENI probability after EVT. Patients with low scores (0\u0026ndash;2 points) have roughly a one-in-five chance of ENI, whereas those with scores of 3\u0026ndash;4 points approach a one-in-two probability. Very high scores appear to identify patients with a high likelihood of early improvement, although this estimate is derived from a small number of individuals and should be interpreted cautiously. Such risk stratification may help clinicians and families form more realistic expectations in the first days after thrombectomy, before 90-day outcomes are known, and can be integrated into post-procedural workflows to support early risk assessment, ICU and ward resource allocation, and shared decision-making. For example, patients with a low predicted probability of ENI may be flagged for closer monitoring of complications and more intensive supportive care, whereas those with higher predicted probability could be prioritised for early rehabilitation planning and discharge optimisation.\u003c/p\u003e \u003cp\u003eIn addition, by providing a quantitative and reproducible estimate of early improvement that does not rely on advanced imaging or specialised software, the score and its corresponding nomogram could be used as stratification or enrichment tools in clinical trials of neuroprotective agents, blood pressure management strategies or post-EVT care bundles, helping to target interventions to patients most likely to demonstrate measurable early benefit. [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e] At the same time, the only moderate AUC and modest net-benefit gain over a NIHSS-only strategy indicate that this tool should be regarded as an adjunct rather than a gatekeeper and should not be used to deny reperfusion therapy or to justify withholding guideline-based standard care.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section2\"\u003e \u003ch2\u003e5.4 Strengths and limitations\u003c/h2\u003e \u003cp\u003eThis study has several strengths. First, it is based on a consecutive, real-world EVT cohort from Shanxi Provincial People\u0026rsquo;s Hospital, a high-volume comprehensive stroke centre in northern China, which enhances the clinical relevance and applicability of the findings. Second, we adhered to key principles of prediction model development and reporting, including the use of multiple performance metrics (AUC, Brier score and calibration indices) and decision curve analysis to quantify clinical utility, in line with TRIPOD and related guidance. Third, we explicitly translated the statistical model into a simplified bedside score and nomogram, thereby moving beyond pure association towards a pragmatic tool that can be implemented without advanced imaging or specialised software.\u003c/p\u003e \u003cp\u003eSeveral important limitations must also be acknowledged. The single-centre retrospective design raises concerns about selection bias and limits the generalisability of the results. The sample size is modest, yielding an events-per-variable ratio close to the lower recommended boundary; as a consequence, some covariate effects\u0026mdash;particularly those related to DPT\u0026thinsp;\u0026gt;\u0026thinsp;12 h\u0026mdash;are estimated imprecisely. Advanced imaging predictors such as infarct core volume, infarct growth rate, collateral scores or CTP-ASPECTS, which have been shown to be relevant for ENI in other studies, were not systematically available in this cohort. Although bootstrap internal validation was used to quantify optimism and we report optimism-corrected discrimination and calibration, external validation in independent multicentre cohorts is still required before routine clinical implementation.Finally, our primary endpoint was ENI rather than the 90-day modified Rankin Scale; although ENI is a validated surrogate marker, it does not fully capture delayed complications, the impact of rehabilitation or long-term quality of life.\u003c/p\u003e \u003cp\u003eFuture work should therefore prioritise external validation and potential updating of this scoring system in larger multicentre cohorts, incorporate advanced imaging variables where available, and assess whether embedding the score and nomogram into clinical workflows can improve both ENI and longer-term functional outcomes in prospective impact studies.[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn this single-centre, real-world EVT cohort, we developed and evaluated an ENI prediction model based entirely on simple, routinely available clinical variables. Baseline NIHSS emerged as the dominant independent predictor of early neurological improvement, and a parsimonious multivariable logistic regression model demonstrated moderate discrimination with acceptable calibration. From this model, we derived a three-component simplified bedside score (NIHSS strata, diabetes and cardioembolic aetiology) and a corresponding nomogram, both of which provided intuitive risk stratification and yielded meaningful net clinical benefit over \u0026ldquo;treat-all\u0026rdquo; and \u0026ldquo;treat-none\u0026rdquo; strategies across a clinically relevant range of threshold probabilities.\u003c/p\u003e \u003cp\u003eThese findings indicate that the proposed model and simplified score may serve as practical tools for early post-thrombectomy risk stratification, supporting clinical decision-making, resource allocation and communication with patients and their families. However, the modest sample size, single-centre retrospective design, use of ENI rather than long-term functional outcome, and absence of external validation mean that the results should be interpreted with caution. Before routine clinical implementation, the score and nomogram should undergo independent validation and, if necessary, updating in larger multicentre cohorts, and should ideally be tested for their impact on both ENI and 90-day functional outcomes.\u003c/p\u003e"},{"header":"Declarations","content":" \u003ch2\u003eEthical Approval\u003c/h2\u003e \u003cp\u003eThis study was approved by the Ethics Committee of Shanxi Provincial People's Hospital (approval No. 891). Owing to the retrospective design and use of de-identified clinical data, the committee waived the requirement for written informed consent.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eCompeting Interests\u003c/strong\u003e \u003cp\u003eThe authors declare that there are no potential conflicts of interest related to the publication of this article.\u003c/p\u003e \u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eCZ and YS conceived and designed the study. CZ, XL, YH, TZ, and XR were responsible for patient screening and data collection. CZ performed the statistical analyses and drafted the manuscript. XL and YH contributed to data interpretation and critical revision of the manuscript. TZ and XR contributed to data curation/verification and helped revise the manuscript. YS supervised the study and critically revised the manuscript for important intellectual content. All authors read and approved the final manuscript.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe de-identified patient-level dataset contains sensitive clinical information and cannot be publicly shared under institutional and legal requirements. Data may be made available from the Ethics Committee of Shanxi Provincial People\u0026rsquo;s Hospital (approval No. 891) upon reasonable request and with appropriate approvals and a data-use agreement.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eBerkhemer OA, et al. A randomized trial of intraarterial treatment for acute ischemic stroke. N Engl J Med. 2015;372(1):11\u0026ndash;20.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGoyal M, et al. Randomized assessment of rapid endovascular treatment of ischemic stroke. N Engl J Med. 2015;372(11):1019\u0026ndash;30.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSaver JL, et al. Stent-retriever thrombectomy after intravenous t-PA vs. t-PA alone in stroke. N Engl J Med. 2015;372(24):2285\u0026ndash;95.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJovin TG, et al. Thrombectomy within 8 hours after symptom onset in ischemic stroke. N Engl J Med. 2015;372(24):2296\u0026ndash;306.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOng CT, et al. Early neurological improvement after intravenous tissue plasminogen activator infusion in patients with ischemic stroke aged 80 years or older. J Chin Med Assoc. 2014;77(4):179\u0026ndash;83.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKurmann CC, et al. Association of the 24-Hour National Institutes of Health Stroke Scale After Mechanical Thrombectomy With Early and Long‐Term Survival. Volume 2. Stroke: Vascular and Interventional Neurology; 2022. p. e000244. 4.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKobeissi H, et al. Early neurological improvement as a predictor of outcomes after endovascular thrombectomy for stroke: a systematic review and meta-analysis. J Neurointerv Surg. 2023;15(6):547\u0026ndash;51.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLai Y, et al. Identifying the predictors of ultra early neurological improvement and its role in functional outcome after endovascular thrombectomy in acute ischemic stroke. Front Neurol. 2025;16:1492013.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003evan Swieten JC, et al. Interobserver agreement for the assessment of handicap in stroke patients. Stroke. 1988;19(5):604\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang M, et al. Early Neurological Improvement Predicts Clinical Outcome After Thrombectomy for Distal Medium Vessel Occlusions. Front Neurol. 2022;13:809066.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang Z et al. Infarct Growth Rate Predicts Early Neurological Improvement in Ischemic Stroke After Endovascular Thrombectomy. Brain Sci, 2025. 15(3).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDai Z, et al. Predictors of unexplained early neurological deterioration after thrombectomy for posterior circulation infarction: a reanalysis of the BASILAR study. J Neurosurg. 2024;141(6):1705\u0026ndash;11.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang X, et al. Nomogram predicting early neurological improvement in ischaemic stroke patients treated with endovascular thrombectomy. Eur J Neurol. 2021;28(1):152\u0026ndash;60.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu Y, et al. Nomogram-Based Prediction of 3-Month Unfavorable Outcome and Early Neurological Deterioration After Endovascular Thrombectomy in Acute Ischemic Stroke. Ther Clin Risk Manag. 2025;21:239\u0026ndash;56.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLv B-H et al. A machine learning-based predictive nomogram for early neurological improvement after thrombolysis in acute ischemic stroke. Front Neurol, 2025. Volume 16\u0026ndash;2025.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu K, et al. A nomogram predicts early neurological deterioration after mechanical thrombectomy in patients with ischemic stroke. Front Neurol. 2023;14:1255476.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLuo B, et al. A novel nomogram predicting early neurological deterioration after intravenous thrombolysis for acute ischemic stroke. Heliyon. 2024;10(1):e23341.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMenon BK, et al. Analysis of Workflow and Time to Treatment on Thrombectomy Outcome in the Endovascular Treatment for Small Core and Proximal Occlusion Ischemic Stroke (ESCAPE) Randomized, Controlled Trial. Circulation. 2016;133(23):2279\u0026ndash;86.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJoundi RA, et al. Time From Hospital Arrival Until Endovascular Thrombectomy and Patient-Reported Outcomes in Acute Ischemic Stroke. JAMA Neurol. 2024;81(7):752\u0026ndash;61.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChung HI, et al. Delayed door to puncture time during off-duty hours is associated with unfavorable outcomes after mechanical thrombectomy in the early window of acute ischemic stroke. BMC Neurol. 2024;24(1):357.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGilbert F, et al. Door-to-puncture time in ischemic stroke with large vessel occlusion in France: Patient and hospital factors. Rev Neurol (Paris). 2025;181(6):556\u0026ndash;62.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu Z, et al. Reducing Door-to-Puncture Times for Mechanical Thrombectomy in a Large Tertiary Hospital. Neurol Clin Pract. 2024;14(5):e200325.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVickers AJ, Elkin EB. Decision curve analysis: a novel method for evaluating prediction models. Med Decis Mak. 2006;26(6):565\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVickers AJ, van Calster B, Steyerberg EW. A simple, step-by-step guide to interpreting decision curve analysis. Diagn Progn Res. 2019;3:18.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrott T, et al. Measurements of acute cerebral infarction: a clinical examination scale. Stroke. 1989;20(7):864\u0026ndash;70.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKattan M. \u003cem\u003eEncyclopedia of Medical Decision Making\u003c/em\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSteyerberg EW, et al. Assessing the performance of prediction models: a framework for traditional and novel measures. Epidemiology. 2010;21(1):128\u0026ndash;38.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCollins GS, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): the TRIPOD statement. Ann Intern Med. 2015;162(1):55\u0026ndash;63.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMoons KG, et al. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med. 2015;162(1):W1\u0026ndash;73.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWeyland CS, et al. Predictors for Failure of Early Neurological Improvement After Successful Thrombectomy in the Anterior Circulation. Stroke. 2021;52(4):1291\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLai Y, et al. Predictors of failure of early neurological improvement in early time window following endovascular thrombectomy: a multi-center study. Front Neurol. 2023;14:1227825.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi Y, et al. Predictors of Early Neurological Improvement in Patients with Anterior Large Vessel Occlusion and Successful Reperfusion Following Endovascular Thrombectomy-Does CT Perfusion Imaging Matter? Clin Neuroradiol. 2022;32(3):839\u0026ndash;47.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDesai S, et al. P-027Ultra-early functional improvement after stroke thrombectomy \u0026ndash; predictors and implications. J NeuroInterventional Surg. 2021;13(Sup1):1.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang Z, et al. Decision curve analysis: a technical note. Ann Transl Med. 2018;6(15):308.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou J, et al. Development and Validation of a Nomogram for Predicting the 6-Year Risk of Cognitive Impairment Among Chinese Older Adults. J Am Med Dir Assoc. 2020;21(6):864\u0026ndash;e8716.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-neurology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"nurl","sideBox":"Learn more about [BMC Neurology](http://bmcneurol.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/nurl","title":"BMC Neurology","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"stroke, thrombectomy, early neurological improvement, prediction model, risk score, nomogram","lastPublishedDoi":"10.21203/rs.3.rs-8536054/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8536054/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003eBackground\u003c/b\u003e\u003c/p\u003e \u003cp\u003eEarly neurological improvement (ENI) after endovascular thrombectomy (EVT) is strongly associated with long-term functional outcome, but simple bedside tools to predict ENI are limited. We aimed to develop and internally evaluate a pragmatic clinical model and simplified risk score for ENI after EVT.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMethods\u003c/b\u003e\u003c/p\u003e \u003cp\u003eWe conducted a single-centre retrospective cohort study including consecutive patients with acute ischemic stroke who underwent emergency mechanical thrombectomy at Shanxi Provincial People\u0026rsquo;s Hospital between January 2020 and December 2022. ENI was defined as a decrease in NIHSS\u0026thinsp;\u0026ge;\u0026thinsp;8 points or NIHSS\u0026thinsp;\u0026le;\u0026thinsp;1 at day 7. Candidate predictors included age, sex, baseline NIHSS, diabetes, cardioembolic etiology, prior cerebrovascular disease and door-to-puncture time (DPT, including DPT\u0026thinsp;\u0026gt;\u0026thinsp;12 h). Multivariable logistic regression was used to build a full prediction model. A simplified bedside score based on NIHSS strata, diabetes and cardioembolic etiology was derived from model coefficients. Calibration, Brier score, and decision curve analysis (DCA) were assessed, and internal validation was performed using bootstrap resampling (1,000 repetitions) to obtain optimism-corrected performance estimates.\u003c/p\u003e\u003cp\u003e\u003cb\u003eResults\u003c/b\u003e\u003c/p\u003e \u003cp\u003eA total of 185 patients were included, of whom 53 (28.6%) achieved early neurological improvement (ENI). In multivariable logistic regression, baseline NIHSS was the dominant predictor of ENI, whereas the incremental effects of other candidate predictors were modest. The full model (Model 2) showed an apparent AUC of 0.706 and an optimism-corrected AUC of 0.657 after 1,000 bootstrap resamples; the Brier score was 0.181 (apparent) and 0.197 (optimism-corrected). The simplified bedside score demonstrated an apparent and optimism-corrected AUC of 0.677, while the NIHSS-only model yielded an apparent AUC of 0.673 and an optimism-corrected AUC of 0.674. Bootstrap-corrected calibration of the full model suggested some overfitting (intercept \u0026minus;\u0026thinsp;0.335, slope 0.591), whereas the simplified score showed acceptable calibration (intercept \u0026minus;\u0026thinsp;0.108, slope 0.893). Decision curve analysis indicated that the full model and simplified score provided net benefit over treat-all and treat-none across clinically relevant threshold probabilities.\u003c/p\u003e\u003cp\u003e\u003cb\u003eConclusions\u003c/b\u003e\u003c/p\u003e \u003cp\u003eA parsimonious clinical model and a simplified bedside score based on baseline NIHSS category and cardioembolic aetiology provided moderate discrimination for ENI after EVT; external validation is warranted.External validation in larger multicentre cohorts is warranted.\u003c/p\u003e","manuscriptTitle":"Development of a Simple Clinical Score to Predict Early Neurological Improvement after Mechanical Thrombectomy: A Single-Centre Cohort Study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-02 05:57:43","doi":"10.21203/rs.3.rs-8536054/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-04-13T13:36:16+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-09T19:48:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"285266487935214397310439561351203766512","date":"2026-03-29T18:14:20+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-02-07T21:35:43+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"54663145419158049059659832066999047454","date":"2026-02-06T03:18:00+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-01-29T00:39:23+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-01-29T00:29:43+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-01-20T16:56:48+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-01-20T15:41:14+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Neurology","date":"2026-01-20T15:17:34+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-neurology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"nurl","sideBox":"Learn more about [BMC Neurology](http://bmcneurol.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/nurl","title":"BMC Neurology","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"7ec96960-e2b9-44b9-9cbb-a04d4952080b","owner":[],"postedDate":"February 2nd, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-22T06:55:34+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-02 05:57:43","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8536054","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8536054","identity":"rs-8536054","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}