Search

Background Post-stroke cognitive impairment (PSCI) brings a heavy burden to patients and their families. An early recognition and intervention can help delay the occurrence and development of PSCI. Therefore, the use of accurate neuropsychological assessment tools to screen for PSCI is essential for the management and treatment of PSCI. Objective To analyze the screening accuracy of assessment tools for PSCI by meta-analysis, thus providing references for an accurate screening of PSCI. Methods Diagnostic trials on screening tools of PSCI published from the establishment of the database to December 2022 were searched in CNKI, VIP, Wanfang Data, SinoMed, PubMed, Embase, Web of Science, Cochrane Library. Two researchers respectively screened literatures, extracted data, and assessed the risk of bias. Stata 17.0 software was used to analyze the data. Results A total of 57 articles were included, involving 7 assessment tools [the National Institute of Neurological Disorders and Stroke-Canadian Stroke Network 5-Minute Battery (NINDS-CSN 5-Minutes), the Montreal Cognitive Assessment (MoCA), the Mini-Mental State Examination (MMSE), the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE), the Addenbrooke's Cognitive Examination-Revised (ACE-R), the Telephone Interview for Cognitive Status Modified (TICS-m) and the Montreal Cognitive Assessment 5-minute protocol (MoCA-5 min) ] to screen 12 113 patients. Meta-analysis results showed that the combined sensitivity and specificity of MoCA in screening PSCI were 0.84 (95%CI=0.80-0.87) and 0.74 (95%CI=0.67-0.80), respectively, with a combined area under the curve (AUC) of 0.87 (95%CI=0.84-0.90). The combined sensitivity and specificity of MMSE in screening PSCI were 0.73 (95%CI=0.67-0.79) and 0.76 (95%CI=0.69-0.82), respectively, with a combined AUC of 0.81 (95%CI=0.77-0.84). The combined sensitivity and specificity of IQCODE in screening PSCI were 0.73 (95%CI=0.48-0.89) and 0.95 (95%CI=0.75-0.99), respectively, with a combined AUC of 0.91 (95%CI=0.88-0.93). The combined sensitivity and specificity of the NINDS-CSN 5-min in screening PSCI were 0.83 (95%CI=0.78-0.87) and 0.69 (95%CI=0.60-0.76), respectively, with a combined AUC of 0.85 (95%CI=0.81-0.88). The combined sensitivity and specificity of the ACE-R in screening PSCI were 0.90 (95%CI=0.80-0.95) and 0.61 (95%CI=0.19-0.91), respectively, with a combined AUC of 0.90 (95%CI=0.87-0.92). The combined sensitivity and specificity of TICS-m in screening PSCI were 0.84 (95%CI=0.75-0.91) and 0.67 (95%CI=0.61-0.74), respectively, with a combined AUC of 0.66 (95%CI=0.60-0.71) . Conclusion The combined AUC of IQCODE and ACE-R is larger, and the former as a higher combined specificity and the latter has a higher combined sensitivity. Therefore, IQCODE and ACE-R are optimal assessment tools to accurately screen PSCI. Due to the limited number of literatures reporting the IQCODE and ACE-R in screening PSCI, our conclusions still need to be validated by multicenter and large-sample studies.

To critically appraise and quantify the performance studies by employing machine learning (ML) to predict delirium. A systematic review and meta-analysis. Articles reporting the use of ML to predict delirium in adult patients were included. Studies were excluded if (1) the primary goal was only the identification of various risk factors for delirium; (2) the full-text article was not found; and (3) the article was published in a language other than English/Chinese. PubMed, Embase, Cochrane Library database, Web of Science, Grey literature, and other relevant databases for the related publications were searched (from inception to December 15, 2021). The data were extracted using a standard checklist, and the risk of bias was assessed through the prediction model risk of bias assessment tool. Meta-analysis with the area under the receiver operating characteristic curve, sensitivity, and specificity as effect measures, was performed with Metadisc software. Cochran Q and I 2 statistics were used to assess the heterogeneity. Meta-regression was performed to determine the potential effect of adjustment for the key covariates. A total of 22 studies were included. Only 4 of 22 studies were quantitatively analyzed. The studies varied widely in reporting about the study participants, features and selection, handling of missing data, sample size calculations, and the intended clinical application of the model. For ML models, the overall pooled area under the receiver operating characteristic curve for predicting delirium was 0.89, sensitivity 0.85 (95% confidence interval 0.84‒0.85), and specificity 0.80 (95% confidence interval 0.81–0.80). We found that the ML model showed excellent performance in predicting delirium. This review highlights the potential shortcomings of the current approaches, including low comparability and reproducibility. Finally, we present the various recommendations on how these challenges can be effectively addressed before deploying these models in prospective analyses. [ABSTRACT FROM AUTHOR]