Skip to main content
Download PDF

Traditional and machine learning models for predicting haemorrhagic transformation in ischaemic stroke: a systematic review and meta-analysis

Systematic Reviews volume 14, Article number: 46 (2025) Cite this article

Abstract

Background

Haemorrhagic transformation (HT) is a severe complication after ischaemic stroke, but identifying patients at high risks remains challenging. Although numerous prediction models have been developed for HT following thrombolysis, thrombectomy, or spontaneous occurrence, a comprehensive summary is lacking. This study aimed to review and compare traditional and machine learning-based HT prediction models, focusing on their development, validation, and diagnostic accuracy.

Methods

PubMed and Ovid-Embase were searched for observational studies or randomised controlled trials related to traditional or machine learning-based models. Data were extracted according to Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies (CHARMS) checklist and risk of bias was assessed using the Prediction model Risk Of Bias ASsessment Tool (PROBAST). Performance data for prediction models that were externally validated at least twice and showed low risk of bias were meta-analysed.

Results

A total of 100 studies were included, with 67 focusing on model development and 33 on model validation. Among 67 model development studies, 44 were traditional model studies involving 47 prediction models (with National Institutes of Health Stroke Scale score being the most frequently used predictor in 35 models), and 23 studies focused on machine learning prediction models (with support vector machines being the most common algorithm, used in 10 models). The 33 validation studies externally validated 34 traditional prediction models. Regarding study quality, 26 studies were assessed as having a low risk of bias, 11 as unclear, and 63 as high risk of bias. Meta-analysis of 15 studies validating eight models showed a pooled area under the receiver operating characteristic curve of approximately 0.70 for predicting HT.

Conclusion

While significant progress has been made in developing HT prediction models, both traditional and machine learning-based models still have limitations in methodological rigour, predictive accuracy, and clinical applicability. Future models should undergo more rigorous validation, adhere to standardised reporting frameworks, and prioritise predictors that are both statistically significant and clinically meaningful. Collaborative efforts across research groups are essential for validating these models in diverse populations and improving their broader applicability in clinical practice.

Systematic review registration

International Prospective Register of Systematic Reviews (CRD42022332816).

Peer Review reports

Background

Stroke is the leading cause of death and disability in the world [1] and haemorrhagic transformation (HT) is a potentially devastating complication after acute ischaemic stroke [2]. HT may occur spontaneously during acute phase of stroke, or as a complication of interventions such as thrombectomy, thrombolysis, dual antiplatelet, and anticoagulation [3]. HT is associated with poor outcome after ischaemic stroke and contributes to the underutilisation of reperfusion therapies [4]. Identifying patients at high risk of HT has so far proved challenging [5]. Numerous prediction models have been developed to predict HT after thrombolysis [6], after thrombectomy, or spontaneously, but none has yet to be incorporated into consensus clinical guidelines because of their less than satisfactory performance [7]. With advancement in technology and medical informatics in recent decades, a large volume of ischaemic stroke data has been generated and stored in structured electronic formats worldwide, facilitating the use of artificial intelligence approaches such as machine learning for developing prediction models [8]. It is not immediately clear how these prediction models have been developed, whether they have been validated, or how they compare to each other. Given the emerging evidence, we aimed to perform a systematic review and meta-analysis to identify all traditional and machine learning models for predicting HT, describe their development and validation, and compare their diagnostic accuracies.

Methods

We reported this systematic review according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement (PRISMA) [9]. The study protocol was registered with PROSPERO International Prospective Register of Systematic Reviews under registration number CRD42022332816.

Literature searching

We systematically searched PubMed and Ovid-Embase for potentially eligible studies from database inception through March 1, 2022. This search was subsequently updated on October 16, 2023 and again October 31, 2024. Search terms included ischemic stroke, ischaemic stroke, haemorrhagic transformation, haemorrhagic transformation, intracerebral haemorrhage, intracerebral haemorrhage, prediction, predicting, predictive, score, and model. The reference lists of potentially eligible studies were manually checked to identify additional studies. Full search strategies were available in Supplement 1.

Eligibility criteria

We included observational studies or randomised controlled trials in Chinese or English that reported new models or validation of existing models to predict HT after ischaemic stroke, regardless of thrombolysis or thrombectomy. We excluded reviews, case studies, editorials, letters, and meeting abstracts. We also excluded original research studies if they only explored predictors of HT without constructing a formal model.

Study selection and quality assessment

Two reviewers (Zengyi Zhang and Zhimeng Zhang) independently screened the databases for eligible studies based on titles and abstracts, followed by reading of the full text. Disagreements about study inclusion were resolved through discussion with a third investigator (Yanan Wang or Junfeng Liu). The quality of included studies was assessed using the Prediction model Risk Of Bias ASsessment Tool (PROBAST) [10]. The risk of bias was assessed across four domains of PROBAST (participants, predictors, outcome, and analysis), while applicability was evaluated across three domains (participants, predictors, and outcome).

Data extraction

Two reviewers (Zengyi Zhang and Zhimeng Zhang) independently extracted data using a predefined form based on the Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies (CHARMS) checklist [11], which included information about study characteristics, predictors, and model development and validation, as follows: (1) Study characteristics: first author, publication year, study country, study population, study type, number of centres, source of data, outcome measure, and events. (2) Predictors: predictors were categorised into various domains, including demographics, clinical characteristics, laboratory examination, imaging findings, and genes. (3) Model development and validation: model name, modelling methods, calibration, discrimination, model validation, and prediction format. Modelling methods included logistic regression, literature review, and machine learning, etc. Calibration of models was conducted by Hosmer-Lemeshow goodness-of-fit test or calibration plots. Discrimination of models was evaluated using the area under the receiver operating characteristic curve (AUC). Model validation included internal and external validation. Presentations of models included risk scores, nomogram, and classifier-based framework, etc.

Outcome measure

The primary outcome measure was any HT, defined as the presence of haemorrhage within the infarct territory or as parenchymal haemorrhage outside the infarct zone. This haemorrhage was not visible on the initial CT or MRI scan after ischaemic stroke but was detected on follow-up imaging. Secondary outcome measures included radiological and clinical subtypes of HT. Radiological subtypes, classified according to criteria of European-Australian Cooperative Acute Stroke Study (ECASS) II, included haemorrhagic infarction (HI) and parenchymal haematoma (PH) [7]. Clinical subtypes consisted of symptomatic intracerebral haemorrhage (sICH) and asymptomatic intracerebral haemorrhage (aICH). The classification criteria for sICH included the criteria of National Institutes of Neurological Diseases and Stroke (NINDS), Safe Implementation of Thrombolysis in Stroke-Monitoring Study (SITS-MOST), ECASS II, ECASS III, the Third International Stroke Trial (IST-3), and Heidelberg Bleeding Classification (HBC) [12].

Statistical analysis

Qualitative data synthesis of prediction models

We categorised the included studies into two types: model development studies (which establish and validate prediction models) and validation studies (which conduct external validation of existing models). A descriptive analysis was then performed for each type. For model development studies, we summarised study population characteristics, types of predictors, types of outcome measures, model development methods, and presentation formats. For studies that use machine learning to develop prediction models, we summarised the machine learning algorithms. We also evaluated model performance, including discrimination and calibration. For validation studies, we focused on summarising study population characteristics, types of outcome measures, and model performance evaluation (including discrimination and calibration). Additionally, we conducted separate qualitative analyses of the risk of bias and applicability for each study type.

Quantitative analysis and comparison of the performance of prediction models

For validation studies, we conducted meta-analyses based on the type of prediction model and type of HT. However, three conditions must be met simultaneously: (1) The model has at least two external validation studies; (2) Risk of bias was assessed as low risk; (3) AUC and 95% confidence interval (CI) were provided. AUC values are interpreted as follows: values less than 0.6 indicate poor performance, 0.6–0.69 suggest moderate performance, 0.7–0.79 reflect good performance, 0.8–0.89 indicate very good performance, and values greater than 0.9 demonstrate excellent performance [13]. We calculated the pooled effect size using the random effects model and evaluated the heterogeneity using the I2 statistic, with thresholds of 25%, 50%, or 75% indicating low, moderate, or high heterogeneity, respectively. Publication bias was assessed through visual inspection of funnel plots. The impact of potential publication bias on pooled estimates was analysed using the ‘trim-and-fill’ method [14]. Data were analysed using Stata 18.0 (Stata Corp, College Station, TX, USA). P < 0.05 was considered statistically significant.

Results

We identified 12,335 articles in our initial database search in March 2022 and included 62 of these in our study. An additional 28 articles were included from an updated search in October 2023, and a final update on October 31, 2024, identified 10 articles for inclusion. Finally, 100 studies were included: 67 focused on model development [15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81] and 33 on model validation [82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114] (Fig. 1).

Fig. 1
figure 1

Flowchart of study selection

Full size image

Model development studies

Among the 67 model development studies, 44 studies developed 47 traditional prediction models, whereas 23 studies developed machine learning models. A total of 61 studies included both derivation and internal validation stages, while six studies focused exclusively on derivation. The study populations varied, with ischaemic stroke patients treated with thrombolysis comprising the largest group (n = 35), followed by those undergoing thrombectomy (n = 14), general ischaemic stroke patients (n = 11), without thrombolysis or thrombectomy (n = 5), and with thrombolysis or thrombectomy (n = 2). Centre settings included 38 multicentre studies, 27 single-centre studies, and two studies that did not specify centre status. Most studies assessed any HT (n = 39) as the outcome measure, with additional radiological subtypes, such as HI (n = 2) and PH (n = 6), and clinical subtypes of sICH (n = 37). Criteria for sICH classification included NINDS (n = 13), ECASS II (n = 16), and ECASS III (n = 5) (Table 1 and Supplemental Table 1).

Table 1 Summary of baseline characteristics of included studies
Full size table

Of the 44 traditional prediction model studies, logistic regression was the predominant method (40 studies), while three studies derived models from literature reviews, and one study did not specify the model development method (Supplemental Table 1). Among the 23 machine learning model studies, support vector machines were the most common algorithm (n = 10), followed by logistic regression (n = 8) and random forest (n = 7) (Supplemental Table 2).

The most frequent predictors in traditional models included the National Institutes of Health Stroke Scale score (NIHSS; n = 35), blood glucose (n = 23), age (n = 18), Alberta Stroke Program Early Computerised Tomography Score (n = 15), atrial fibrillation (n = 12), and systolic blood pressure (n = 11) (Fig. 2 and Supplemental Table 3). Presentation methods for traditional prediction models primarily included risk scores (n = 24) and nomograms (n = 17). All studies reported AUC for model discrimination, with 38 studies also reporting calibration results using the Hosmer–Lemeshow test (n = 13), calibration plots (n = 14), or both (n = 11) (Supplemental Table 1).

Fig. 2
figure 2

Main categories of predictors reported ≥ 5 times in traditional model development studies. ASPECTS, Alberta Stroke Program Early Computerised Tomography Score

Full size image

Risk of bias was assessed across the 67 studies, with 56 rated as high risk, 10 as unclear, and only one as low risk. Bias was primarily concentrated in the analysis domain, with 53 studies exhibiting high risk due to issues such as inadequate sample sizes, dichotomisation of continuous variables, improper handling of missing data, and reliance on univariate analysis for predictor selection. In the participant domain, 19 studies were rated as high risk, primarily due to reliance on retrospective data sources. All 67 studies demonstrated a low risk of bias in the predictor domain, and 66 showed a low risk of bias in the outcome domain. Applicability was generally favourable across participants, predictors, and outcomes (Supplemental Table 4).

Model validation studies

All 33 studies focused exclusively on external validation of traditional prediction models; no machine learning models were included. The primary study population was ischaemic stroke patients treated with thrombolysis (n = 23), with additional groups including general ischaemic stroke patients (n = 5), patients treated with thrombectomy (n = 4), and a small group with specific indications for anticoagulation (n = 1). Centre settings included 20 multicentre and 13 single-centre studies. The most frequently reported outcome measure was sICH (n = 29). Criteria for sICH classification included NINDS (n = 15), ECASS II (n = 16), and SITS-MOST (n = 12) (Table 1. and Supplemental Table 1). AUC for discrimination was reported in 31 studies, and calibration was reported in 11 studies, with eight using the Hosmer–Lemeshow test, two using calibration plots or curves, and one study using both (Supplemental Table 1).

Of the 33 validation studies, 25 were rated as low risk of bias, one as unclear, and seven as high risk. In the participant domain, three studies were rated as high risk, mainly due to reliance on retrospective data sources. All studies showed low risk of bias in the predictor and outcome domains. In the analysis domain, five studies were rated as high risk, with two studies having inadequate sample sizes (n < 100) and four not fully evaluating prediction model performance. Applicability was generally favourable across participants, predictors, and outcomes (Supplemental Table 4).

Meta-analysis of prediction model performance

A total of 15 studies [87, 89,90,91, 94, 96, 98, 99, 101, 103, 105, 107, 112,113,114] were included in the meta-analysis, encompassing the external validation of eight models. For predicting any HT, three studies that validated seven models identified the ‘Sugar-Early infarct signs-Dense artery sign-Age-NIHSS’ (SEDAN) score as having the highest discrimination (AUC 0.70, 95% CI 0.67–0.73; I2 = 0.06%; Fig. 3A). For predicting sICH per NINDS criteria, nine studies validating seven models showed that the ‘Haemorrhage After Thrombolysis’ (HAT) score and the ‘Glucose-Race-Age-Sex-Pressure-Stroke severity’ (GRASPS) achieved the best discrimination (AUC 0.69 Fig. 3B). For predicting sICH per SITS-MOST criteria, seven studies validating seven models found similar AUCs (around 0.68) for the HAT (Haemorrhage after Thrombolysis), GRASPS, Multicentre Stroke Survey (MSS), and Safe Implementation of Treatments in Stroke Symptomatic Intracerebral Haemorrhage Risk (SITS-SICH) scores (Fig. 3C). For predicting sICH ECASS II criteria, ten studies validating seven models demonstrated that the HAT and MSS score had the best discrimination (AUC 0.69 Fig. 3D). No publication bias was detected (Supplemental Fig. 1).

Fig. 3
figure 3

Meta-analysis of the area under the receiver operating characteristic curve of different models for predicting haemorrhagic transformation after acute ischaemic stroke, where such outcome was defined as any type of haemorrhagic transformation (A) or as symptomatic intracerebral haemorrhage diagnosed according to the criteria of the National Institutes of Neurological Diseases and Stroke (B), Safe Implementation of Thrombolysis in Stroke-Monitoring Study (C), or European-Australian Cooperative Acute Stroke Study II (D)

Full size image

Discussion

This systematic review and meta-analysis identified 47 traditional and 23 machine learning-based prediction models for HT after ischaemic stroke. Traditional models predominantly employed logistic regression, with key predictors such as NIHSS, blood glucose, and age being most commonly included. Among these traditional models, 34 were externally validated, and eight were validated at least twice in low-risk studies, achieving pooled AUCs of approximately 0.70. In contrast, machine learning models exhibited substantial variability in performance, with AUCs ranging from 0.42 to 0.99. Regarding study quality, 26 studies were assessed as having a low risk of bias, 11 as unclear, and 63 as high risk of bias.

Despite the promise of these models, the selection of predictors remains a critical challenge. Most traditional models rely heavily on statistical methods for predictor selection, often without considering clinical relevance, which may introduce selection bias [115]. Only a small subset of studies reviewed the literature to identify predictors that were both statistically and clinically significant [16, 22, 70]. Among the frequently used predictors in traditional models were NIHSS, blood glucose, and age, all of which have robust support in the literature for their association with HT risk [3, 7, 116,117,118]. Systolic blood pressure was another commonly included predictor and has shown a consistent relationship with HT risk [119,120,121,122]. Stroke type, such as those based on the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) classification and Oxfordshire community stroke project (OCSP) classification, was also frequently used [35, 45, 50, 63]; however, the accurate determination of stroke type shortly after admission remains challenging and limits their utility in clinical practice. Additionally, genetic and imaging predictors, including single-nucleotide polymorphisms [17, 110] and the Alberta Stroke Program Early Computerised Tomography Score, show promise, but are often impractical in the acute phase of stroke due to time and resource constraints. These challenges highlight the need for consensus on a core set of reliable and feasible predictors that can be standardised across models to enhance clinical utility.

Regarding the AUC performance of the models, traditional models showed an AUC around 0.70, indicating that these models have a moderate or good ability to predict HT. While this suggests that they offer reliable predictions for identifying high-risk patients, there is significant room for improvement in terms of predictive accuracy and clinical utility. In contrast, machine learning models exhibited a wide range of AUC values (0.42–0.99), reflecting substantial variability in performance. Some models demonstrated excellent performance (AUC > 0.9), while others had poor predictive value (AUC < 0.5). This variation could be attributed to differences in model type, the quality of data used, and the selection of predictors.

Although machine learning models have the potential to improve prediction accuracy, they still face challenges related to their clinical feasibility and generalisability, especially when incorporating complex data such as radiomic and other imaging data [18, 23, 24, 32, 44, 54, 55, 66, 68, 72, 73, 75,76,77, 80, 81]. The reliance on these data types introduces practical challenges, as the collection and processing of radiomic data can be time-consuming and resource-intensive. These factors limit the implementation of such models in acute stroke care. Therefore, future machine learning models should focus on simplifying input requirements without sacrificing predictive accuracy. Additionally, integrating diverse data sources—such as clinical, imaging, and laboratory variables—may improve model comprehensiveness but also presents challenges related to data harmonisation and standardisation.

Methodological weaknesses were evident across many of the included studies. A substantial proportion relied on retrospective data, which is associated with a higher risk of bias. Small sample sizes were also common, with many studies failing to meet the recommended sample size ratio for predictor-to-outcome events. The dichotomisation of continuous variables often led to the loss of valuable information, further increasing the risk of bias. Furthermore, many studies inadequately addressed missing data, either by excluding patients with incomplete data or using inappropriate imputation methods. Calibration assessments were often insufficient, with 51 studies failing to assess calibration altogether, and many that did used the Hosmer-Lemeshow test, which may not be suitable for complex models involving multiple predictors [123]. These methodological issues raise concerns about the reliability and generalisability of the models, while the PROBAST tool published in 2019 [10] has improved bias assessment, its inconsistent application in studies published after 2019 highlights the need for broader adoption of standardised reporting frameworks. Moreover, future studies should prioritise external validation cohorts to ensure that prediction models are robust and applicable across diverse clinical settings.

Future research should focus on addressing the methodological limitations identified in this review. Specifically, studies should prioritise prospective, multicentre studies with larger, more diverse sample sizes to improve generalisability. External validation cohorts should be used to ensure the reliability of prediction models across various clinical settings. Additionally, the selection and definition of key predictors, particularly stroke classification and imaging biomarkers, should be more consistent across studies. Simplifying prediction models, while maintaining predictive power, will be critical for enhancing their utility in acute stroke care. Furthermore, it is essential that future studies not only assess the accuracy of models but also their clinical feasibility, ensuring that models can be easily implemented in routine practice without requiring significant resources or specialised expertise.

There are several limitations to this systematic review and meta-analysis. First, in addition to AUC, other metrics like the net reclassification index, integrated discrimination index, net benefit, and decision curve analysis [124] can also assess model performance. However, these metrics were reported in only a few studies, so we did not conduct a pooled analysis for them. Second, there was significant heterogeneity in the meta-analysis results. This variability, due to differences in study populations, treatment protocols, and outcome definitions, limited the ability to perform meaningful subgroup analyses. Third, the variability in the predictors used in machine learning models and the lack of external validation for these models prevented their inclusion in the meta-analysis. Finally, we included only studies published in English and Chinese, which may have introduced language bias.

Conclusion

In conclusion, while substantial progress has been made in developing HT prediction models, both traditional and machine learning-based models still face significant limitations, particularly in terms of methodological rigour, predictive performance, and clinical applicability. To enhance their clinical utility, future models must undergo more rigorous validation, adhere to standardised reporting frameworks, and incorporate predictors that are both statistically significant and clinically meaningful. Collaborative efforts across research groups are essential to validate these models in diverse patient populations, ultimately improving their broader applicability in clinical practice.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Abbreviations

aICH:

Asymptomatic intracerebral haemorrhage

AUC:

Area under the receiver operating characteristic curve

CHARMS:

Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies

CI:

Confidence interval

ECASS:

European-Australian Cooperative Acute Stroke Study

GRASPS:

Glucose-Race-Age-Sex-Pressure-Stroke severity

HAT:

Haemorrhage After Thrombolysis

HBC:

Heidelberg Bleeding Classification

HI:

Haemorrhagic infarction

HT:

Haemorrhagic transformation

IST-3:

The Third International Stroke Trial

MSS:

Multicentre Stroke Survey

NIHSS:

National Institutes of Health Stroke Scale

NINDS:

National Institutes of Neurological Diseases and Stroke

OCSP:

Oxfordshire community stroke project

PH:

Parenchymal haematoma

PRISMA:

Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement

PROBAST:

Prediction model Risk Of Bias ASsessment Tool

SEDAN:

Sugar-Early infarct signs-Dense artery sign-Age-NIHSS

sICH:

Symptomatic intracerebral haemorrhage

SITS-MOST:

Safe Implementation of Thrombolysis in Stroke-Monitoring Study

SITS-SICH:

Safe Implementation of Treatments in Stroke Symptomatic Intracerebral Haemorrhage Risk

TOAST:

Trial of ORG 10172 in Acute Stroke Treatment

References

  1. Wu S, Liu M. Global burden of stroke: dynamic estimates to inform action. Lancet Neurol. 2024;23(10):952–3.

    Article  PubMed  Google Scholar 

  2. Álvarez-Sabín J, Maisterra O, Santamarina E, Kase CS. Factors influencing haemorrhagic transformation in ischaemic stroke. Lancet Neurol. 2013;12(7):689–705.

    Article  PubMed  Google Scholar 

  3. Chinese Society of Neurology, Chinese Stroke Society. Consensus on diagnosis and treatment of hemorrhagic transformation after acute ischemic stroke in China 2019. Chin J Neurol. 2019;52(4):252–65. [Chinese].

  4. Wang Y, Maeda T, You S, Chen C, Liu L, Zhou Z, et al. Patterns and Clinical Implications of Hemorrhagic Transformation After Thrombolysis in Acute Ischemic Stroke: Results From the ENCHANTED Study. Neurology. 2024;103(11):e210020.

  5. Wu S, Wu B, Liu M, Chen Z, Wang W, Anderson CS, et al. Stroke in China: advances and challenges in epidemiology, prevention, and management. Lancet Neurol. 2019;18(4):394–405.

    Article  PubMed  Google Scholar 

  6. Echouffo-Tcheugui JB, Woodward M, Kengne AP. Predicting a post-thrombolysis intracerebral hemorrhage: a systematic review. J Thromb Haemost. 2013;11(5):862–71.

    Article  CAS  PubMed  Google Scholar 

  7. Yaghi S, Willey JZ, Cucchiara B, Goldstein JN, Gonzales NR, Khatri P, et al. Treatment and outcome of hemorrhagic transformation after intravenous alteplase in acute ischemic stroke: a scientific statement for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2017;48(12):e343–61.

    Article  PubMed  Google Scholar 

  8. Obermeyer Z, Emanuel EJ. Predicting the future - big data, machine learning, and clinical medicine. N Engl J Med. 2016;375(13):1216–9.

    Article  PubMed  PubMed Central  Google Scholar 

  9. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71.

    Article  PubMed  PubMed Central  Google Scholar 

  10. Wolff RF, Moons KGM, Riley RD, Whiting PF, Westwood M, Collins GS, et al. PROBAST: a tool to assess the risk of bias and applicability of prediction model studies. Ann Intern Med. 2019;170(1):51–8.

    Article  PubMed  Google Scholar 

  11. Moons KG, de Groot JA, Bouwmeester W, Vergouwe Y, Mallett S, Altman DG, et al. Critical appraisal and data extraction for systematic reviews of prediction modelling studies: the CHARMS checklist. PLoS Med. 2014;11(10):e1001744.

    Article  PubMed  PubMed Central  Google Scholar 

  12. von Kummer R, Broderick JP, Campbell BC, Demchuk A, Goyal M, Hill MD, et al. The Heidelberg Bleeding Classification: Classification of Bleeding Events After Ischemic Stroke and Reperfusion Therapy. Stroke. 2015;46(10):2981–6.

    Article  Google Scholar 

  13. de Hond AAH, Steyerberg EW, van Calster B. Interpreting area under the receiver operating characteristic curve. Lancet Digit Health. 2022;4(12):e853–5.

    Article  PubMed  Google Scholar 

  14. Taylor S. Tweedie RLJAe. Practical estimates of the effect of publication bias in meta-analysis. Australas epidemiol;1998;5: 4-17.

    Google Scholar 

  15. Cucchiara B, Tanne D, Levine SR, Demchuk AM, Kasner S. A risk score to predict intracranial hemorrhage after recombinant tissue plasminogen activator for acute ischemic stroke. J Stroke Cerebrovasc Dis. 2008;17(6):331–3.

    Article  PubMed  Google Scholar 

  16. Lou M, Safdar A, Mehdiratta M, Kumar S, Schlaug G, Caplan L, et al. The HAT score: a simple grading scale for predicting hemorrhage after thrombolysis. Neurology. 2008;71(18):1417–23.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. del Río-Espínola A, Fernández-Cadenas I, Giralt D, Quiroga A, Gutiérrez-Agulló M, Quintana M, et al. A predictive clinical-genetic model of tissue plasminogen activator response in acute ischemic stroke. Ann Neurol. 2012;72(5):716–29.

    Article  PubMed  Google Scholar 

  18. Dharmasaroja P, Dharmasaroja PA. Prediction of intracerebral hemorrhage following thrombolytic therapy for acute ischemic stroke using multiple artificial neural networks. Neurol Res. 2012;34(2):120–8.

    Article  PubMed  Google Scholar 

  19. Mazya M, Egido JA, Ford GA, Lees KR, Mikulik R, Toni D, et al. Predicting the risk of symptomatic intracerebral hemorrhage in ischemic stroke treated with intravenous alteplase: Safe Implementation of Treatments in Stroke (SITS) symptomatic intracerebral hemorrhage risk score. Stroke. 2012;43(6):1524–31.

    Article  CAS  PubMed  Google Scholar 

  20. Menon BK, Saver JL, Prabhakaran S, Reeves M, Liang L, Olson DM, et al. Risk score for intracranial hemorrhage in patients with acute ischemic stroke treated with intravenous tissue-type plasminogen activator. Stroke. 2012;43(9):2293–9.

    Article  CAS  PubMed  Google Scholar 

  21. Strbian D, Engelter S, Michel P, Meretoja A, Sekoranja L, Ahlhelm FJ, et al. Symptomatic intracranial hemorrhage after stroke thrombolysis: the SEDAN score. Ann Neurol. 2012;71(5):634–41.

    Article  PubMed  Google Scholar 

  22. Saposnik G, Guzik AK, Reeves M, Ovbiagele B, Johnston SC. Stroke prognostication using age and NIH Stroke Scale: SPAN-100. Neurology. 2013;80(1):21–8.

    Article  PubMed  PubMed Central  Google Scholar 

  23. Scalzo F, Alger JR, Hu X, Saver JL, Dani KA, Muir KW, et al. Multi-center prediction of hemorrhagic transformation in acute ischemic stroke using permeability imaging features. Magn Reson Imaging. 2013;31(6):961–9.

    Article  PubMed  PubMed Central  Google Scholar 

  24. Bentley P, Ganesalingam J, Carlton Jones AL, Mahady K, Epton S, Rinne P, et al. Prediction of stroke thrombolysis outcome using CT brain machine learning. NeuroImage Clin. 2014;4:635–40.

    Article  PubMed  PubMed Central  Google Scholar 

  25. Whiteley WN, Thompson D, Murray G, Cohen G, Lindley RI, Wardlaw J, et al. Targeting recombinant tissue-type plasminogen activator in acute ischemic stroke based on risk of intracranial hemorrhage or poor functional outcome: an analysis of the third international stroke trial. Stroke. 2014;45(4):1000–6.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Asuzu D, Nyström K, Amin H, Schindler J, Wira C, Greer D, et al. TURN: a simple predictor of symptomatic intracerebral hemorrhage after IV thrombolysis. Neurocrit Care. 2015;23(2):166–71.

    Article  CAS  PubMed  Google Scholar 

  27. Kalinin MN, Khasanova DR, Ibatullin MM. The hemorrhagic transformation index score: a prediction tool in middle cerebral artery ischemic stroke. BMC Neurol. 2017;17(1):177.

    Article  PubMed  PubMed Central  Google Scholar 

  28. Lokeskrawee T, Muengtaweepongsa S, Patumanond J, Tiamkao S, Thamangraksat T, Phankhian P, et al. Prediction of symptomatic intracranial hemorrhage after intravenous thrombolysis in acute ischemic stroke: the symptomatic intracranial hemorrhage score. J Stroke Cerebrovasc Dis. 2017;26(11):2622–9.

    Article  PubMed  Google Scholar 

  29. Yeo LLL, Chien SC, Lin JR, Liow CW, Lee JD, Peng TI, et al. Derivation and validation of a scoring system for intravenous tissue plasminogen activator use in Asian patients. J Stroke Cerebrovasc Dis. 2017;26(8):1695–703.

    Article  PubMed  Google Scholar 

  30. Cappellari M, Turcato G, Forlivesi S, Zivelonghi C, Bovi P, Bonetti B, et al. STARTING-SICH nomogram to predict symptomatic intracerebral hemorrhage after intravenous thrombolysis for stroke. Stroke. 2018;49(2):397–404.

    Article  PubMed  Google Scholar 

  31. Erdur H, Polymeris A, Grittner U, Scheitz JF, Tutuncu S, Seiffge DJ, et al. A score for risk of thrombolysis-associated hemorrhage including pretreatment with statins. Front Neurol. 2018;9:74.

  32. Yu Y, Guo D, Lou M, Liebeskind D, Scalzo F. Prediction of hemorrhagic transformation severity in acute stroke from source perfusion MRI. IEEE Trans Biomed Eng. 2018;65(9):2058–65.

    Article  PubMed  Google Scholar 

  33. Cappellari M, Mangiafico S, Saia V, Pracucci G, Nappini S, Nencini P, et al. IER-SICH nomogram to predict symptomatic intracerebral hemorrhage after thrombectomy for stroke. Stroke. 2019;50(4):909–16.

    Article  PubMed  Google Scholar 

  34. Montalvo M, Mistry E, Chang AD, Yakhkind A, Dakay K, Azher I, et al. Predicting symptomatic intracranial haemorrhage after mechanical thrombectomy: the TAG score. J Neurol Neurosurg Psychiatry. 2019;90(12):1370–4.

    PubMed  Google Scholar 

  35. de Andrade JBC, Mohr JP, Lima FO, Carvalho JJF, de Farias VAE, Oliveira-Filho J, et al. Predicting hemorrhagic transformation in patients not submitted to reperfusion therapies. J Stroke Cerebrovasc Dis. 2020;29(8):104940.

  36. Wang F, Huang Y, Xia Y, Zhang W, Fang K, Zhou X, et al. Personalized risk prediction of symptomatic intracerebral hemorrhage after stroke thrombolysis using a machine-learning model. Ther Adv Neurol Disord. 2020;13:1756286420902358.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  37. Wang Q, Reps JM, Kostka KF, Ryan PB, Zou Y, Voss EA, et al. Development and validation of a prognostic model predicting symptomatic hemorrhagic transformation in acute ischemic stroke at scale in the OHDSI network. PLoS One. 2020;15(1):e0226718.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  38. Wu Y, Chen H, Liu X, Cai X, Kong Y, Wang H, et al. A new nomogram for individualized prediction of the probability of hemorrhagic transformation after intravenous thrombolysis for ischemic stroke patients. BMC Neurol. 2020;20(1):426.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  39. Zhang WB, Zeng YY, Wang F, Cheng L, Tang WJ, Wang XQ. A high neutrophil-to-lymphocyte ratio predicts hemorrhagic transformation of large atherosclerotic infarction in patients with acute ischemic stroke. Aging. 2020;12(3):2428–39.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  40. Zhang X, Xie Y, Wang H, Yang D, Jiang T, Yuan K, et al. Symptomatic intracranial hemorrhage after mechanical thrombectomy in Chinese ischemic stroke patients: the ASIAN score. Stroke. 2020;51(9):2690–6.

  41. Zhou Z, Yin X, Niu Q, Liang S, Mu C, Zhang Y. Risk factors and a nomogram for predicting intracranial hemorrhage in stroke patients undergoing thrombolysis. Neuropsychiatr Dis Treat. 2020;16:1189–97.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  42. Choi JM, Seo SY, Kim PJ, Kim YS, Lee SH, Sohn JH, et al. Prediction of hemorrhagic transformation after ischemic stroke using machine learning. J Pers Med. 2021;11(9):863.

  43. Guo H, Xu W, Zhang X, Zhang S, Dai Z, Li S, et al. A nomogram to predict symptomatic intracranial hemorrhage after intravenous thrombolysis in Chinese patients. Neuropsychiatr Dis Treat. 2021;17:2183–90.

    Article  PubMed  PubMed Central  Google Scholar 

  44. Jiang L, Zhou L, Yong W, Cui J, Geng W, Chen H, et al. A deep learning-based model for prediction of hemorrhagic transformation after stroke. Brain Pathol. 2023;33(2):e13023.

  45. Liu J, Wang Y, Jin Y, Guo W, Song Q, Wei C, et al. Prediction of hemorrhagic transformation after ischemic stroke: development and validation study of a novel multi-biomarker model. Front Aging Neurosci. 2021;13:667934.

    Article  PubMed  PubMed Central  Google Scholar 

  46. Soni M, Wijeratne T, Ackland DC. A risk score for prediction of symptomatic intracerebral haemorrhage following thrombolysis. Int J Med Inform. 2021;156:104586.

    Article  Google Scholar 

  47. Wei C, Liu J, Guo W, Jin Y, Song Q, Wang Y, et al. Development and validation of a predictive model for spontaneous hemorrhagic transformation after ischemic stroke. Front Neurol. 2021;12:747026.

    Article  PubMed  PubMed Central  Google Scholar 

  48. Yu X, Pan J, Zhao X, Hou Q, Liu B. Predicting hemorrhagic transformation after thrombectomy in acute ischemic stroke: a multimodal score of the regional pial collateral. Neuroradiology. 2022;64(3):493–502.

    Article  PubMed  Google Scholar 

  49. Zhang XX, Yao FR, Zhu JH, Chen ZG, Shen YP, Qiao YN, et al. Nomogram to predict haemorrhagic transformation after stroke thrombolysis: a combined brain imaging and clinical study. Clin Radiol. 2022;77(1):e92–8.

    Article  PubMed  Google Scholar 

  50. Xie X, Yang J, Ren L, Hu S, Lian W, Xiao J, et al. Nomogram to predict symptomatic intracranial hemorrhage after intravenous thrombolysis in acute ischemic stroke in Asian population. Curr Neurovasc Res. 2021;18(5):543–51.

    Article  CAS  PubMed  Google Scholar 

  51. Elsaid AF, Fahmi RM, Shehta N, Ramadan BM. Machine learning approach for hemorrhagic transformation prediction: capturing predictors’ interaction. Front Neurol. 2022;13:951401.

    Article  PubMed  PubMed Central  Google Scholar 

  52. Janvier P, Kerleroux B, Turc G, Pasi M, Farhat W, Bricout N, et al. TAGE score for symptomatic intracranial hemorrhage prediction after successful endovascular treatment in acute ischemic stroke. Stroke. 2022;53(9):2809–17.

    Article  CAS  PubMed  Google Scholar 

  53. Kang Z, Nie C, Ouyang K, Wu X, Yin J, Sun D, et al. A nomogram for predicting symptomatic intracranial hemorrhage after endovascular thrombectomy. Clin Neurol Neurosurg. 2022;218:107298.

    Article  PubMed  Google Scholar 

  54. Liu J, Chen X, Guo X, Xu R, Wang Y, Liu M. Machine learning prediction of symptomatic intracerebral hemorrhage after stroke thrombolysis: a cross-cultural validation in Caucasian and Han Chinese cohort. Ther Adv Neurol Disord. 2022;15:17562864221129380.

  55. Meng Y, Wang H, Wu C, Liu X, Qu L, Shi Y. Prediction model of hemorrhage transformation in patient with acute ischemic stroke based on multiparametric MRI radiomics and machine learning. Brain Sci. 2022;12(7):858.

  56. Ren Y, He Z, Du X, Liu J, Zhou L, Bai X, et al. The SON(2)A(2) score: a novel grading scale for predicting hemorrhage and outcomes after thrombolysis. Front Neurol. 2022;13:952843.

    Article  PubMed  PubMed Central  Google Scholar 

  57. Weng ZA, Huang XX, Deng D, Yang ZG, Li SY, Zang JK, et al. A new nomogram for predicting the risk of intracranial hemorrhage in acute ischemic stroke patients after intravenous thrombolysis. Front Neurol. 2022;13:774654.

    Article  PubMed  PubMed Central  Google Scholar 

  58. Xu Y, Li X, Wu D, Zhang Z, Jiang A. Machine learning-based model for prediction of hemorrhage transformation in acute ischemic stroke after alteplase. Front Neurol. 2022;13: 897903.

    Article  PubMed  PubMed Central  Google Scholar 

  59. Yang M, Zhong W, Zou W, Peng J, Tang X. A novel nomogram to predict hemorrhagic transformation in ischemic stroke patients after intravenous thrombolysis. Front Neurol. 2022;13:913442.

  60. Zhang K, Luan J, Li C, Chen M. Nomogram to predict hemorrhagic transformation for acute ischemic stroke in Western China: a retrospective analysis. BMC Neurol. 2022;22(1):156.

  61. Chang GC, Nguyen TN, Qiu J, Li W, Zhao YG, Sun XH, et al. Predicting symptomatic intracranial hemorrhage in anterior circulation stroke patients with contrast enhancement after thrombectomy: the CAGA score. J Neurointerv Surg. 2023;15(e3):e356–62.

  62. Elsaid N, Bigliardi G, Dell’Acqua ML, Vandelli L, Ciolli L, Picchetto L, et al. Proposal of multimodal computed tomography-based scoring system in prediction of hemorrhagic transformation in acute ischemic stroke. Acta Neurol Belg. 2023;123(4):1405–11.

    Article  PubMed  Google Scholar 

  63. Grifoni E, Bini C, Signorini I, Cosentino E, Micheletti I, Dei A, et al. Predictive factors for hemorrhagic transformation in acute ischemic stroke in the real-world clinical practice. Neurologist. 2023;28(3):150–6.

    PubMed  Google Scholar 

  64. Lei C, Li Y, Zhou X, Lin S, Zhu X, Yang X, et al. A simple grading scale for predicting symptomatic intracranial hemorrhage after mechanical thrombectomy. Cerebrovasc Dis. 2023;52(4):401–8.

    Article  CAS  PubMed  Google Scholar 

  65. Li X, Xu C, Shang C, Wang Y, Xu J, Zhou Q. Machine learning predicts the risk of hemorrhagic transformation of acute cerebral infarction and in-hospital death. Comput Methods Programs Biomed. 2023;237:107582.

    Article  PubMed  Google Scholar 

  66. Ren H, Song H, Wang J, Xiong H, Long B, Gong M, et al. A clinical-radiomics model based on noncontrast computed tomography to predict hemorrhagic transformation after stroke by machine learning: a multicenter study. Insights Imaging. 2023;14(1):52.

    Article  PubMed  PubMed Central  Google Scholar 

  67. Shen Y, Xiong Y, Cao Q, Li Y, Xiang W, Wang L, et al. Construction and validation of a nomogram model to predict symptomatic intracranial hemorrhage after intravenous thrombolysis in severe white matter lesions. J Thromb Thrombolysis. 2023;56(1):111–20.

    Article  CAS  PubMed  Google Scholar 

  68. Wen X, Xiao Y, Hu X, Chen J, Song F. Prediction of hemorrhagic transformation via pre-treatment CT radiomics in acute ischemic stroke patients receiving endovascular therapy. Br J Radiol. 2023;96(1147):20220439.

    Article  PubMed  PubMed Central  Google Scholar 

  69. Sung SF, Chen SC, Lin HJ, Chen CH, Tseng MC, Wu CS, et al. Oxfordshire community stroke project classification improves prediction of post-thrombolysis symptomatic intracerebral hemorrhage. BMC Neurol. 2014;14:39.

    Article  PubMed  PubMed Central  Google Scholar 

  70. Lee JS, Kim CK, Kang J, Park JM, Park TH, Lee KB, et al. A novel computerized clinical decision support system for treating thrombolysis in patients with acute ischemic stroke. J Stroke. 2015;17(2):199–209.

    Article  PubMed  PubMed Central  Google Scholar 

  71. Qian Y, Qian ZT, Huang CH, Wang HY, Lu X, Cao K, et al. Predictive factors and nomogram to evaluate the risk of symptomatic intracerebral hemorrhage for stroke patients receiving thrombectomy. World Neurosurg. 2020;144:e466–74.

    Article  PubMed  Google Scholar 

  72. Chung CC, Chan L, Bamodu OA, Hong CT, Chiu HW. Artificial neural network based prediction of postthrombolysis intracerebral hemorrhage and death. Sci Rep. 2020;10(1):20501.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  73. Cui S, Song H, Ren H, Wang X, Xie Z, Wen H, et al. Prediction of hemorrhagic complication after thrombolytic therapy based on multimodal data from multiple centers: an approach to machine learning and system implementation. J Pers Med. 2022;12(12):2052.

  74. Duan Q, Li W, Zhang Y, Zhuang W, Long J, Wu B, et al. Nomogram established on account of Lasso-logistic regression for predicting hemorrhagic transformation in patients with acute ischemic stroke after endovascular thrombectomy. Clin Neurol Neurosurg. 2024;243:108389.

    Article  PubMed  Google Scholar 

  75. Heo J, Sim Y, Kim BM, Kim DJ, Kim YD, Nam HS, et al. Radiomics using non-contrast CT to predict hemorrhagic transformation risk in stroke patients undergoing revascularization. Eur Radiol. 2024;34(9):6005–15.

    Article  PubMed  Google Scholar 

  76. Heo J, Yoon Y, Han HJ, Kim JJ, Park KY, Kim BM, et al. Prediction of cerebral hemorrhagic transformation after thrombectomy using a deep learning of dual-energy CT. Eur Radiol. 2024;34(6):3840–8.

    Article  PubMed  Google Scholar 

  77. Huang YH, Chen ZJ, Chen YF, Cai C, Lin YY, Lin ZQ, et al. The value of CT-based radiomics in predicting hemorrhagic transformation in acute ischemic stroke patients without recanalization therapy. Front Neurol. 2024;15:1255621.

    Article  PubMed  PubMed Central  Google Scholar 

  78. Kuang Y, Zhang L, Ye K, Jiang Z, Shi C, Luo L. Clinical and imaging predictors for hemorrhagic transformation of acute ischemic stroke after endovascular thrombectomy. J Neuroimaging. 2024;34(3):339–47.

    Article  PubMed  Google Scholar 

  79. Ma Y, Xu DY, Liu Q, Chen HC, Chai EQ. Nomogram prediction model for the risk of intracranial hemorrhagic transformation after intravenous thrombolysis in patients with acute ischemic stroke. Front Neurol. 2024;15:1361035.

    Article  PubMed  PubMed Central  Google Scholar 

  80. Zhang Y, Xie G, Zhang L, Li J, Tang W, Wang D, et al. Constructing machine learning models based on non-contrast CT radiomics to predict hemorrhagic transformation after stoke: a two-center study. Front Neurol. 2024;15:1413795.

    Article  PubMed  PubMed Central  Google Scholar 

  81. Ren H, Song H, Liu J, Cui S, Gong M, Li Y. Deep learning using one-stop-shop CT scan to predict hemorrhagic transformation in stroke patients undergoing reperfusion therapy: a multicenter study. Acad Radiol. 2024. https://doiorg.publicaciones.saludcastillayleon.es/10.1016/j.acra.2024.09.052.

  82. Cucchiara B, Kasner S, Tanne D, Levine S, Demchuk A, Messe S, et al. Validation assessment of risk scores to predict postthrombolysis intracerebral haemorrhage. Int J Stroke. 2011;6(2):109–11.

    Article  PubMed  Google Scholar 

  83. Saposnik G, Fang J, Kapral MK, Tu JV, Mamdani M, Austin P, et al. The iScore predicts effectiveness of thrombolytic therapy for acute ischemic stroke. Stroke. 2012;43(5):1315–22.

    Article  PubMed  Google Scholar 

  84. Flint AC, Faigeles BS, Cullen SP, Kamel H, Rao VA, Gupta R, et al. THRIVE score predicts ischemic stroke outcomes and thrombolytic hemorrhage risk in VISTA. Stroke. 2013;44(12):3365–9.

    Article  CAS  PubMed  Google Scholar 

  85. Mazya MV, Bovi P, Castillo J, Jatuzis D, Kobayashi A, Wahlgren N, et al. External validation of the SEDAN score for prediction of intracerebral hemorrhage in stroke thrombolysis. Stroke. 2013;44(6):1595–600.

    Article  CAS  PubMed  Google Scholar 

  86. Saposnik G, Demchuk A, Tu JV, Johnston SC; Stroke Outcomes Research Canada (SORCan) Working Group. The iScore predicts efficacy and risk of bleeding in the national institute of neurological disorders and stroke tissue plasminogen activator stroke trial. J Stroke Cerebrovasc Dis. 2013;22(6):876–82.

    Article  PubMed  Google Scholar 

  87. Sung SF, Chen SCC, Lin HJ, Chen YW, Tseng MC, Chen CH. Comparison of risk-scoring systems in predicting symptomatic intracerebral hemorrhage after intravenous thrombolysis. Stroke. 2013;44(6):1561–6.

    Article  CAS  PubMed  Google Scholar 

  88. Boehme AK, Rawal PV, Lyerly MJ, Albright KC, Bavarsad Shahripour R, Palazzo P, et al. Investigating the utility of previously developed prediction scores in acute ischemic stroke patients in the stroke belt. J Stroke Cerebrovasc Dis. 2014;23(8):2001–6.

    Article  PubMed  PubMed Central  Google Scholar 

  89. Flint AC, Gupta R, Smith WS, Kamel H, Faigeles BS, Cullen SP, et al. The THRIVE score predicts symptomatic intracerebral hemorrhage after intravenous tPA administration in SITS-MOST. Int J Stroke. 2014;9(6):705–10.

    Article  PubMed  Google Scholar 

  90. Strbian D, Michel P, Seiffge DJ, Saver JL, Numminen H, Meretoja A, et al. Symptomatic intracranial hemorrhage after stroke thrombolysis: comparison of prediction scores. Stroke. 2014;45(3):752–8.

    Article  CAS  PubMed  Google Scholar 

  91. Van Hooff RJ, Nieboer K, De Smedt A, Moens M, De Deyn PP, De Keyser J, et al. Validation assessment of risk tools to predict outcome after thrombolytic therapy for acute ischemic stroke. Clin Neurol Neurosurg. 2014;125:189–93.

    Article  PubMed  Google Scholar 

  92. von Klemperer A, Bateman K, Owen J, Bryer A. Thrombolysis risk prediction: applying the SITS-SICH and SEDAN scores in South African patients. Cardiovasc J Afr. 2014;25(5):224–7.

    Article  Google Scholar 

  93. Asuzu D, Nystrom K, Amin H, Schindler J, Wira C, Greer D, et al. Comparison of 8 scores for predicting symptomatic intracerebral hemorrhage after IV thrombolysis. Neurocrit Care. 2015;22(2):229–33.

    Article  CAS  PubMed  Google Scholar 

  94. Li M, Wang-Qin RQ, Wang YL, Liu LB, Pan YS, Liao XL, et al. Symptomatic intracerebral hemorrhage after intravenous thrombolysis in Chinese patients: comparison of prediction models. J Stroke Cerebrovasc Dis. 2015;24(6):1235–43.

    Article  PubMed  Google Scholar 

  95. Muengtaweepongsa S, Prapa-Anantachai P, Dharmasaroja PA, Rukkul P, Yodvisitsak P. External validation of the SEDAN score: the real world practice of a single center. Ann Indian Acad Neurol. 2015;18(2):181–6.

    Article  PubMed  PubMed Central  Google Scholar 

  96. Watson-Fargie T, Dai D, MacLeod MJ, Reid JM. Comparison of predictive scores of symptomatic intracerebral haemorrhage after stroke thrombolysis in a single centre. J R Coll Physicians Edinb. 2015;45(2):127–32.

    Article  CAS  PubMed  Google Scholar 

  97. Wei H, Yu YF, Zhou R, Yin HX, Du JC, Yang X, et al. Clinical study on HAT and SEDAN score scales and related risk factors for predicting hemorrhagic transformation following thrombolysis in acute ischemic stroke. Chin J Contemp Neurol Neurosurg. 2015;15(2):126–32. [Chinese].

    Google Scholar 

  98. Al-Khaled M, Langner B, Brüning T. Predicting risk of symptomatic intracerebral hemorrhage and mortality after treatment with recombinant tissue-plasminogen activator using SEDAN score. Acta Neurol Scand. 2016;133(4):239–44.

    Article  CAS  PubMed  Google Scholar 

  99. Asuzu D, Nystrom K, Amin H, Schindler J, Wira C, Greer D, et al. Validation of TURN, a simple predictor of symptomatic intracerebral hemorrhage after IV thrombolysis. Clin Neurol Neurosurg. 2016;146:71–5.

    Article  PubMed  Google Scholar 

  100. Marsh EB, Llinas RH, Schneider ALC, Hillis AE, Lawrence E, Dziedzic P, et al. Predicting hemorrhagic transformation of acute ischemic stroke: prospective validation of the HeRS score. Medicine. 2016;95(2):e2430.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  101. Li F, Wang J, Peng F, Xiao L, Sun W, Liu X. Comparison of risk models in predicting intracranial hemorrhage and poor outcomes in acute anterior circulation ischemic stroke after mechanical thrombectomy. Chin J Cerebrovasc Dis. 2017;14(4):175–82. [Chinese].

    Google Scholar 

  102. Mobius C, Blinzler C, Schwab S, Kohrmann M, Breuer L. Re-evaluation of the stroke prognostication using age and NIH Stroke Scale index (SPAN-100 index) in IVT patients - the-SPAN 10065 index. BMC Neurol. 2018;18(1):129.

  103. Pan Y, Peng Y, Chen W, Wang Y, Lin Y, He Y, et al. THRIVE-c score predicts clinical outcomes in Chinese patients after thrombolysis. Brain Behav. 2018;8(2):e00927.

  104. Nisar T, Hanumanthu R, Khandelwal P. Symptomatic intracerebral hemorrhage after intravenous thrombolysis: predictive factors and validation of prediction models. J Stroke Cerebrovasc Dis. 2019;28(11):104360.

    Article  PubMed  Google Scholar 

  105. Miao CQ, Yin XY, Liang SM, Mu CY, Zhang YR. MSS score and atrial fibrillation are the risk factors of intracranial hemorrhage in stroke patients. Acta Medica Mediterr. 2020;36(6):3295–300.

    Google Scholar 

  106. Chang X, Zhang X, Zhang G. Different scores predict the value of hemorrhagic transformation after intravenous thrombolysis in patients with acute ischemic stroke. Evid Based Complement Alternat Med. 2021;2021:2468052.

    Article  PubMed  PubMed Central  Google Scholar 

  107. Fu CH, Chen CH, Lin CH, Lee CW, Lee M, Tang SC, et al. Comparison of risk scores in predicting symptomatic intracerebral hemorrhage after endovascular thrombectomy. J Formos Med Assoc. 2022;121(7):1257–65.

  108. Lei C, Wu B, Liu M, Chen Y, Yang H, Wang D, et al. Totaled health risks in vascular events score predicts clinical outcomes in patients with cardioembolic and other subtypes of ischemic stroke. Stroke. 2014;45(6):1689–94.

    Article  PubMed  Google Scholar 

  109. Chen W, Pan Y, Zhao X, Liao X, Liu L, Wang C, et al. Totaled health risks in vascular events score predicts clinical outcome and symptomatic intracranial hemorrhage in Chinese patients after thrombolysis. Stroke. 2015;46(3):864–6.

    Article  PubMed  Google Scholar 

  110. Carrera C, Cullell N, Torres-Águila N, Muiño E, Bustamante A, Dávalos A, et al. Validation of a clinical-genetics score to predict hemorrhagic transformations after rtPA. Neurology. 2019;93(9):e851–63.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  111. Alsbrook DL, Heiferman DM, Dornbos D, Inoa V, Hoit D, Krishnaiah B, et al. External validation of TICI-ASPECTS-glucose score as a predictive model for symptomatic intracranial hemorrhage following mechanical thrombectomy. J Stroke Cerebrovasc Dis. 2022;31(11):106796.

    Article  Google Scholar 

  112. de Andrade JBC, Mohr JP, Ahmad M, Lima FO, Barros LCM, Silva GS. Accuracy of predictive scores of hemorrhagic transformation in patients with acute ischemic stroke. Arq Neuropsiquiatr. 2022;80(5):455–61.

    Article  PubMed  PubMed Central  Google Scholar 

  113. Van Der Ende NAM, Kremers FCC, Van Der Steen W, Venema E, Kappelhof M, Majoie CBLM, et al. Symptomatic intracranial hemorrhage after endovascular stroke treatment: external validation of prediction models. Stroke. 2023;54(2):476–87.

    Article  PubMed  PubMed Central  Google Scholar 

  114. Wang Y, Liu J, Wu Q, Cheng Y, Liu M, Collaborators V. Validation and comparison of multiple risk scores for prediction of symptomatic intracerebral hemorrhage after intravenous thrombolysis in VISTA. Int J Stroke. 2023;18(3):338–45.

    Article  PubMed  Google Scholar 

  115. Sun GW, Shook TL, Kay GL. Inappropriate use of bivariable analysis to screen risk factors for use in multivariable analysis. J Clin Epidemiol. 1996;49(8):907–16.

    Article  CAS  PubMed  Google Scholar 

  116. Whiteley WN, Slot KB, Fernandes P, Sandercock P, Wardlaw J. Risk factors for intracranial hemorrhage in acute ischemic stroke patients treated with recombinant tissue plasminogen activator: a systematic review and meta-analysis of 55 studies. Stroke. 2012;43(11):2904–9.

    Article  CAS  PubMed  Google Scholar 

  117. Wang R, Zeng J, Wang F, Zhuang X, Chen X, Miao J. Risk factors of hemorrhagic transformation after intravenous thrombolysis with rt-PA in acute cerebral infarction. QJM. 2019;112(5):323–6.

    Article  CAS  PubMed  Google Scholar 

  118. Kidwell CS, Saver JL, Carneado J, Sayre J, Starkman S, Duckwiler G, et al. Predictors of hemorrhagic transformation in patients receiving intra-arterial thrombolysis. Stroke. 2002;33(3):717–24.

    Article  PubMed  Google Scholar 

  119. Butcher K, Christensen S, Parsons M, De Silva DA, Ebinger M, Levi C, et al. Postthrombolysis blood pressure elevation is associated with hemorrhagic transformation. Stroke. 2010;41(1):72–7.

    Article  PubMed  Google Scholar 

  120. Ahmed N, Wahlgren N, Brainin M, Castillo J, Ford GA, Kaste M, et al. Relationship of blood pressure, antihypertensive therapy, and outcome in ischemic stroke treated with intravenous thrombolysis: retrospective analysis from Safe Implementation of Thrombolysis in Stroke-International Stroke Thrombolysis Register (SITS-ISTR). Stroke. 2009;40(7):2442–9.

    Article  PubMed  Google Scholar 

  121. Wang X, Minhas JS, Moullaali TJ, Di Tanna GL, Lindley RI, Chen X, et al. Associations of early systolic blood pressure control and outcome after thrombolysis-eligible acute ischemic stroke: results from the ENCHANTED study. Stroke. 2022;53(3):779–87.

    Article  PubMed  Google Scholar 

  122. Yong M, Kaste M. Association of characteristics of blood pressure profiles and stroke outcomes in the ECASS-II trial. Stroke. 2008;39(2):366–72.

    Article  PubMed  Google Scholar 

  123. Huang Y, Li W, Macheret F, Gabriel RA, Ohno-Machado L. A tutorial on calibration measurements and calibration models for clinical prediction models. J Am Med Inform Assoc. 2020;27(4):621–33.

    Article  PubMed  PubMed Central  Google Scholar 

  124. Steyerberg EW, Vickers AJ, Cook NR, Gerds T, Gonen M, Obuchowski N, et al. Assessing the performance of prediction models: a framework for traditional and novel measures. Epidemiology. 2010;21(1):128–38.

    Article  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

None.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the National Natural Science Foundation of China (grant No. 82371323 and 81974181), the Sichuan Science and Technology Program (2023NSFSC1558), the Major International (Regional) Joint Research Project, National Natural Science Foundation of China (81620108009), and the Postdoctor Research Fund of West China Hospital, Sichuan University (2024HXBH139). The funders have not in any way been involved in study design, data collection, data analysis, manuscript preparation, and/or publication decision.

Author information

Author notes

  1. Yanan Wang, Zengyi Zhang and Zhimeng Zhang contributed equally to this work.

Authors and Affiliations

  1. Department of Neurology, West China Hospital, Sichuan University, No. 37 Guo Xue Xiang, Chengdu, Sichuan, 610041, China

    Yanan Wang, Junfeng Liu & Ming Liu

  2. West China School of Medicine, Sichuan University, Chengdu, Sichuan, China

    Zengyi Zhang & Zhimeng Zhang

  3. Faculty of Medicine, The George Institute for Global Health, University of New South Wales, Sydney, Australia

    Xiaoying Chen

  4. Centre of Cerebrovascular Diseases, West China Hospital, Sichuan University, Chengdu, Sichuan, China

    Junfeng Liu & Ming Liu

Authors
  1. Yanan Wang
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Zengyi Zhang
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Zhimeng Zhang
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Xiaoying Chen
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Junfeng Liu
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Ming Liu
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

JL and ML researched literature and conceived the study. YW, ZZ, ZZ, XC, and JL were involved in protocol development, literature search data extraction, and data analysis. YW and JL wrote the first draft of the manuscript. All authors reviewed and edited the manuscript and approved the final version of the manuscript.

Corresponding authors

Correspondence to Junfeng Liu or Ming Liu.

Ethics declarations

Ethics approval and consent to participate

As no individual patient-level data were used, institutional review board approval and informed consent were not required.

Consent for publication

Not applicable.

Competing interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Wang, Y., Zhang, Z., Zhang, Z. et al. Traditional and machine learning models for predicting haemorrhagic transformation in ischaemic stroke: a systematic review and meta-analysis. Syst Rev 14, 46 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13643-025-02771-w

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13643-025-02771-w

Keywords

Systematic Reviews

ISSN: 2046-4053