Effects of cerebral infarction on cognitive and speech performance Authors: Yang Niu, Mingming Hu, Xiaoyan Zhao, Xiaoping Gao, Yongmei Zhu Published: 2026 ABSTRACT AbstractObjectiveThe present study sought to identify the cognitive features and speech features that might distinguish cerebral infarction patients by using artificial intelligence.MethodsA total of 117 patients were divided into the lacunar group, non-lacunar cerebral infarction group, and control group from Anhui No.2 Provincial People’s Hospital. The cognitive features were created from the Chinese version of Montreal Cognitive Assessment Basic (MoCA-B) and the speech features were created from the cookie theft picture description task.ResultsCerebral infarction patients, including lacunar and non-lacunar groups, had lower cognitive scores than the control group, with the non-lacunar group performing more poorly. The fluency task showed significant differences among the three groups, with the non-lacunar cerebral infarction group differing most from the control group. Non-lacunar and lacunar cerebral infarction groups performed worse than the control group in memory tasks, with the non-lacunar infarction group performing more poorly. Artificial intelligence technology was used to analyze speech data from patients completing tasks, revealing differences in acoustic features between the non-lacunar and lacunar cerebral infarction groups.ConclusionCognitive performance is a important distinguishing features of cerebral infarction. Some acoustic features may play an important role in distinguishing lacunar infarction from non-lacunar infarction.Keywords: Cerebral infarction, Lacunar infarction, Cognitive feature, Artificial intelligence technology, Acoustic features Introduction Stroke, which is the second leading cause of death in the world, has been ranked as the first cause of death for decades in China. Ischemic strokes are the most common, comprising about 62% of strokes worldwide [1].And cerebral infarction as a type of ischemic stroke has been recognized as a common disease in older adults, caused by ischemia or hypoxia due to the obstruction of blood supply to the brain [2]. With the development of economics, the incidence and mortality rates of cerebral infarction have been increased rapidly. Usually, cerebral infarction can result in speech difficulty, headache, motor weakness, and even death [3]. In fact, cerebral infarction can easily cause serious complications and sequelae, and bring a huge burden for patient’s family and society [1]. Previous research has found that a quarter of all cerebral infarction are lacunar [4]. Lacunar infarctions, which typically range from 2 to 20 mm in diameter, occur in the basal ganglia, deep cerebral white matter, or pons [4]. These infarctions are thought to be caused by a blockage of a solitary small perforating artery that provides blood supply to the subcortical regions of the brain [5]. Distinctive risk factors, natural progression, and clinical management are traditionally associated with them compared to other forms of brain infarctions [6, 7]. So Understanding the characteristics of lacunar and non-lacunar strokes can also help us provide more targeted treatments and interventions. As medical science has progressed, the predominant medical model has transitioned from a purely biological perspective to an integrated approach that incorporates biological, societal, and psychological factors. Therefore, the concern for life and health of cerebral infarction patients not only focuses on the recovery of limb function but also on nonlimb dysfunction. Cognition and language are the advanced functions of brain. If cognitive and language decline starts in the phase of cerebral infarction, it could be hypothesized that cognitive and language impairment presents a predict factor for cerebral infarction. Debaun et al. were able to effectively identify the cerebral infarction with a sensitivity of 86% and specificity of 81% through a short attention measure [8]. White et al. effectively screened children with cerebral infarction through a brief cognitive battery with 75% accuracy, 75% sensitivity, and 76% specificity [9].Although few studies have used speech features to classify cerebral infarction, attempts have been made to develop artificial intelligence classification models based on speech features in the field of related cognitive disorders.The study successfully achieved an 81.25% accuracy rate in the binary classification of AD and healthy participants conducted by Koo et al. utilized acoustic and linguistic features, as well as the attention mechanism and BiLSTM neural network [10]. The study of Yang et al. also realized the classification of AD based on multi-modal features including acoustic features [11]. Therefore, understanding the differences in acoustic features among cerebral infarction patients can also provide support for building multimodal diagnoses in the future. After all, the traditional methods of diagnosing cognitive impairment, such as scales, brain magnetic resonance imaging, cerebrospinal fluid, have deficiencies in balancing among the accuracy, cost, and convenience of the testing [11, 12]. So, the cognitive features and language features related to cerebral infarction can provide ideas and experiment possibilities for screening models based on intelligent language. To summarize, the present study aims to investigate the cognitive characteristics of different cerebral infarction patients firstly. And further, the study is one of the initial efforts to use artificial intelligence to find differences in speech features of cerebral infarction patients. In general, this research contributes to the field in several innovative ways. Firstly, it provide a theoretical basis for clinical studies and demonstrated the relationship between cerebral infarction and cognitive and language function through empirical research. Secondly, our feature selection process aims to identify the most relevant indicators of cerebral infarction, potentially uncovering new digital biomarkers for the condition. Finally, these features can provide the basis for us to development a machine learning-based classification model, potentially enabling earlier diagnosis and intervention. Literature review Overview of the relationship between cognitive impairment and cerebral infarction It is widely recognized that cognitive impairment is one of the most important characteristics of cerebral infarction.Cerebral infarction can lead to cognitive impairment and an increased risk of dementia [13, 14]. The majority of patients with cerebral infarction fall below the cognitive performance level expected according to premorbid functioning level. Researchers examined hospitalized patients with ischemic stroke and found that 10 percent to 38 percent had varying degrees of memory loss, inattention, disorientation, visuospatial impairment, and executive dysfunction [15]. Firstly, in terms of the prevalence of neuropsychological impairment in lacunar stroke patient, Grau-Olivares et al. found that 57.5% of the patients with acute lacunar infarction had mild neuropsychological disorders in a study of 40 patients with lacunar infarction, especially in patients with atypical lacunar syndrome and pure motor hemiparesis [16]. This suggested that neuropsychological impairment should be considered a common clinical feature of acute lacunar infarction. Coincidentally, a study by Jacova et al. demonstrated that 47% of 1636 patients with lacunar infarction had mild impairment, with the greatest deficits found in episodic memory, verbal fluency, and motor dexterity [17]. Secondly, in the specific aspect of cognition performance, Kramer et al. indicated that small vessel disease patients(including cerebral infarction) display poor performance on attention, processing speed, executive function and memory based on neuropsychological measures [18]. Jessica et al. also found that patients with stroke experience significant decline in global cognition and all cognitive domains [19].Even minor stroke affects daily functioning, executive functions, and cognition, consequently affecting participation, quality of life, and return to work [20].Especially, lacunar infarction has a more severe effect on cognition profile due to its repeated attacks [21].However, the relationship between cerebral infarction and cognitive function remains controversial. The study conducted by Schneider [22] founded that cerebral infarctions do not affect all cognitive systems equally, showing the strongest association with perceptual speed and the weakest with episodic memory. And the study of Araki and Ito founded that patients with cerebral infarction showed more show more severe impairments in attention/complex psychomotor skills and visual memory [23]. Therefore, it is important to investigate the association of different types of cerebral infarction with cognitive dysfunction, providing more evidence for clinical screening and treatment strategies. Overview of the relationship between language and speech dysfunctions and cerebral infarction On the other hand, the language and speech dysfunctions associated with cognitive impairment and dementia are under increasing scrutiny by researchers. Medical research shows that speech disorder is one of the typical clinical symptoms of cerebral infarctions at the early stage.Firstly, from the brain mechanism, Sunghyon et al. found that less lateralized language function related to the ventral pathway was influenced in the bilateral brain areas during both subacute and chronic phases after stroke [24]. A study explored the changes in language-related brain areas of cerebral infarction from both structural integrity and functional neural activity.The results showed that there were some differences in both structural and functional outcomes between patients with cerebral infarction and normal subjects [25]. Secondly, from the prevalence of speech disturbance, a study by Kazuo et al. demonstrated that 52.6% of 1247 patients with cerebral infarction had speech disturbance.This suggested that speech disturbance e was frequently observed in stroke patients at the onset and therefore could be useful to identify the problem at the earliest stage [26].However, the research on the correlation between language and different types of cerebral infarction from the perspective of speech feature is relatively scarce. Speech feature analysis may be a less uncomfortable ways of screening for disease of brain. Recording the interviews between doctors/neuropsychologists and patients and subsequently analyzing the patient’s speech features based on that recording would obviate the need for the patient need to undergo a cognitive test [27]. Therefore, by focusing on these non-invasive, easily obtainable features, we can make early screening for cerebral infarction easier. Method Subjects Data collection was carried out at the Department of Rehabilitation Medicine of Anhui No.2 Provincial People’s Hospital.Subjects were excluded from the study if they met any of the following exclusion criteria: (I) were aged below 40 years; (2) had a definite history of stroke, because stroke is a risk factor for dementia and dementia likely confounds the relationship between infarcts and cognitive function; (3) had a definite history of mental illness, such as schizophrenia, major depressive disorder; (4) had an alcohol or drug addiction; (5) had a serious physical disease; (6) had a definite history of brain diseases, other serious somatic diseases; (7) had severe aphasia and unconsciousness, and were unable to complete the neuropsychological tests; (8) had liver and kidney dysfunctions, Parkinson disease, multiple sclerosis, and other known diseases that may lead to cognitive impairments and dysphasia. A total of 117 patients (36 males, mean age = 65.17 years) were recruited from Anhui No.2 Provincial People’s Hospital. Professional clinical neurologists diagnosed all the subjects via neuroimaging tests, including computed tomography and magnetic resonance image. Among them, 95 patients were diagnosed as cerebral infarction. Subsequently, for the analysis, 95 patients with cerebral infarction were summarized by the size and location of infarct lesions, and 54 patients were identified as lacunar cerebral infarction. The rest of 22 patients were included as a control group. And a comprehensive neuropsychological battery that included the Chinese version of Montreal Cognitive Assessment Basic (MoCA-B) and the cookie theft picture description task. Their demographic and clinical information were also recorded. The stud The study was approved by Anhui No.2 Provincial People’s Hospital and Biomedical Ethics Committee, University of Science and Technology of China (IRB approval number: 2021-N(H)-213), and all the investigations were permitted by the participants and the informed consents were signed. It is believed that this study complies with the Helsinki Principles and is in line with medical ethics (as revised in 2013). Material and procedure To understand the patient’s cognitive function, all of them completed the Chinese version of Montreal Cognitive Assessment Basic (MoCA-B) under the guidance of trained research staff. MoCA-B is free for clinical use, assessing nine cognitive domains including language, attention, calculation, orientation, memory, abstraction, executive function, naming, and visual perception. And we collected speech recordings for this study based on the cookie theft picture description task (Fig. 1) which was taken from Boston Diagnostic Aphasia Examination-Third Edition (BDAE-3) [11].The cookie theft picture description task requires each subject to say as much as possible about the content in the picture, and it is allowed that the subjects are encouraged by the examiner to restate when they encounter difficulties. “Cookie Theft” picture description task Statistical analysis For the voice data, we used the openSMILE toolkit to extract the acoustic features through artificial intelligence technology. The openSMILE (open-source Speech and Music Interpretation by Large-space Extraction) is an open-source toolkit for audio feature extraction and classification of speech and music signals. The openSMILE is widely applied in automatic emotion recognition for affective computing. The eGeMAPS feature set was extracted by the openSMILE toolkit, which contains 88-dimensional high-level statistical features calculated based on low-level descriptors, MFCC, Spectral flux, and frequency features. The eGeMAPS feature set is an extension set to the minimalistic set [19]. The minimalistic acoustic parameter set was calculated from 18 low-level descriptors(LLD), including 6 frequency related parameters, 3 energy/amplitude related parameters and 9 spectral parameters. All LLDs are smoothed over time using a symmetric moving average filter with a length of 3 frames (for pitch, jitter, and shimmer, the smoothing is only performed within voiced regions, meaning that the transitions from 0 (unvoiced) to non-0 are not smoothed). The arithmetic mean and coefficient of variation (standard deviation normalized by the arithmetic mean) are applied as functionals to all 18 LLDs, resulting in 36 parameters.In addition to loudness and pitch, the following eight functionals are applied: 20th, 50th, and 80th percentiles, the range between 20th and 80th percentiles, and the mean and standard deviation of the slope of rising/falling signal parts. All functionals are applied only to voiced regions (non-zero F0), except for those that are applied to loudness. In addition, the arithmetic mean of the Alpha Ratio, the Hammarberg Index, and the spectral slopes from 0 to 500 Hz and 500–1500 Hz over all unvoiced segments are included, resulting in a total of 56 parameters. In addition, six temporal features are included. So in total, 62 parameters are contained in the Geneva Minimalistic Standard Parameter Set. The eGeMAPS feature set contains seven LLD in addition to the 18 LLD in the minimalistic set, including 5 spectral (balance/shape/dynamics) parameters and 2 frequency related parameters. The arithmetic mean and coefficient of variation are applied as functionals to all seven additional LLDs for all segments (voiced and unvoiced together), except for formant bandwidths, which are only applied in voiced regions. This results in an additional 14 descriptors. Furthermore, the arithmetic mean of the spectral flux in only unvoiced regions, as well as the arithmetic mean and coefficient of variation of the spectral flux and MFCC 1–4 in only voiced regions are included. This adds another 11 descriptors. Additionally, the equivalent sound level is included, resulting in a total of 26 extra parameters. When combined with the Minimalistic Set, the extended Geneva Minimalistic Acoustic Parameter Set (eGeMAPS) contains a total of 88 parameters. We used chi-square and one-way ANOVA to assess possible group differences between the three groups (lacunar cerebral infarction, non-lacunar cerebral infarction, and control group) in cognitive test performance and speech performance. P < 0.05was considered a statistical significance. Analyses were all conducted using SPSS version 26 (IBM Corporation, Armonk, NY, USA). Results The background information of the three groups The demographic characteristics of lacunar cerebral infarction, non-lacunar cerebral infarction, and control group are given in Table 1. Eleven patients with lacunar cerebral infarction were male and forty-three patients with lacunar cerebral infarction were female. The mean age of the lacunar cerebral infarction group was 64.87 years and the average education duration was 9.24 years. Nineteen patients with non-lacunar cerebral infarction were male and twenty-two patients with non-lacunar cerebral infarction were female. The mean age of the non-lacunar cerebral infarction group was 67.66 years and the average education duration was 8.66 years. Five patients from the control group were male and seventeen patients were female. The mean age of the control group was 62.81 years and the average education duration was 10.06 years. Table 1The demographic characteristics of the three groups of patientsCharacteristicsLacunar cerebral infarction groupNon-lacunar cerebral infarction groupControl groupN544122Age(year)64.87(7.47)67.66(7.01)62.81(5.40)Sex(male)11195Education(year)9.24(2.83)8.66(3.75)10.06(2.35) The demographic characteristics of the three groups of patients Comparisons of the three groups for cognitive performance The means and standard deviations for the nine cognitive domains are given in Table 2. One-way analysis of variance (ANOVA) was conducted to examine group differences on cognitive measures. A significant difference was found for the effect of group differences in total cognitive scores, F(2,114) = 4.838,p < 0.05. Using post hoc revealed that lacunar and non-lacunar infarction groups were significantly worse than the control group in total cognitive scores. And the non-lacunar infarction group performed worse than the lacunar cerebral infarction group in cognitive scores. Separate analyses of variance were conducted on each cognitive domain (see Table 2). From the results, there were significant differences in the fluency task among the three groups. Specifically, the differences were mainly between the non-lacunar cerebral infarction and the control group. And we also found significant differences in the memory task. Both the non-lacunar infarction and lacunar cerebral infarction groups were significantly worse than the control group in memory task. And the non-lacunar infarction group performed worse than the lacunar cerebral infarction group in the memory task. Table 2Mean scores and standard deviations on various cognitive domains for lacunar cerebral infarction group, non-lacunar cerebral infarction group, control group and F values for group differences on various measuresDomain2Lacunar cerebral infarction group(n = 54)1Non-lacunar cerebral infarction group(n = 41)3Control group(n = 22)FPost hoc comparisonMSDMSDMSDTotal score223.15620.374.68923.453.8264.838*L < C,N < C,L < NExecutive function0.330.4760.370.4880.450.5100.486—Language/ Fluency1.080.6750.850.5271.250.6393.065*L = C, N < COrientation5.670.5835.610.775.770.430.475—Calculation1.831.1781.541.0982.180.9582.470—Abstraction1.701.0021.821.0061.760.9690.109—Memory3.411.3252.731.6593.911.4775.011*L < C,N < C,L < NVisual perception2.020.8121.610.8621.860.8342.804—Naming3.610.5963.610.7033.770.4290.631—Attention2.350.99932.290.902.550.8000.545—L: Lacunar cerebral infarction group; N: Non-lacunar cerebral infarction group; C: Control group*p < 0.05; **p < 0.01; ***p < 0.001 Mean scores and standard deviations on various cognitive domains for lacunar cerebral infarction group, non-lacunar cerebral infarction group, control group and F values for group differences on various measures L: Lacunar cerebral infarction group; N: Non-lacunar cerebral infarction group; C: Control group *p < 0.05; **p < 0.01; ***p < 0.001 Comparisons of the three groups for speech performance We extracted 88-dimensional acoustic features using artificial intelligence techniques from individual speech data based on the cookie theft picture description task. It was found that there were significant differences in some of these acoustic characteristics, including F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope (standard deviation of the slope of falling signal parts of F0), loudness_sma3_percentile80.0 (percentile 80-th of estimate of perceived signal intensity from an auditory spectrum), loudness_sma3_meanRisingSlope(mean of the slope of rising signal parts of loudness), loudness_sma3_meanFallingSlope(mean of the slope of falling signal parts of loudness), loudness_sma3_stddevFallingSlope(standard deviation of the slope of falling signal parts of loudness), slopeUV0500_sma3nz_amean (mean of linear regression slope of the logarithmic power spectrum within 0–500 Hz band in voiced regions), MeanVoicedSegmentLengthSec (mean of continuously voiced regions) and equivalentSoundLevel_dBp, among the three groups of patients. Using post hoc revealed that the difference was mainly from the lacunar group and the non-lacunar group, but in the non-lacunar cerebral infarction group and the normal group, there was no significant difference in the acoustic characteristics of speech (Table 3). Table 3Mean scores and standard deviations on acoustic features for lacunar cerebral infarction group, non-lacunar cerebral infarction group, control group and F values for group differences on various featuresAcoustic FeaturesLacunar cerebral infarction group(n = 54)Non-lacunar cerebral infarction group(n = 41)Control group(n = 22)FPost hoc comparisonMSDMSDMSDF0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope49.09754.38876.68673.45435.64439.3233.732*L < Nloudness_sma3_percentile80.00.6590.3330.4490.3270.5490.3504.074*L > Nloudness_sma3_meanRisingSlope6.9033.0704.7743.0385.7843.4264.809**L > Nloudness_sma3_meanFallingSlope5.0502.3903.5132.3724.4192.5904.143*L > Nloudness_sma3_stddevFallingSlope3.1162.3391.6501.5572.1732.5094.993**L > NslopeUV0500_sma3nz_amean0.0100.029-0.0070.0160.0090.0196.160**L > N, C > NMeanVoicedSegmentLengthSec0.2060.0590.1770.0390.1800.0573.959*L > NequivalentSoundLevel_dBp-24.5807.621-34.40011.319-31.26612.9817.425***L > NOnly the acoustic features with significant differences are presented in the table*p < 0.05; **p < 0.01; ***p < 0.001 Mean scores and standard deviations on acoustic features for lacunar cerebral infarction group, non-lacunar cerebral infarction group, control group and F values for group differences on various features Only the acoustic features with significant differences are presented in the table *p < 0.05; **p < 0.01; ***p < 0.001 Discussion The present study investigated the cognitive characteristics and was the first attempt to establish the speech features of cerebral infarction patients. We compared lacunar cerebral infarction, non-lacunar cerebral infarction and control groups and found that the cerebral infarction group, including lacunar and non-lacunar, performed less well than the control groups on total cognitive score. There were significant differences in the fluency task and memory task among the three groups. This is also consistent with previous studies, cerebral infarction caused by cognitive impairment is mainly partial impairment, rather than a total decline in cognitive function [28]. The review by Almkvist found that there was a slight excess of deficits in executive functions, verbal fluency, attention and functions for cerebral infarction patients [29]. Jeffery et al. made a comprehensive analysis of 27 studies and found that cognitive impairment after cerebral infarction was mainly reflected in executive function, information processing speed and working memory [30].It can be seen that hypomnesis is one of the prominent symptoms of cerebral infarction, and studies suggest that learning disabilities after cerebral infarction are the result of impaired brain function [31]. A large number of researchers used clinical memory scales to test cerebral infarction patients and showed that each memory component and total score were significantly different from the control group [32, 33]. Although existing studies have found that the most common cognitive differences in cerebral infarction are in memory and executive function [34], fluency tasks are also often considered to be important means of executive function in studies. Therefore, although the differences in executive function between the three groups were not directly obtained in this study, we found differences in fluency, which may also reflect individual differences in executive function. And the non-lacunar infarction group performed worse than the lacunar cerebral infarction group in total cognitive performance and the memory task. This is also consistent with our understanding. By clinical definition, lacunar infarction is less symptomatic than other cerebral infarction. From the results, we found that there were significant differences in some of these acoustic characteristics, including F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope, loudness_sma3_percentile80.0, loudness_sma3_meanRisingSlope, loudness_sma3_meanFallingSlope, loudness_sma3_stddevFallingSlope, slopeUV0500_sma3nz_amean, MeanVoicedSegmentLengthSec and equivalentSoundLevel_dBp, among the three groups of patients based on intelligent speech analysis. Using post hoc revealed that the difference was mainly between the lacunar group and the non-lacunar group. This provides evidence for us to screen cerebral infarction with acoustic features. Although acoustic features alone may not be good at distinguishing cerebral infarction from non-cerebral infarction, the distinction between lacunar and non-lacunar cerebral infarction may be a good indicator. Few researchers have attempted to use speech features as indicators in the early screening of cerebral infarction. In the study of Yang et al., acoustic features have been found to be an important feature in the diagnosis of cognitive impairment in the elderly [11]. Although acoustic features alone may not be good at distinguishing cerebral infarction from non-cerebral infarction, the distinction between lacunar and non-lacunar cerebral infarction may be a good indicator. Especially, a study found that although lacunar syndromes are highly suggestive of small deep cerebral infarctions, lacunar syndromes not due to lacunar infarcts are found in 16.6% of cases [35]. This finding suggests that although most lacunar syndromes are associated with lacunar infarction, there are some cases that do not arise from this cause. These atypical lacunar syndromes may be associated with specific clinical predictors. Perhaps acoustic features can serve as a new attempt in future research. So understanding the differences in speech features between different groups could help us make more accurate classifications of diseases. Despite novel contributions, our study has some limitations. Firstly, in the study, we only examined lacunar cerebral infarction, non-lacunar cerebral infarction and control groups, it’s more meaningful if we could make a more detailed classification of cerebral infarction patients. Unfortunately, the small sample size does not allow further stratification of non-lacunar cerebral infarction patients. Future studies with a larger sample size especially multiple types of cerebral infarction datasets should be carried out for this purpose, we may get more interesting findings. Secondly, we did not further explore the source of cognitive and linguistic differences in cerebral infarction, as it may be due to individual factors. Further studies should explore possible mechanisms to make interpretations more firmed. Thirdly, the study did not present the cognitive and speech features of different cerebral infarction patients over time. Individuals’ profile could exhibit a year-to-year variability due to the possible factors such as enviromental change. Well-designed and well-executed large-scale cohort studies are an indispensable means for future research.Finally, the results can be made more interpretable through feature correlation analysis, but it may often overlook some hidden features. So, in future studies, we can attempt to use deep learning methods for feature representation for modeling, which may lead to new discoveries in our understanding of cerebral infarction. Acknowledgements Thanks to all the healthcare workers who participated in the study from Anhui No.2 Provincial People’s Hospital. Author contributions Yang Niu and Mingming Hu devised the project and the main conceptual ideas, along with the data analysis. Xiaoyan Zhao conducted the data collection and the preprocessing, Yongmei Zhu and Xiaoping Gao worked on all of the technical details and provided the clinical theritial support. All authors reviewed the manuscript. Funding Not applicable. Data availability No datasets were generated or analysed during the current study. Declarations Ethics approval and consent to participate The study was approved by Anhui No.2 Provincial People’s Hospital and Biomedical Ethics Committee, University of Science and Technology of China (IRB approval number: 2021-N(H)-213), and all the investigations were permitted by the participants and the informed consents were signed. It is believed that this study complies with the Helsinki Principles and is in line with medical ethics. “University of Science and Technology of China” is an ethics committee/ Institutional review board that approved my study. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement No datasets were generated or analysed during the current study.