Principal Component Scores Research Articles

The construction of an intelligent English grading teaching model based on partial least squares method

Abstract Building an intelligent English grading teaching model helps to improve English teaching and students’ English performance in colleges and universities. In this paper, the PLS-ELM model is constructed from partial least squares as the base algorithm, the concept of an extreme learning machine is introduced, and the principal component score of the PLS algorithm is used to the number of neuron nodes of ELM, and then the adjustment of weights is realized. The principal components of PLS were used to analyze the selection of independent variables for the English-graded teaching model. The effectiveness of the English-graded teaching model was verified by comparing students’ satisfaction with teaching evaluation and performance after teaching implementation. Regarding student satisfaction, the percentages of those evaluated as very satisfied, generally satisfied, and dissatisfied were 53.71%, 32.86%, and 13.43%, respectively. Regarding students’ performance, the average score of students in the experimental group was 88.6, 14.62% higher than the English performance of the control group. This shows that the English grading teaching mode can improve students’ English learning performance, enhancing students’ own quality and professional ability, improving university English teaching, and cultivating students’ English application ability.

Relationship between gait parameters and cognitive indexes in adult aging.

This study aimed to identify the most effective summary cognitive index predicted from spatio-temporal gait features (STGF) extracted from gait patterns. The study involved 125 participants, including 40 young (mean age: 27.65 years, 50% women), and 85 older adults (mean age: 73.25 years, 62.35% women). The group of older adults included both healthy adults and those with Mild Cognitive Impairment (MCI). Participant´s performance in various cognitive domains was evaluated using 12 cognitive measures from five neuropsychological tests. Four summary cognitive indexes were calculated for each case: 1) the z-score of Mini-Mental State Examination (MMSE) from a population norm (MMSE z-score); 2) the sum of the absolute z-scores of the patients' neuropsychological measures from a population norm (ZSum); 3) the first principal component scores obtained from the individual cognitive variables z-scores (PCCog); and 4) the Mahalanobis distance between the vector that represents the subject's cognitive state (defined by the 12 cognitive variables) and the vector corresponding to a population norm (MDCog). The gait patterns were recorded using a body-fixed Inertial Measurement Unit while participants executed four walking tasks (normal, fast, easy- and hard-dual tasks). Sixteen STGF for each walking task, and the dual-task costs for the dual tasks (when a subject performs an attention-demanding task and walks at the same time) were computed. After applied Principal Component Analysis to gait measures (96 features), a robust regression was used to predict each cognitive index and individual cognitive variable. The adjusted proportion of variance (adjusted-R2) coefficients were reported, and confidence intervals were estimated using the bootstrap procedure. The mean values of adjusted-R2 for the summary cognitive indexes were as follows: 0.0248 for MMSE z-score, 0.0080 for ZSum, 0.0033 for PCCog, and 0.4445 for MDCog. The mean adjusted-R2 values for the z-scores of individual cognitive variables ranged between 0.0009 and 0.0693. Multiple linear regression was only statistically significant for MDCog, with the highest estimated adjusted-R2 value. The association between individual cognitive variables and most of the summary cognitive indexes with gait parameters was weak. However, the MDCog index showed a stronger and significant association with the STGF, exhibiting the highest value of the proportion of the variance that can be explained by the predictor variables. These findings suggest that the MDCog index may be a useful tool in studying the relationship between gait patterns and cognition.

MULTIDIMENSIONAL PROfiLES OF COGNITIVE DETERIORATION AND RECOVERY ASSOCIATED WITH GLIOBLASTOMA SURGERY

Abstract AIMS Patients with glioblastoma suffer a high incidence of cognitive deficits. Cognition is the result of multiple neural mechanisms and glioblastoma is diffusely infiltrative with the capacity to invade multiple functional areas in the brain. We hypothesise that cognitive deterioration and recovery in these patients will be multidimensional in nature and therefore different components of cognition will show distinct recovery profiles following surgery. METHOD 28 patients with high-grade glioma, recruited at a single centre underwent preoperative (T0), early postoperative (within one week) (T1) and delayed (six to eight weeks) (T2) testing of their cognition with a neurocognitive screening tests; Oxford Cognitive Screen and Cambridge Attention, Memory and Perception Screen (OCS-Bridge) tool. Varimax principal component analysis was performed to determine if there are low-dimensional domains of cognition examined by the tests. Changes in cognitive scores between time points was calculated and correlated with principal component scores (PCs). RESULTS We found six principal components which explained 71% of the variance in OCS scores. and six different com- ponents explained 63% of the variance in Bridge scores. Cognitive deterioration was largely multidimensional with few significant correlations between PC change scores at T1 except tests loading on to visual field and spatial neglect. Cognitive recovery measured with T2 change scores were not significantly correlated between attention, praxis and language domains and between verbal working memory and visual memory and attention. CONCLUSIONS Longitudinal cognitive signatures of deterioration and recovery are heterogenous between patients. This high- lights the importance of targeting potentially distinct neural mechanisms for the rehabilitation of these patients.

Evaluation of Spatiotemporal Patterns and Water Quality Conditions Using Multivariate Statistical Analysis in the Yangtze River, China

As a crucial surface water resource, the Yangtze River has raised concerns about its water quality due to its importance in economic and social development, environmental conservation, and agricultural development. The principal component analysis (PCA), hierarchical clustering analysis (HCA), and the water quality index (WQI) were utilized to assess the overall condition and detect spatiotemporal patterns and the key parameters of water quality in the Yangtze River. All usage data were determined monthly from samples taken in 2021 at the 33 Yangtze River water quality monitoring stations. The results demonstrated that 85% of the monitoring stations in the whole Yangtze River were maintained at a “good” condition, with average WQI values ranging from 71.16 to 81.25. The water quality was slightly poorer in the summer, with 56.6% of monitoring stations being in “medium” condition. Spatially, there was a downward trend in the water quality from upstream to downstream. Two significant principal component scores (PCs) were produced as a result of PCA and HCA, explaining 60.3% of the total variance in the upstream, 67.4% in the transition zone, and 50.4% in the downstream, respectively. In addition, the middle–upper reaches of water quality were found to correlated with CODMn, whereas the water quality in the downstream were mainly influenced by TUR, TP, T, and DO. The results primarily motivated our understanding of the Yangtze River’s water quality status and suggested the main targets for water quality improvement in different monitoring areas.

Solar Radiation and Weather Analysis of Meteorological Satellite Data by Tensor Decomposition

In this study, the data obtained from meteorological satellites were analyzed using tensor decomposition. The data used in this paper are meteorological image data observed by the Himawari-8 satellite and solar radiation data generated from Himawari Standard Data. First, we applied Higher-Order Singular Value Decomposition (HOSVD), a type of tensor decomposition, to the original image data and analyzed the features of the data, called the core tensor, obtained from the decomposition. As a result, it was found that the maximum value of the core tensor element is related to the cloud cover in the observed area. We then applied Multidimensional Principal Component Analysis (MPCA), an extension of principal component analysis computed using HOSVD, to the solar radiation data and analyzed the Principal Components (PC) obtained from MPCA. We also found that the PC with the highest contribution rate is related to the solar radiation in the entire observation area. The resulting PC score was compared to actual weather data. From the result, it was confirmed that the temporal transition of the amount of solar radiation in this area can be expressed almost correctly by using the PC score.

Robust joint modelling of sparsely observed paired functional data

AbstractA reduced‐rank mixed‐effects model is developed for robust modelling of sparsely observed paired functional data. In this model, the curves for each functional variable are summarized using a few functional principal components, and the association of the two functional variables is modelled through the association of the principal component scores. A multivariate‐scale mixture of normal distributions is used to model the principal component scores and the measurement errors in order to handle outlying observations and achieve robust inference. The mean functions and principal component functions are modelled using splines, and roughness penalties are applied to avoid overfitting. An EM algorithm is developed for computation of model fitting and prediction. A simulation study shows that the proposed method outperforms an existing method, which is not designed for robust estimation. The effectiveness of the proposed method is illustrated through an application of fitting multiband light curves of Type Ia supernovae.

Fit-free analysis of fluorescence lifetime imaging data using chemometrics approach for rapid and nondestructive wood species classification

Abstract Conventional fluorescence spectroscopy has been suggested as a valuable tool for classifying wood species rapidly and non-destructively. However, because it is challenging to conduct absolute emission intensity measurements, fluorescence analysis statistics are difficult to obtain. In this study, another dimension of fluorescence, that is, fluorescence lifetime, was further evaluated to address this issue. A time-resolved fluorescence spectroscopic measurement system was first designed, mainly using a streak camera, picosecond pulsed laser at 403 nm, and a spectroscope, to collect the fluorescence time-delay (FTD) profiles and steady-state fluorescence intensity (FI) spectra simultaneously from 15 wood species. For data analysis, principal component analysis was used to “compress” the mean-centered FTD and FI spectra. Then, support vector machine classification analysis was utilized to train the wood species classification model based on their principal component scores. To avoid overfitting, ten-fold cross-validation was used to train the calibration model using 70 % of the total samples, and the remaining 30 % hold-out validation was used to test its reproducibility. The cross-validation accuracies were 100 % (5 softwoods) and 96 % (10 hardwoods), with test-validation accuracies of 96 % and 89 %.

Stress Chain Analysis for an ABA Triblock Copolymer Using Principal Component Scores.

Researchers characterize the mechanical properties of ABA triblock copolymers by structures such as chain conformation. During elongation, bridge chains are stretched and act as a stress chain. Some loop chains also act as a stress chain because of the transmission of stress through an entanglement of loop chains. The stress chain, including the entangled loop chains, in an ABA triblock copolymer that exhibits a body-centered cubic structure was analyzed by principal component analysis (PCA), using the physical data for the B block obtained by coarse-grained molecular dynamics simulations. Local deformation of the A domains caused by the stress chains was also analyzed by PCA of the A block. The dynamics of the stress chain strongly corresponded to the recombination of the A domains; shrinkage because of domain breakage, replacement of stress chains, and biased stress distribution as well as its time dependence were observed.

Construction of Air Quality Level Prediction Model Based on STEPDISC-PCA-BP

Air quality level has a complex nonlinear relationship with air pollutant and meteorological conditions, including multiple factors, overlapping information, and difficulty solving equations. In order to identify significant factors, remove correlations, reduce data dimensionality, and simplify the model structure, a BP neural network model for air quality level prediction optimized by stepwise discriminant analysis (STEPDISC) and principal component analysis (PCA) is proposed with 12 factors of historical daily meteorology and air pollutants in Bayannur city as samples. The results showed that, at the significance level of 0.01, the STEPDISC method retained 9 significant impact factors. The PCA method made an orthogonal linear combination of the 9 factors to form the principal components, and the contribution of the top 5 principal components were 37.6%, 19.2%, 15.3%, 8.8%, and 7.7%. At a contribution threshold of 0.85, the top 5 principal component scores were used as input nodes to construct the STEPDISC-PCA-BP model, which had a prediction accuracy of 85.5%.Compared with the PCA-BP and BP model, which had a prediction accuracy of 61.8% and 56.7%, respectively, the STEPDISC-PCA-BP model has a higher prediction accuracy, shorter time, and lower complexity of structure and data dimensionality, and can provide the necessary technical support for the local air quality improvement.

A principal component analysis assisted machine learning modeling and validation of methanol formation over Cu-based catalysts in direct CO2 hydrogenation

With the advent of powerful machine learning algorithms strongly supporting complex non-linear regression modeling, catalyst design features for a wide and customized set of catalysts used for a specific reaction have been made easy. Herein, we use these techniques in the methanol synthesis by CO2 reduction over Cu-based binary and ternary catalysts with the help of three machine learning algorithms: artificial neural network, support vector machine regression, and gaussian process regression. 227 catalytic performance dataset points from existing literature on CO2 hydrogenation to methanol were compiled and initially accessed by Principal Component Analysis (PCA) for training and preliminary evaluation of the algorithms, which was further guided using a 10-fold cross-validation method. The predictive model and its insights were validated experimentally over 30 datasets derived from experimental runs of this reaction over a ternary Cu/ZnO/ZrO2 laboratory-synthesized catalyst at varying conditions of temperature, pressure, and space velocity in a continuous mode fixed-bed plug-flow reactor. The assessment of the space of input and laboratory data was aided by Principal Component Analysis (PCA), scores, and loadings plot. This work shows how experimentalists can predict typical heterogeneous catalytic reaction outputs (R2 greater than 0.9 for three variables) with fair accuracy using a combination of machine learning and PCA.

Identification of Impacts from Meteorology and Local and Transported Photochemical Generation on Ozone Trends in Changsha from 2018 to 2020

Ozone (O3) pollution in Hunan province has become the most important factor among the six common conventional pollutants (i.e., NO2, SO2, CO, O3, PM10, and PM2.5) in the atmospheric environment. Further investigation has indicated that the relevant studies of O3 are insufficient. Therefore, it is essential to clarify the key driving factors of O3 variations for government regulators. In this study, a combined method consisting of a generalized additive model (GAM), empirical orthogonal function (EOF), and absolute principal component scores (APCs) model was employed to identify and quantify the impacts of meteorology and local photochemical generation (local) and that transported from outside (nonlocal) on O3 variations from 2018-2020. Simultaneously, the driving factors of O3 annual values from 2018 to 2019 and from 2019 to 2020 in Changsha were analyzed. The results showed that O3 episodes were commonly caused by meteorology when the relative contribution from precursors was high, on the short-term time scale. Overall, on the temporal scale, meteorology and local were the driving factors for the increasing annual O3 from 2018 to 2019. Additionally, the contribution from meteorology, local, and nonlocal decreased from 2019 to 2020, leading to a lower level of O3 concentration in 2020. Geographically, the east, north, and south of Changsha were mainly affected by meteorology, local, and nonlocal, respectively. Throughout the three years, nonlocal exhibited a sustained decreasing trend, whereas the tendencies from meteorology and local varied by year and geography. Local contribution in the north of Changsha increased from 2018 to 2019, which was likely attributed to the increasing biogenic volatile organic compound emission (BVOCs), and it became lower in the south owing to the strengthened consumption by NOx. Impacts from meteorology on O3 in all sites were enhanced from 2018 to 2019. By contrast, local contribution decreased in the north and increased in the south with the decline in BVOC and NOx emissions from 2019 to 2020, when the meteorological impacts on O3 in the whole area became weak.

Poster 139: Utilization of 3D Statistical Shape Modeling of Anatomic Variation Across a Cohort of Patella Instability Patients

Poster 139: Utilization of 3D Statistical Shape Modeling of Anatomic Variation Across a Cohort of Patella Instability Patients

The effect of severe obesity on three-dimensional ground reaction force signals during walking

BackgroundThe gait pattern of adults with class I obesity [30 ≤ body mass index < 35kg/m2] was characterized by altered three-dimensional ground reaction force signals compared to lean adults (18.5 ≤ body mass index < 25 kg/m2). However, results might not be generalizable to adults with severe obesity (class II and III; body mass index ≥ 35 kg/m2). Hence, the purpose of the present study was to investigate the differences in relative ground reaction force signals, i.e., normalized by body weight, between adults with severe obesity and lean adults using functional principal component analysis. MethodsThirteen lean and eighteen sedentary adults with severe obesity performed a 5-min walking trial (1.11 m/s) on an instrumented treadmill. The first five functional principal components of the relative force signals (mediolateral, anterior-posterior, and vertical directions) were obtained using functional principal component analysis. Functional principal component scores were compared between groups using an analysis of covariance with age as covariable. FindingsFunctional principal component analysis reported a statistically significant group effect for first functional principal component score for mediolateral (P = 0.004), and second and fifth functional principal component scores for anterior-posterior (P ≤ 0.02) force signals. Adults with severe obesity displayed a greater mediolateral force during most of the stance but similar magnitudes of the anterior-posterior and vertical forces compared to lean adults. InterpretationTherefore, increasing the obesity level accentuates differences in mediolateral force but promotes no specific changes in anterior-posterior force likely due to chronic loading adaptation.

A Comprehensive Evaluation Model Based on the Environmental Impact of Saihanba

China Saihanba Forest Farm has become an ecological green farm with stable sand control function, and an ecological environmental impact assessment model is constructed according to the ecological model of Saihanba Forest Farm. According to the seven data indicators collected, this article uses principal component analysis to analyze. From the principal component scores, it can be seen that there is a significant difference in the scores before and after restoration, indicating that the impact of Saihanba Forest Farm on the ecological environment before and after restoration is very significant. At the same time, in order to establish ecological regions more accurately and extend the ecological protection model to the whole country, based on the AHP (Analytic Hierarchy Process) and TOPSIS (Distance between superior and inferior solutions), a model is established to calculate the comprehensive evaluation indicators of the ecological environment in Liaoning, Ningxia, and Shanxi provinces. The scores of the comprehensive evaluation indicators are used to determine the establishment of ecological zones in Ningxia and Shanxi provinces.

590-P: Assessment of a Comprehensive Physical Fitness Index to Predict Future Diabetes Mellitus (DM) and Metabolic Syndrome (MetS)

Impaired muscle strength, balance, and flexibility have been associated with the development of type 2 diabetes and atherosclerosis. A physical score (PS), which integrates these physical fitness indices, could be expected to better predict future metabolic diseases. We investigated the longitudinal relationship between the baseline PS and changes in PS and future DM and MetS. Analyzed were 5,718 persons (4,068 men) aged 30 to 69 y without DM who underwent physical fitness tests. Principal component analysis was performed on the correlation matrix of the physical fitness test results according to age and sex. The PS was defined as the first principal component score. Associations between PSs at the start of observation (year −2), change in the PS two years later (year 0) and the incidence of DM at the end of observation (until year 3) were examined by logistic regression analysis. The same analysis was performed for 5,304 persons (3,729 men) without MetS at baseline. Although no significant difference was found in the incidence of DM, a significant difference was found in the incidence of MetS; odds ratio was 1.39 (1.20, 1.61) for each decrease of 1 in the PS in year −2 and 1.51 (1.18, 1.92) for each decrease of 1 in the change in PS from year -2 to year 0. = Results showed that the PS was a simple and non-invasive predictor of MetS. Disclosure T.Sato: None. K.Kato: None. H.Sone: Research Support; Novo Nordisk, Astellas Pharma Inc., Kyowa Kirin Co., Ltd., Taisho Pharmaceutical Holdings Co., Ltd., Ono Pharmaceutical Co., Ltd., Eisai Co., Ltd., Takeda Pharmaceutical Co., Ltd. K.Fujihara: None. M.H.Yamada: None. Y.Yaguchi: None. M.Yamamoto: None. M.Kitazawa: None. H.Ishiguro: None. T.Osawa: None. T.Yamada: None. Funding Japan Society for the Promotion of Science; Japan Ministry of Health, Labour and Welfare (21K11569)

Divergent temporal changes of heavy metals in the soil induced by natural versus anthropogenic forces: A case study in the Yangtze River delta area, China

The contributions of natural versus anthropogenic forces on temporal changes of metals in the soil of the Yangtze River delta region were successfully quantified by combining repeated soil sampling, geostatistics and the modified principal component scores and multiple linear regressions approach. The findings show that the mean concentrations of Cu, Cd, As, Hg, Cr and Ni generally exhibited a decrease trend from 2010 to 2020. The decline of soil Hg was most outstanding, decreased by 20 % as a whole. The result of the modified principal component scores and multiple linear regressions approach suggests that the decrease of Hg content was predominately driven by the geochemical processes, with 48 % contribution. The mean soil Pb concentration increased by 40 % from 2010 to 2020. Approximately 76 % of the Pb change was attributed to the strengthened development of Pb-containing battery industry in the south part of the Yangtze River delta region. The Pb battery industry development also contributed 48 % of the change of Cd. The anthropogenic activities involving Cu, Zn or fossil fuel consumption contributed 32–35 % of the changes of As, Cu and Zn concentrations in the soil, and the steel industries contributed 82 % of Cr and 60 % of Ni changes, respectively.

Application of robust principal component analysis–multivariate adaptive regression splines for the determination of °API gravity in crude oil samples using ATR-FTIR spectroscopy

The robust principal component analysis-multivariate adaptive regression splines (r-PCA-MARS) has been applied and verified for analysis of the quantitative determination of American Petroleum Institute (°API) gravity values in crude oils. Seven principal component (PC) scores using 95.00% variance by principal component analysis (PCA) were applied as inputs in the MARS model. The calibration and prediction sets were obtained using duplex algorithm for the construction of the model and the then for the validation of the model. The calibration set (67*7) was used for the r-PCA-MARS model. The partial least squares regression (PLS-R), and support vector machine regression (SVM-R) models were utilized for comparison the quantitative value of the °API gravity in crude oils. In this paper, we also conducted a comparison study of Kennard-stone (KS) and duplex splitting methods on PLS-R and SVM-R models. The efficiency of the r-PCA-MARS model was evaluated using coefficient of determination (R2), R2 estimated by generalized cross-validation (R2GCV), root mean square error of calibration (RMSEC), root mean square error of prediction (RMSEP), and mean absolute error (MAE). The optimal r-PCA-MARS model uses 32 basis functions to characterize the °API gravity values in crude oils. The correlation coefficients value for calibration and prediction sets were 0.997 and 0.926, respectively. The RMSEC, RMSEP, MAE, and R2GCV in the piecewise-cubic r-PCA-MARS model was 6.726*10-13, 0.538, 0.290 and 0.988, respectively. According to the results, the r-PCA-MARS model provided high efficiency than commonly used regression models for prediction of °API gravity values in crude oils. The result of this study confirmed that the r-PCA-MARS model is the best model with more successful than the PLS-R and SVM-R models. It can be concluded that the r-PCA-MARS model is an appropriate model for describing the physicochemical properties of crude oil samples in the oil industry.

Characterizing Ultrafine Particle Mobile Monitoring Data for Epidemiology.

Mobile monitoring is increasingly used to assess exposure to traffic-related air pollutants (TRAPs), including ultrafine particles (UFPs). Due to the rapid spatial decrease in the concentration of UFPs and other TRAPs with distance from roadways, mobile measurements may be non-representative of residential exposures, which are commonly used for epidemiologic studies. Our goal was to develop, apply, and test one possible approach for using mobile measurements in exposure assessment for epidemiology. We used an absolute principal component score model to adjust the contribution of on-road sources in mobile measurements to provide exposure predictions representative of cohort locations. We then compared UFP predictions at residential locations from mobile on-road plume-adjusted versus stationary measurements to understand the contribution of mobile measurements and characterize their differences. We found that predictions from mobile measurements are more representative of cohort locations after down-weighting the contribution of localized on-road plumes. Further, predictions at cohort locations derived from mobile measurements incorporate more spatial variation compared to those from short-term stationary data. Sensitivity analyses suggest that this additional spatial information captures features in the exposure surface not identified from the stationary data alone. We recommend the correction of mobile measurements to create exposure predictions representative of residential exposure for epidemiology.

Dietary Patterns and Their Associations with Symptom Levels Among People with Multiple Sclerosis: A Real-World Digital Study.

The objective of the study was to investigate long-term food intake patterns and establish possible associations between the inferred dietary habits and levels of reported symptoms among people with multiple sclerosis (MS) in Denmark. The present study was designed as a prospective cohort study. Participants were invited to register daily food intake and MS symptoms and were observed during a period of 100days. Dropout and inclusion probabilities were addressed using generalized linear models. Dietary clusters were identified among 163 participants using hierarchical clustering on principal component scores. Associations between the dietary clusters and the levels of self-assessed MS symptoms were estimated using inverse probability weighting. Furthermore, the effect of a person's position on the first and second principal dietary component axis on symptom burden was investigated. Three dietary clusters were identified: a Western dietary cluster, a plant-rich dietary cluster and a varied dietary cluster. Analyses further indicated a vegetables-fish-fruit-whole grain axis and a red-meat-processed-meat axis. The plant-rich dietary cluster showed reduction in symptom burden in nine pre-defined MS symptoms compared to the Western dietary cluster (between 19 and 90% reduction). This reduction was significant for pain and bladder dysfunction as well as across all nine symptoms (pooled p value = 0.012). Related to the two dietary axes, high intake of vegetables resulted in 32-74% reduction in symptom burden compared to low levels of vegetable intake. Across symptoms, this was significant (pooled p value = 0.015), also regarding walking difficulty and fatigue. Three dietary clusters were identified. Compared to levels of self-assessed MS-related symptoms, and adjusted for potential confounders, the results suggested less symptom burden with increased intake of vegetables. Although the research design limits the possibilities of establishing causal inference, the results indicate that general guidelines for healthy diet may be relevant as a tool in coping with MS symptoms.

Late Holocene vegetation and climate change in the northern Tibetan Plateau region based on a palynological analysis from Lake Ayakum

In this paper, we undertake a high-resolution palynological analysis of a sediment core covering the last 1500 years obtained from Lake Ayakum in the northern Tibetan Plateau, to improve the understanding of regional climate fluctuations and their driving mechanisms. The pollen spectra are dominated by herbaceous species such as Artemisia, Chenopodiaceae, Cyperaceae and Poaceae, which grew in various alpine desert settings. Four pollen assemblage zones are identified using stratigraphically constrained cluster analysis, representing climate history at different stages. The Artemisia/Chenopodiaceae (A/C) ratio and axis score of the first principal component are used to infer regional moisture conditions and reveal a temporal pattern that is consistent with various records from central Asia. The source of Cyperaceae pollen is mainly from marsh wetlands distributed around the lake, which responded sensitively to the lake level fluctuations. In contrast, arboreal pollen is mainly a long-distance transported component, reflecting changes in regional vegetation. Our findings show that the northern Tibetan Plateau experienced relatively wet conditions during the Dark Age Cold Period and the Little Ice Age, while relatively dry conditions occurred during the Medieval Warm Period and the last two centuries. Fluctuations in moisture levels are broadly consistent with moisture records elsewhere in western China, and show a good correlation with phase changes of the North Atlantic Oscillation. The climate changes in the northern Tibetan Plateau are supposed to be primarily triggered by the NAO phases through south-north migrations of the mid-latitude westerlies during the late Holocene.