In this study, based on a morpheme segmentation framework, we researched a text keyword extraction method for Uyghur, Kazakh and Kirghiz languages, which have similar grammatical and lexical structures. In these languages, affixes and a stem are joined together to form a word. A stem is a word particle with a notional meaning, while the affixes perform grammatical functions. Because of these derivative properties, the vocabularies used for these languages are huge. Therefore, pre-processing is a necessary step in NLP tasks for Uyghur, Kazakh and Kirghiz. Morpheme segmentation enabled us to remove the suffixes as the auxiliary unit while retaining the meaningful stem and it reduced the dimension of the feature space present in the keyword extraction task for Uyghur, Kazakh and Kirghiz texts. We transformed the morpheme segmentation task into the problem of labeling the morpheme sequences, and we used the Bi-LSTM network to bidirectionally obtain the position feature information of character sequences. We applied CRF to effectively learn the information of the preceding and following label sequences to build a highly accurate Bi-LSTM_CRF morpheme segmentation model, and we prepared morpheme-based experimental text sets by using this model. Subsequently, we used the stem vectors’ similarity to modify the TextRank algorithm, subsequent to the training of the stem embedding vector using the Doc2vec algorithm, and then we performed a text keyword extraction experiment. In this experiment, the highest F1 scores of 43.8%, 44% and 43.9% were obtained for three datasets. The experimental results show that the morpheme-based approach provides much better results than the word-based approach, which shows the stem vector similarity weighting is an efficient method for the text keyword extraction task, thus proving the efficiency of morpheme sequence for morphologically derivative languages.

Drought characterization in hyper-arid endorheic basins requires multi-index approaches that capture distinct hydrometeorological processes. This study investigates spatio-temporal drought dynamics in the Tarim Basin (TB)—China’s largest inland arid region—using two complementary remote sensing indices: the Temperature Vegetation Dryness Index (TVDI) for landscape-scale moisture and the Crop Water Stress Index (CWSI) for agricultural drought. Based on 2000–2024 remote sensing and meteorological data, we employed a pixel-wise Random Forest framework with spatial cross-validation and permutation importance analysis to quantify climatic drivers across the TB. Results reveal a fundamental “core-periphery” dichotomy: TVDI identifies persistent extreme drought in the Taklamakan Desert core, while CWSI reveals alleviating water stress in peripheral oasis farmlands (73.21% showing significant decrease, p < 0.05). Despite regional warming-wetting trends, TVDI exhibited an insignificant decrease (54.72% of the basin), contrasting with CWSI's significant agricultural drought alleviation. Vapor Pressure Deficit (VPD)—a key atmospheric dryness indicator—exhibited high relative permutation importance for both drought indices (72–75%), considerably exceeding the values obtained for precipitation (6–8%) within the Tarim Basin. Secondary drivers diverge by land surface type: TVDI responds to Relative Humidity (8.2%) and Precipitation (6.1%), while CWSI is modulated by Land Surface Temperature (9.4%) and Sunshine Hours (7.8%). Partial correlation analyses controlling for topography and temperature confirm VPD’s independent effect on drought severity. Large-scale climate oscillations, particularly the Arctic Oscillation (AO) and ENSO-PDO interactions, significantly modulate interannual drought variability (r = 0.74–0.75, p < 0.01). This study provides the first pixel-scale quantification of the relative dominance of atmospheric water demand over precipitation in driving drought evolution in the Tarim Basin, with VPD contributing 72–75% of the total permutation importance compared to 6–8% for precipitation. This quantitative benchmark offers actionable parameters for drought monitoring systems in arid regions and underscores the need to integrate VPD and large-scale climate signals into early warning frameworks.

Bulk skin transcriptomes cannot distinguish cell-intrinsic gene regulation from shifts in tissue-type composition, and computational deconvolution is widely applied to resolve this limitation. In this study, we performed deconvolution on published Hetian sheep skin RNA-seq data from three high-wool-density and three low-wool-density animals, using the available ovine skin single-cell reference. All hypotheses and analytical workflows were prospectively specified and timestamped prior to estimating cell-type fractions. Our work identifies two major methodological pitfalls for livestock tissue deconvolution. First, gene-length bias distorts deconvolution outputs derived from mixed bulk and 3′ single-cell references. After bias correction, the estimated hair-follicle epithelial fraction changed substantially. This bias cannot be uncovered by standard benchmark workflows, highlighting a hidden failure mode for deconvolution. Second, limited signature separability within the reference prevents reliable discrimination between related epidermal compartments. This non-identifiability strongly disturbs apparent compartment-level effects. When analysed at the granularity the reference reliably supports, we found no compositional difference between high- and low-wool-density groups for the epithelial hypothesis. The dermal hypothesis could not be meaningfully evaluated due to incompatible reference compartment merging requirements. Group-level differential expression remained robust despite this null compositional outcome. We recovered a large set of differentially expressed genes consistent with the original study, including most qPCR-validated candidates. Adjusting for cell-type composition altered part of this gene list, as expected given negligible group-level compositional divergence. Co-expression modules correlated more strongly with inter-individual compositional variation than with wool-density phenotype. This indicates compositional heterogeneity shapes transcriptome co-expression even without measurable mean differences across experimental groups. To avoid misleading biological inference, we propose a set of practical diagnostic checks for deconvolution when developmental or limited single-cell atlases are used for livestock tissues. Collectively, these transcriptomic patterns support cell-state-driven regulation rather than detectable mean shifts in tissue compartment composition.

Flash drought (FD) is an extreme climate event that intensifies within days and exerts severe socio-environmental impacts. Its onset and evolution remain difficult to predict. Here, we quantify the spatio-temporal dynamics of FD across northern Xinjiang from 1961 to 2023 and identify the dominant driving factors. We apply linear trend detection, wavelet analysis, change-point detection, random forest (RF) modeling, and Pearson correlation. Results show that winter is becoming significantly wetter, whereas the annual signal and the other three seasons exhibit drying trends. After 1980, both FD frequency and FD duration increased; the longest single event lasted 40 days. Spatially, FD is concentrated in the Ili River Valley and the Altay region; the Akdala station recorded the highest count (nine events). Duration, rather than frequency, peaks on the northern slope of the Tianshan Mountains, where the maximum length reaches 40 days. RF importance ranks the Pacific Decadal Oscillation (PDO) as the leading driver (20.9%), followed by air temperature (17.8%); the sunspot index contributes only 6.1%.

Drought characterization in hyper-arid endorheic basins requires multi-index approaches that capture distinct hydrological processes. This study investigates spatio-temporal drought dynamics in the Tarim Basin (TB), China's largest inland arid region, using two complementary remote sensing indices: the Temperature Vegetation Dryness Index (TVDI) for landscape-scale moisture and the Crop Water Stress Index (CWSI) for agricultural drought. Based on 2000–2024 remote sensing and simulated data, we employed a pixel-wise Random Forest framework with spatial cross-validation and permutation importance analysis to quantify drivers across TB. Results reveal a fundamental "core-periphery" dichotomy: TVDI identifies persistent extreme drought in the Taklamakan Desert core, while CWSI highlights intensifying water stress in peripheral oasis farmlands (73.21% showing significant increase, p<0.05). Despite regional warming-wetting trends, TVDI exhibited insignificant decrease (54.72% of basin), contrasting with CWSI's significant agricultural drought intensification. Vapor Pressure Deficit (VPD) dominates both indices (Permutation Importance: 72–75%), but secondary drivers diverge: TVDI responds to Relative Humidity (8.2%) and Precipitation (6.1%), while CWSI is modulated by Land Surface Temperature (9.4%) and Sunshine Hours (7.8%). Partial correlation analyses controlling for topography and temperature confirm VPD's independent effect on drought severity. Large-scale oscillations, particularly Arctic Oscillation (AO) and ENSO-PDO interactions, significantly modulate interannual variability.This study demonstrates that VPD, rather than precipitation alone, governs drought evolution in this warming-wetting transition zone, emphasizing the necessity of complementary indices for water security assessment in arid regions.

Background: Single Chen’s chaotic system or Advanced Encryption Standard (AES) cannot provide sufficient anti-attack performance when separately used for medical image encryption. This study built a hybrid encryption framework combining four-dimensional (4D) Chen hyperchaotic system, Arnold transform, logistic map and AES-256 to fix security flaws of single encryption modules and protect privacy of clinical medical imaging data. Methods: A three-stage encryption workflow driven by logistic chaotic sequences was designed: (I) the logistic map ran 4,300 iterations (initial X=0.1, μ: 3.57–4.00, step =0.0001). The 100th and 1000th sequence values were Arnold mapping parameters; the 1200th, 1500th, 1800th and 2000th outputs served as four initial states of the 4D Chen hyperchaotic system to guarantee high chaotic parameter sensitivity; (II) Arnold transform realized pixel scrambling, followed by exclusive-OR (XOR) diffusion with Chen hyperchaotic sequences to eliminate intra-image pixel correlation; (III) we extracted 256 continuous bits (positions 2050–2305) from logistic sequences as the unique AES-256 secret key, and implemented multiround block encryption under AES Electronic Codebook (ECB) mode to boost cipher randomness. Simulations were run on MATLAB R2019b with two self-generated medical phantom images: a 743×720 liver lesion phantom (46.1 kB) and a 236×180 splenic trauma phantom (24.6 kB). Results: Quantitative metrics show excellent encryption performance. For liver lesion ciphertext, number of pixels change rate (NPCR) =99.66% and unified average changing intensity (UACI) =34.73%; splenic trauma ciphertext reached NPCR =99.81% and UACI =33.55%, both above the 99.6% and 33.3% security thresholds. Ciphertext χ² values were 275.5194 (liver lesion phantom) and 243.3932 (splenic trauma phantom), much lower than plaintext values (1.4977×10⁵, 4.4232×10⁶), proving uniform grayscale distribution. Ciphertext information entropy was 7.9953 (liver lesion phantom) and 7.9969 (splenic trauma phantom), nearly the ideal maximum of 8.00. Ciphertext pixel correlation coefficients (horizontal, vertical, diagonal) ranged −0.0042 to 0.0039, close to zero and far smaller than plaintext coefficients (>0.93). Compared with 2025–2026 state-of-the-art chaotic encryption schemes, our method achieved the best NPCR, UACI and correlation results. Extra tests confirmed strong resistance against chosen-plaintext, chosen-ciphertext, noise and occlusion attacks, with stable efficiency for computed tomography, magnetic resonance imaging, ultrasound and X-ray multimodal medical images. Conclusions: The proposed linked chaotic-AES hybrid algorithm unifies Arnold parameters, Chen hyperchaotic initial values and AES keys via logistic mapping, overcoming drawbacks of independent random module design. It achieves near-ideal randomness, ultra-low pixel correlation and strong multi-attack resistance versus standalone Chen chaos or AES. This lightweight, low-overhead scheme provides reliable end-to-end security for clinical medical image transmission and storage in smart healthcare systems.