Risk Decision Research Articles

Predicting High-Flow Nasal Cannula Oxygen Therapy Failure in Patients With Acute Hypoxaemic Respiratory Failure Using Machine Learning: Model Development and External Validation.

To develop and validate a prediction model for high-flow nasal cannula (HFNC) failure in patients with acute hypoxaemic respiratory failure (AHRF). AHRF accounts for a major proportion of intensive care unit (ICU) admissions and is associated with high mortality. HFNC is a non-invasive respiratory support technique that can improve patient oxygenation. However, HFNC failure, defined as the need for escalation to invasive mechanical ventilation, can lead to delayed intubation, prolonged mechanical ventilation and increased risk of mortality. Timely and accurate prediction of HFNC failure has important clinical implications. Machine learning (ML) can improve clinical prediction. Multicentre observational study. This study analysed 581 patients from an academic medical centre in Boston and 180 patients from Guangzhou, China treated with HFNC for AHRF. The Boston dataset was randomly divided into a training set (90%, n = 522) and an internal validation set (10%, n = 59), and the model was externally validated using the Guangzhou dataset (n = 180). A random forest (RF)-based feature selection method was used to identify predictive factors. Nine machine learning algorithms were selected to build the predictive model. The area under the receiver operating characteristic curve (AUC) and performance evaluation parameters were used to evaluate the models. The final model included 38 features selected using the RF method, with additional input from clinical specialists. Models based on ensemble learning outperformed other models (internal validation AUC: 0.83; external validation AUC: 0.75). Important predictors of HFNC failure include Glasgow Coma Scale scores and Sequential Organ Failure Assessment scores, albumin levels measured during HFNC treatment, ROX index at ICU admission and sepsis. This study developed an interpretable ML model that accurately predicts the risk of HFNC failure in patients with AHRF. Clinicians and nurses can use ML models for early risk assessment and decision support in AHRF patients receiving HFNC. TRIPOD checklist for prediction model studies was followed in this study. Patients were involved in the sample of the study.

Can an Expert Opinion Mitigate Racially Biased Diversion Decisions? An Empirical Examination in the Context of Re-Offense Risk Assessment

Using a vignette-based experimental design, this study tested whether the racial or ethnic identity of a justice-impacted person influences jury-eligible individuals’ perceptions of post-conviction risk, placement (incarceration vs. community supervision), and need for mandated treatment. Furthermore, we examined whether the inclusion of data from a reoffense risk tool could mitigate racial bias. Participants recruited through Amazon’s MTurk ( N = 448) were randomly assigned to conditions varying by race/ethnicity of an examinee (Black, Latino, White) and if/how risk assessment data were included (no risk data, risk data only, and risk data reported by a qualified examiner). There was no statistically significant effect of the examinee’s racial/ethnic identity on ratings of reoffense risk or placement decisions; however, the Black examinee was more likely to be mandated with treatment. Regardless of examinee identity, participants exposed to risk information proffered by an expert more often calibrated their risk decisions to the examinee’s actual risk level.

Fats and oils contaminants: an analysis of EU RASFF notifications from 1999 to 2023

Between 1999 and 2023, an extensive analysis of 713 notifications from the EU Rapid Alert System for Food and Feed (RASFF) related to fats and oils was conducted. The analysis covered the most frequently reported products, associated hazards, countries of origin, notification classifications, risk decisions, and actions taken in response. The data demonstrate considerable variability in RASFF notifications throughout the studied period, with significant spikes and a recent upward trend. Of all notifications, 44.32 % were categorized as alerts, and 23.42 % as border rejections. In terms of risk decisions, 34.08 % were deemed serious and 60.59 % remained undecided. Palm oil emerged as the most frequently notified product, representing 37.87 % of all notifications, followed by olive oil at 15.71 % and sunflower oil at 11.36 %. The Sudan 4 dye was identified as the most common hazard, constituting 30.01 % of all fats and oils notifications. Other notable hazards included benzo(a)pyrene (26.65 %), glycidyl esters (6.03 %), and chlorpyrifos (3.37 %). The top three countries of origin for RASFF notifications on fats and oils were Ghana (22.30 %), Ukraine (13.18 %), and Italy (9.96 %). These findings could benefit further research and influence decision-making, thereby enhancing food safety protocols in the global production, processing, and distribution of fats and oils.

Clinical and demographic factors determining patient fracture risk decision point (FRDP): The improving risk communication in osteoporosis (RICO) project.

This part of the Improving Risk Communication in Osteoporosis (RICO) study aims to understand patients' willingness to initiate osteoporosis treatment given a hypothetical fracture probability-derived from the FRAX® Risk Assessment Tool-and how age, fracture history, and numeric literacy may influence this. In 2022-2023, 332 postmenopausal women at risk of fracture were interviewed from nine countries to determine participants' Fracture Risk Decision Point (FRDP), the lowest probability of major osteoporotic fracture at which they would accept an osteoporosis medication. Participants' FRDP was evaluated given eight hypothetical 10-year FRAX scores. In countries with FRAX-based treatment thresholds, over half of the participants per country reported an FRDP that was below the threshold. Collectively, 79.2% demonstrated FRDPs at or below their respective threshold. Age and fracture history did not have a strong influence on FRDP; however, those who demonstrated higher levels of numeric literacy reported a significantly higher median FRDP (10%) compared to those who showed lower levels (5%, p < 0.001). Most patients were willing to accept an osteoporosis medication prescription at a hypothetical FRAX probability that was even lower than that of their nationally recommended treatment threshold. Literacy scores had a significant influence on FRDP whereas age and fracture history did not.

Assessment and forecasting of water ecological security and obstacle factor diagnosis in the Hexi Corridor of Northwest China

The water ecological security pattern is a core factor. A scientific, accurate, and practical evaluation of water ecological security provides a theoretical basis for regional water ecological management. Using water resource data from five cities in the Hexi Corridor of Gansu Province (Jiuquan (JQ), Jiayuguan (JYG), Zhangye (ZY), Jinchang (JC), and Wuwei (WW)) from 2006 to 2021, a water ecological security evaluation index system based on the PSR (pressure-state-response) framework was constructed, covering 27 factors related to water resources, socio-economics, and the ecological environment. The main obstacle factors of water ecological security were identified using the obstacle degree model, and the grey GM(1,1) model was employed to predict water ecological security. Results indicated that the comprehensive assessment index of water ecology in the Hexi Corridor increased from 2006 to 2021, showing a transition from relatively unsafe (0.319) to basic safety and then to relatively safe (0.672). The pressure and response systems were the main limiting factors affecting water ecological security in the Hexi Corridor. After a slight decline in 2008, the overall spatial distribution continued to rise, with WW City and ZY City leading since 2016. ZY had a higher safety grade proportion (25%) compared to other areas in the Hexi region. The pressure system was the most significant obstacle to water ecological security after 2006. Prediction results indicated that the comprehensive evaluation index of water ecological security would continue to rise annually from 2022 to 2031, reaching a very safe level by 2025. The evaluation results provide a scientific basis for ecological security and risk decision-making in the study area.

A case study of highly pathogenic avian influenza (HPAI) H5N1 at Bird Island, South Georgia: the first documented outbreak in the subantarctic region

ABSTRACT Capsule HPAI H5N1 was documented for the first time in the subantarctic region on Bird Island, South Georgia, resulting in the mortality of Brown Skuas Stercorarius antarcticus, Gentoo Penguins Pygoscelis papua, Snowy Albatrosses Diomedea exulans, and Antarctic Fur Seals Arctocephalus gazella. Aims The spread of the HPAI H5N1 subtype has had dramatic impacts on numerous populations of wild birds and mammals. We describe a case study that can inform the management of HPAI for conservation practitioners and researchers globally. Methods We documented the detection, monitoring, and impact of the first known outbreak of H5N1 HPAI in the subantarctic region, at Bird Island in South Georgia (−54.3582, −36.5112) during 2023–2024. Deaths from HPAI were first suspected in September 2023 and later confirmed by genetic analysis. Results In total, 77 Brown Skuas, 38 Gentoo penguins, and 58 Snowy Albatrosses were suspected to have died from HPAI infection, and HPAI was confirmed in 5 dead Antarctic Fur Seals. Total mortality was unknown for all species, as other individuals will have been scavenged before discovery, or died at sea. Conclusion This case study provides lessons for the management, risk, safety considerations, and ethical decisions regarding animal welfare that may help guide research and management responses to HPAI outbreaks elsewhere, particularly in remote areas or in species of conservation concern.

Industrial robots and corporate risk-taking value

This paper empirically examines the impact of industrial robot application on corporate risk-taking value using data from Chinese A-share listed manufacturing companies, measured by the Sharpe ratio. The findings indicate that industrial robots significantly enhance corporate risk-taking value through information effects, innovation, and production optimisation. Heterogeneity tests reveal that firms with higher capital density, non-state ownership, lower financing constraints, and more government subsidies experience a more substantial increase in their risk-taking value. These insights contribute to understanding the evolution of enterprise risk decisions in industrial intelligence and provide theoretical references for policies promoting industrial development.

Resilience Assessment of Deep Excavation Engineering Systems Based on Bayesian Network

As a key component of infrastructure, the construction safety of deep excavation is particularly important. In order to improve the safety and stability of Deep excavation construction, research on the risk resistance and rapid recovery capabilities of Deep excavation projects is carried out. By introducing the concept of resilience, we analyze the key factors that affect the safety and resilience of Deep excavation projects, use Bayesian networks to establish a resilience model that expresses the causal relationship between factors, and learn the parameters of the network model through data samples collected through surveys; by adjusting the network Analyze the relationship between each factor and system resilience at each node, and clarify the measures that should be taken and key control objects, thereby improving the system resilience and ensuring that normal construction can be carried out safely and orderly. Research shows that this model can be used to quantify the safety toughness value of deep excavation and reason and analyze the impact of various factors on the toughness value, thereby improving risk management and decision support capabilities.

The role of gene–environment interaction in the formation of risk attitudes

Understanding the formation of risk preferences is crucial for elucidating the roots of economic, social, and health inequalities. However, this area remains inadequately explored. This study employs a risk preference measure directly linked to the labor market to examine whether previous experiences with high unemployment rates influence current risk decision-making among the elderly, and whether this impact varies by genotype. The findings indicate that individuals with low genetic predispositions for risk tolerance are more significantly influenced by historical fluctuations in unemployment rates than those with high genetic predispositions for risk tolerance. Consequently, this paper identifies genetic endowment as a crucial moderating factor that shapes how past experiences impact current decision-making processes. This disparity in how past experiences shape risk preferences based on genetic predisposition may further amplify inequalities in health, wealth, income, and other outcomes associated with risk preferences.

Integrated risk management and maintenance planning in Oil and Gas Supply Chain operations under market uncertainty

The Oil and Gas Supply Chain (OGSC) is a multifaceted network comprising diverse activities and echelons. Instability or interruptions can cause economic fluctuations, impacting industries, markets, and consumers. Maintenance activities, which pause production but extend facilities' life, are recommended during non-peak demand periods to avoid production losses and meet customer demand. To mitigate these effects, decisions on operations planning, maintenance scheduling, and maintenance team assignments should be optimized in a risk management framework. The proposed model adopts a mixed-integer linear programming (MILP) framework and is solved via a sequential approach that incorporates the relax-and-fix (RF) heuristic in order to find a solution that is close to optimal. Subsequently, the solution serves as an initial solution for the CPLEX solver, which employs a branch-and-cut algorithm to attain the exact optimal solution. The practicality of this model has been showcased through its application to the supply chain in Saudi Arabia. The model efficiently schedules maintenance activities evenly and consistently across the OGSC plants over the planning period to reduce lost sales by keeping plants operational during high-demand periods. Furthermore, a sensitivity analysis was conducted to investigate the influence of the decision-maker's risk attitude on the outcomes that were obtained.

A New Application with Conditional Risk Matrix and Decision Rules: An Assessment of the Sustainability and Performance Electric Power Company

Abstract The 2030 Agenda for Sustainable Development of the United Nations General Assembly focuses on economic development, poverty cessation, environmental sustainability, and good governance at all levels, including peace and safety. For electric power distribution concessionary companies in Brazil, it is necessary to evaluate their sustainability and performance according to the parameters laid by the National Electric Power Agency (ANEEL). For that purpose, sixty-two indicators were assembled that were directly and indirectly associated with sustainability for 38 companies in the period–2013-2021. This study proposes a new application with a conditional risk matrix and decision rules aimed at the integrated assessment of sustainability and performance of electric power companies using the conditional probability of indicators in the ascent or descent of the companies’ classification. A set of “if..., then...” decision rules were inferred by the algorithm VC-DomLEM using concepts from the Rough Sets Theory and the Dominance principle, Dominance-based Rough Set Approach (DRSA).

An Integrated Quantitative Method of Risk Analysis and Decision Making for Safety Manipulation of a Forest Firefighting Aircraft by Using Physical Models of Multi-Hazard Factors

Although firefighting aircrafts are effective for fighting forest fires, they are affected by a variety of risks related to forest fires during flight operations such as thermal radiation, temperature rise, decrease in air density, and changes in updraft aerodynamics. However, quantifying these influences poses a challenge. Our research reveals that safety standards predominantly hinge on technical indicators like flight height, speed, and angle of attack, which makes it possible to quantify the above effects. A general method framework for analysing firefighting flight safety criteria was constructed in this thesis. Using relevant physical models, such as heat radiation, temperature, air density, and updraft in a forest fire environment, safety criteria with an analytical form were established. The corresponding safety constraints of each criterion were calculated quantitatively, and the most stringent constraint was determined based on multiple factors to obtain a comprehensive safety constraint boundary composed of the critical altitude, stall speed, and critical angle of attack. Finally, the effectiveness of the method is verified in a simulation case. The research results show that the constrained boundary calculation method proposed in this study can provide a scientific basis for the formulation of technical specifications for forest aviation fire safety and aviation fire task planning.

Financing Risks and Preventive Measures for E-Commerce Corporates

The e-commerce industry has achieved tremendous development in the past decade. However, in this rapid development process, there are some problems that have caused huge hidden dangers, and financing is the most important one among them. Therefore, the topic of this article is the dangers and prevention methods of e-commerce corporations under different financing methods. The method of this paper is as follows: first, segment the financing models of e-commerce corporates, then combine the financing situation of several well-known e-commerce corporates in China, analyze relevant data, and discuss the possible problems under these financing models separately. The research shows that the short-term debt payment risk, financial leverage risk and investment decision risk are the major risks that have influences on the firms’ financing, and the first two have a greater impact. To resolve these dangers, e-commerce firms can adjust debt structure, diversify financing channels and establish a sound risk prevention mechanism.

Individualized MRI-based stroke PRediction scOre using plaque Vulnerability for symptomatic carotid artEry disease patients (IMPROVE).

In TIA and stroke patients with carotid stenosis, estimations of future ipsilateral ischemic stroke risk and treatment decisions are currently primarily based on the degree of stenosis. Intraplaque hemorrhage (IPH), which can be readily visualized on carotid MRI, is increasingly established as an easy to assess and a very strong and independent predictor for ipsilateral stroke risk, stronger than any clinical risk factor. We developed a clinical prediction model (IMPROVE) incorporating IPH, degree of stenosis, and clinical risk factors to select patients with symptomatic carotid stenosis at high risk for stroke. IMPROVE was developed on pooled clinical and MRI data from five cohort studies of 760 recent TIA or minor stroke patients with carotid plaque who received optimal medical treatment. We used Cox proportional hazards models to determine the coefficients of IMPROVE. IMPROVE was internally validated using bootstrapping and converted to one- and three-year ipsilateral ischemic stroke risk. The development dataset contained 65 ipsilateral incident ischemic strokes that occurred during a median follow-up of 1.2 years (IQR: 0.5-4.1). The IMPROVE model includes five predictors, which are in order of importance: degree of stenosis, presence of IPH on MRI, classification of last event (cerebral vs ocular), sex, and age. Internal validation revealed a good accuracy (C-statistic: 0.82; 95% CI: 0.77-0.87) and no evidence for miscalibration (calibration slope: 0.93). Using presence of IPH on MRI and only four conventional parameters, the IMPROVE model provides accurate individual stroke risk estimates, which may facilitate stratification for revascularization.

Geotechnical risk modeling using an explainable transfer learning model incorporating physical guidance

While Artificial intelligence (AI) has been successfully applied in assessing geotechnical risk, such methods heavily rely on data quality to achieve satisfactory performance, and their results hardly can be interpreted due to their opaque design. With this in mind, this paper aims to address the following research gap: How can we accurately model geotechnical risks using limited data and domain knowledge, and efficiently explain the results of AI model? We develop a physics-guided transfer learning (PGTL) model to enhance the explainability and accuracy of geotechnical risk modeling. With the help of a physical model that simulates the tunnel excavation, a physics-guided dataset with 1000 samples is established and used to train a deep neural network. On these bases, transfer learning is adopted to fuse the features of physics mechanisms and monitoring data, constructing an explainable prediction model of geotechnical risk. To further support risk decision-making, feature relevance techniques are employed to assess the contribution of input parameters to risk. A shield tunnel construction in Wuhan is selected as a case to validate the effectiveness of the proposed method. The PGTL exhibits a more promising accuracy with R2 of 0.777 in contrast to three popular machine learning approaches, and provides insights into parameters that significantly induce risk, enhancing site managers’ understanding of tunnel construction and being conducive to tunnel safety.

Effectiveness of Updated 2023-2024 (Monovalent XBB.1.5) COVID-19 Vaccination Against SARS-CoV-2 Omicron XBB and BA.2.86/JN.1 Lineage Hospitalization and a Comparison of Clinical Severity-IVY Network, 26 Hospitals, October 18, 2023-March 9, 2024.

Assessing variant-specific COVID-19 vaccine effectiveness (VE) and severity can inform public health risk assessments and decisions about vaccine composition. BA.2.86 and its descendants, including JN.1 (referred to collectively as "JN lineages"), emerged in late 2023 and exhibited substantial divergence from co-circulating XBB lineages. We analyzed patients hospitalized with COVID-19-like illness at 26 hospitals in 20 U.S. states admitted October 18, 2023-March 9, 2024. Using a test-negative, case-control design, we estimated effectiveness of an updated 2023-2024 (Monovalent XBB.1.5) COVID-19 vaccine dose against sequence-confirmed XBB and JN lineage hospitalization using logistic regression. Odds of severe outcomes, including intensive care unit (ICU) admission and invasive mechanical ventilation (IMV) or death, were compared for JN versus XBB lineage hospitalizations using logistic regression. 585 case-patients with XBB lineages, 397 case-patients with JN lineages, and 4,580 control-patients were included. VE in the first 7-89 days after receipt of an updated dose was 54.2% (95% CI = 36.1%-67.1%) against XBB lineage hospitalization and 32.7% (95% CI = 1.9%-53.8%) against JN lineage hospitalization. Odds of ICU admission (adjusted odds ratio [aOR] 0.80; 95% CI = 0.46-1.38) and IMV or death (aOR 0.69; 95% CI = 0.34-1.40) were not significantly different among JN compared to XBB lineage hospitalizations. Updated 2023-2024 COVID-19 vaccination provided protection against both XBB and JN lineage hospitalization, but protection against the latter may be attenuated by immune escape. Clinical severity of JN lineage hospitalizations was not higher relative to XBB.

Climate storylines as a tool for interdisciplinary dialogue on risk decision-making: Analyzing a severe drought in southeastern South America

Recently, we are witnessing a growing interest in the study of climate narratives and storylines. These are seen as critical tools both for improving shared understandings involving different stakeholders (including scientists from different disciplines) and guiding initiatives for transformative adaptation actions and decision making. This paper approaches the storylines about a severe drought event that took place in southeastern South America (SESA) in 2008/2009 from an interdisciplinary dialogue between climatologists and anthropologists. This approach responds to the need for a more robust understanding of the event and the responses adopted locally, i.e: not restricted to physical aspects but considering socio-economic and political aspects, with the aim of contributing to the development of socially relevant climatic information for realistic climate risk decision making contexts. In particular it intends to contribute to the decision-making processes of government agencies focused on climate communication and public climate services, such as national and weather services and regional agencies. Considering disciplinary differences, we account for the interpretative polyphony regarding this event, i.e.: what each actor/sector is talking about (for what and to whom they are talking about) when referring to "the drought", and analyze climate storylines as an agonistic field of meanings. This implies recognizing the disputes and contradictions among them, how different storylines promote certain lines of argumentation and meanings to the detriment of others, knowledge gaps, the role of social power relations, socio-political order factors and their impact in terms of the necessary transformations towards an environmentally sustainable and socially just horizon.

Interpretable feature-temporal transformer for short-term wind power forecasting with multivariate time series

The inherent randomness and volatility of wind power generation present significant challenges to the reliable and secure operation of the power system. Therefore, it is crucial to have interpretable wind power forecasting (WPF) to ensure seamless grid integration and effective risk assessment. Existing forecasting models often focus on improving WPF performance and ignore the interpretability of the model, resulting in ambiguous forecasting. In this paper, the interpretable feature-temporal transformer (IFTT) for short-term wind power forecasting with multivariate time series is presented. The model uses an encoder-decoder architecture to effectively integrate historical information and future prior information from multiple variables. The designed decoupled feature-temporal self-attention (DFTA) module and variable attention network (VAN) effectively realize the interpretability of temporal information and multi-variable inputs while extracting important features. The Auxiliary Forecasting Network (AFN) plays a key role in providing pseudo-future wind speed predictions, which serve as an essential input for the model's decoder, and enhancing forecasting accuracy through multi-task learning. Experimental results on multiple datasets in different geographical locations show that the proposed algorithm is superior to various advanced methods. Besides, the interpretability of the IFTT model offers valuable insights for ensuring the safety of wind power utilization and supporting informed risk decision-making.

Risk prediction and prognostic analysis of post-implantation syndrome after thoracic endovascular aortic repair

This study aimed to establish a predictive model for the risk of post-thoracic endovascular aortic repair (TEVAR) post-implantation syndrome (PIS) in type B aortic dissection (TBAD) patients, assisting clinical physicians in early risk stratification and decision management for high-risk PIS patients. This study retrospectively analyzed the clinical data of 547 consecutive TBAD patients who underwent TEVAR treatment at our hospital. Feature variables were selected through LASSO regression and logistic regression analysis to construct a nomogram predictive model, and the model's performance was evaluated. The optimal cutoff value for the PIS risk nomogram score was calculated through receiver operating characteristic (ROC) curve analysis, further dividing patients into high-risk group (HRG) and low-risk group (LRG), and comparing the short to midterm postoperative outcomes between the two groups. In the end, a total of 158 cases (28.9%) experienced PIS. Through LASSO regression analysis and multivariable logistic regression analysis, variables including age, emergency surgery, operative time, contrast medium volume, and number of main prosthesis stents were selected to construct the nomogram predictive model. The model achieved an area under the curve (AUC) of 0.86 in the training set and 0.82 in the test set. Results from calibration curve, decision curve analysis (DCA) and clinical impact curve (CIC) demonstrated that the predictive model exhibited good performance and clinical utility. Furthermore, after comparing the postoperative outcomes of HRG and LRG patients, we found that the incidence of postoperative PIS significantly increased in HRG patients. The duration of ICU stay and mechanical assistance time was prolonged, and the incidence of postoperative type II entry flow and acute kidney injury (AKI) was higher. The risk of aortic-related adverse events (ARAEs) and major adverse events (MAEs) at the first and twelfth months of follow-up also significantly increased. However, there was no significant difference in the mortality rate during hospitalization. This study established a nomogram model for predicting the risk of PIS in patients with TBAD undergoing TEVAR. It serves as a practical tool to assist clinicians in early risk stratification and decision-making management for patients.

Evolution Characteristics of Meteorological and Hydrological Drought in an Arid Oasis of Northwest China

In the context of global warming, the acceleration of the water cycle increases the risk of meteorological drought (MD) and hydrological drought (HD) in the arid region of Northwest China. The Manas River Basin is a typical agricultural oasis and the largest oasis farming area in Xinjiang, Northwest China. Droughts in this basin have significant implications for both agricultural production and the livelihoods of inhabitants. To evaluate the MD and HD and provide information for drought relief in the MRB, the standardized precipitation evapotranspiration index (SPEI) and standardized runoff index (SRI) were calculated using long-term rainfall and runoff data. Subsequently, combined with ArcGIS 10.3 software and the trend analysis method, the SPEI and SRI characteristics were evaluated at different time scales (1-, 3-, 6-, and 12-month). There were three main findings. First, both MD and HD were alleviated, with significantly more HD alleviation. MDs in spring and autumn exhibited a trend of aggravation. The SRIs in summer, autumn, and winter increased significantly at a confidence level of p &lt; 0.01, with an insignificant decline in spring. In the 2010s, the frequency of light drought of MD was stable at 10% to 20%, while severe and extreme droughts increased. The frequency of HDs has decreased since the 1990s. Second, on annual and seasonal scales, MDs occurred mainly as light and moderate droughts. The highest frequency of MD was 24% of moderate droughts in winter. Spatially, the northern region of the MRB was characterized by more frequent light and extreme droughts. Third, runoff in the Manas River Basin increased significantly during the 1990s, which may have been related to the acceleration of glacial retreat in the Tianshan Mountains. This study can effectively reveal the changes in meteorological and hydrological drought in NWC and provide the basis for risk decision-making and management for watershed managers.