Accurate delineation of liver tumors is the key to medical image-guided precision radiotherapy. In this paper we propose and study an automatic biological target delineation algorithm based on RASUnet. We propose innovative optimization schemes at the data set level and network architecture level, and verify the effect of the improved scheme through experiments. We apply the residual idea to the Unet network architecture to form the ResUnet network, which speeds up the network convergence process. To introduce an attention gate for ResUnet, we form a RAUnet network, which can “emphasize regions of interest and suppress irrelevant regions” on the input image. Finally, we add the compression activation mechanism to form the ResSE optimization module, and improve the network to RASUnet. This method brings a weighting mechanism acting on the feature channel, which further improves the attention performance. The experimental results show that RASUnet has the best accuracy in the optimal delineation, and has high accuracy and robustness in the automatic segmentation and delineation of tumor biological targets.

Strengthening international collaboration is essential to achieving the United Nations’ SDGs. The Group of Seven (G7) is recognized for acting and enhancing cooperation to achieve the SDGs. However, the current understanding of G7’s cooperation is rather subjective without quantitative measurements. Here we show a comprehensive and quantitative analysis of G7’s cooperation with regards to the economic and environmental SDGs over the period of 2000-2020. The results suggest that G7 countries have all contributed positively to economic indicators thanks to their closely binding relationship. By contrast, significant discrepancies and uncooperative performances in environmental indicators have been revealed. Particularly, Canada and Germany have shown considerable negative synergy contributions to environmental indicators, which might offset the positive contributions brought by France and Italy and lead to an overall negative synergy. Our results highlight the need for further collaboration among G7 to tackle emerging environmental issues, such as climate change and shrinking biodiversity. Wang and colleagues assess the G7’s collaboration in achieving the SDGs from 2000–2020. They find that all G7 members contribute positively economics-related SDGs, but performance on environmental goals shows less collaboration and discrepancies in contributions.

In order to build a low-risk Fund of Funds (FOF), from the perspective of correlation, the principal component factor is used to improve the traditional risk parity model. Principal component analysis is used to decompose the underlying assets and generate unrelated principal component factors, and then the authors can construct a principal component risk parity portfolio. The proposed empirical results based on China’s mutual fund market show that the performance of principal component risk parity portfolio (PCRPP) is better than that of equal weight portfolio (EWP) and traditional risk parity portfolio (RPP). That is to say, not only the PCRPP in this paper has much lower risk than EWP and RPP, but also slightly better than EWP and RPP in terms of average return. Moreover, the study of dividing the underlying assets shows that the PCRPP in this paper is not sensitive to the underlying assets. The PCRPP in this paper is better than EWP and RPP for both the better performing funds and the worse performing funds. In addition, the empirical results on dynamic portfolio adjustments show that it is not appropriate to adjust asset allocation too frequently when the expected rate of return is calculated using the arithmetic mean.

Accurate prediction of potato yield is essential for informed agricultural decision-making, ensuring food security, and supporting farmers’ livelihoods. This is particularly critical in regions like Prince Edward Island (PEI), where potato production is not only a staple of local agriculture but also a cornerstone of the regional economy, accounting for a significant proportion of agricultural revenue and employment. Although machine learning algorithms have been extensively applied in agricultural yield prediction, previous studies have not fully leveraged the potential of capturing both short- and long-term dependencies. This research highlights the efficacy of integrating these temporal dependencies into machine learning models to enhance the accuracy of potato yield predictions. The methodology adopted in this research, including data collection, model selection, and scenario-based projections, can be applied to other regions and crops. Our projections for PEI toward the end of the century indicate a substantial decline in potato yields across different climate scenarios. Under the high-emission SSP5-8.5 scenario, our models predict a potential potato yield reduction of up to 70%. In contrast, the SSP1 and SSP2 scenarios suggest a more moderate decline in potato yield, ranging from 4% to 15%. These findings underscore the urgent need for reducing greenhouse gas emissions to mitigate the adverse impacts on potato production. Furthermore, they highlight the importance of implementing adaptive farming practices to sustain potato yield in the face of climate change.

OBJECTIVES:To conduct a comprehensive analysis of demographic information, medical history, and blood pressure (BP) and heart rate (HR) variability during hospitalisation so as to establish a predictive model for preoperative in-hospital mortality of patients with acute aortic dissection (AD) by using machine learning techniques.DESIGN:Retrospective cohort study.SETTING:Data were collected from the electronic records and the databases of Shanghai Ninth People's Hospital Affiliated to Shanghai Jiao Tong University School of Medicine and the First Affiliated Hospital of Anhui Medical University between 2004 and 2018.PARTICIPANTS:380 inpatients diagnosed with acute AD were included in the study.PRIMARY OUTCOME:Preoperative in-hospital mortality rate.RESULTS:A total of 55 patients (14.47%) died in the hospital before surgery. The results of the areas under the receiver operating characteristic curves, decision curve analysis and calibration curves indicated that the eXtreme Gradient Boosting (XGBoost) model had the highest accuracy and robustness. According to the SHapley Additive exPlanations analysis of the XGBoost model, Stanford type A, maximum aortic diameter >5.5 cm, high variability in HR, high variability in diastolic BP and involvement of the aortic arch had the greatest impact on the occurrence of in-hospital deaths before surgery. Moreover, the predictive model can accurately predict the preoperative in-hospital mortality rate at the individual level.CONCLUSION:In the current study, we successfully constructed machine learning models to predict the preoperative in-hospital mortality of patients with acute AD, which can help identify high-risk patients and optimise the clinical decision-making. Further applications in clinical practice require the validation of these models using a large-sample, prospective database.TRIAL REGISTRATION NUMBER:ChiCTR1900025818.

This paper focuses on capturing the impacts of leptokurtic phenomenon and heterogeneous preferences in higher moments on asset allocation. To achieve this, we propose a utility maximization asset allocation framework under the multivariate affine generalized hyperbolic (MAGH) asset prices dynamics. With the investor’s preference given by the exponential utility, we derive the closed-form optimal asset allocations for mixed multivariate affine normal inverse Gaussian-normal model and mixed multivariate affine variance gamma-normal model, which covers Markowitz’s mean–variance model as our special case. Extensive empirical studies are conducted to assess the effectiveness of the proposed asset allocation models relative to other portfolio strategies based on the Markowitz’s mean–variance theory and the equally weighted 1/N rule. Using the out-of-sample Sharpe ratio, the certainty-equivalent return, quantile and tail metrics as the performance measures, the proposed methods are found to be very effective and robust.

Accurate land use land cover (LULC) classification is vital for the sustainable management of natural resources and to learn how the landscape is changing due to climate. For accurate and efficient LULC classification, high-quality datasets and robust classification methods are required. With the increasing availability of satellite data, geospatial analysis tools, and classification methods, it is essential to systematically assess the performance of different combinations of satellite data and classification methods to help select the best approach for LULC classification. Therefore, this study aims to evaluate the LULC classification performance of two commonly used platforms (i.e., ArcGIS Pro and Google Earth Engine) with different satellite datasets (i.e., Landsat, Sentinel, and Planet) through a case study for the city of Charlottetown in Canada. Specifically, three classifiers in ArcGIS Pro, including support vector machine (SVM), maximum likelihood (ML), and random forest/random tree (RF/RT), are utilized to develop LULC maps over the period of 2017–2021. Whereas four classifiers in Google Earth Engine, including SVM, RF/RT, minimum distance (MD), and classification and regression tree (CART), are used to develop LULC maps for the same period. To identify the most efficient and accurate classifier, the overall accuracy and kappa coefficient for each classifier is calculated throughout the study period for all combinations of satellite data, classification platforms, and methods. Change detection is then conducted using the best classifier to quantify the LULC changes over the study period. Results show that the SVM classifier in both ArcGIS Pro and Google Earth Engine presents the best performance compared to other classifiers. In particular, the SVM in ArcGIS Pro shows an overall accuracy of 89% with Landsat, 91% with Sentinel, and 94% with Planet. Similarly, in Google Earth Engine, the SVM shows an accuracy of 87% with Landsat 8 and 92% with Sentinel 2. Furthermore, change detection results show that 13.80% and 14.10% of forest areas have been turned into bare land and urban class, respectively, and 3.90% of the land has been converted into the urban area from 2017 to 2021, suggesting the intensive urbanization. The results of this study will provide the scientific basis for selecting the remote sensing classifier and satellite imagery to develop accurate LULC maps.

LAY SUMMARY This research explored the demographic, military service, and health characteristics associated with cannabis for medical purposes (CMP) reimbursements among Veterans Affairs Canada (VAC) clients and respondents of the Life After Service Survey 2016 (LASS). Of the initial number of indicators selected contained in the LASS 2016, some specific variables were significantly associated with CMP reimbursement, from which physical/mental health and well-being indicators, such as anxiety, posttraumatic stress disorder (PTSD), depression, bowel ulcer, traumatic brain injury, chronic pain, needing help with tasks, psychological distress, and having three or more conditions of the PTSD diagnosis, were positively associated with CMP. Moreover, unemployment, having low income (< $5,000), a difficult adjustment, being very dissatisfied with life, having low social support, a weak community belonging, and reporting high stress also increased the odds of being reimbursed. These results will help to identify a preliminary profile of VAC clients with higher need for CMP reimbursement.

Objectives Hypertension is a predominant risk factor for aortic dissection (AD), and blood pressure (BP) control plays a vital role in the management of AD. However, the correlation between BP change and the prognosis for AD remains unclear. This study aims to demonstrate the impact of BP change patterns on AD prognosis. Methods This retrospective study included AD patients at two institutions (Shanghai Ninth People's Hospital Affiliated with Shanghai Jiao Tong University School of Medicine and the Vascular Department of the First Affiliated Hospital of Anhui Medical University) between 2004 and 2018. The systolic BP (SBP) change patterns of these patients were analyzed by functional data analysis (FDA). The relationship between BP change patterns and the risk of adverse events (AEs) was assessed using survival analysis. Results A total of 458 patients with AD were eligible for analysis. The logistic regression analysis indicated that compared with that in patients with low SBP variation (SBPV), the incidence of AEs in patients with high SBPV was significantly higher (35.84 vs. 20.35%, OR 2.19, P < 0.001). The patients were divided into four categories (accelerating rise, accelerating drop, decelerating rise, and decelerating drop) based on their SBP patterns after FDA fitting. The results of Kaplan–Meier analysis showed that at the 15- and 20-min time points, the incidence of AEs in the decelerating-drop group was significantly lower than that in the accelerating-rise group (OR 0.19, P = 0.031 and OR 0.23, P = 0.050). However, at the 25- and 30-min time points, the difference between these four groups was not significant (OR 0.26, P = 0.08 and OR 0.29, P = 0.10). Conclusions This study classified AD patients into four groups according to the SBP change patterns the first 30 min following admission, of which those with accelerating rises in SBP are at the highest risk of AEs, while those with decelerating drops have the best prognosis in the first 24 h after admission. Clinical practitioners may benefit from analyzing patterns of in-hospital SBP.

The majority of insurance investment funds are derived from policy liability debt funds. It differs from other institutional investors in a number of ways, including investment size, horizon, duration, risk, and so on. However, only a small portion of the extant literature focuses on in-depth and extensive analysis of Insurance Institutional Investors' holdings (IIIs). This study analyses the impact of shareholding by insurance institutions on the value of Shanghai and Shenzhen A-share listed companies in China's capital market. The paper offers three major contributions. First, we discovered that long-term equity-holding IIIs have both value selection and value creation functions. Second, the value creation function becomes more significant among long-term stock-holding IIIs with an increase in the period during which they retain the company's shares; Third, fast-in and fast-out (FIFO) IIIs have a value-inhibiting effect on the held company and serve a value selection role, rather than a value creation function. This study provides more insight on the lack of academic interest in insurance institutions and serves as a foundation and reference for the design of regulatory policies for insurance institutions' involvement in stock markets. It also gives empirical evidence for corporations to accurately analyze shareholding by insurance institutions. Furthermore, since this study concentrates on China's capital market, it can serve as a benchmark for other nations, particularly, those designated as developing market economies.

The two opposing investment strategies, diversification and concentration, have often been directly compared. While there is much less dispute regarding Markowitz’s approach as the benchmark for diversification, the precise meaning of concentration in portfolio selection remains unclear. This paper offers a novel definition of concentration, along with an extreme value theory-based estimator for its implementation. When overlaying the performances derived from diversification (in Markowitz’s sense) and concentration (in our definition), we find an implied risk threshold, at which the two polar investment strategies reconcile—diversification has a higher expected return in situations where risk is below the threshold, while concentration becomes the preferred strategy when the risk exceeds the threshold. Different from the conventional concave shape, the estimated frontier resembles the shape of a seagull, which is piecewise concave. Further, taking the equity premium puzzle as an example, we demonstrate how the family of frontiers nested inbetween the estimated curves can provide new perspectives for research involving market portfolios.

Markov switching regime models have played an increasingly important role in finance and economics, especially for business cycles and long swings in currencies. Regime-switching models provide a simple way to capture stochastic volatility and thus overcomes the drawback of the classical lognormality assumption characterized by constant volatility. This paper considers multivariate Black and Scholes type models with a Markov regime-switching mechanism. We show that the pricing of some multivariate derivatives under models where the Markov chain has two or three states, can be approximated accurately in closed-form, based on linear and quadratic Taylor polynomials. Closed form approximation methods are computationally advantageous as they perform in constant time, compared with alternative methods such as Monte-Carlo, where the accuracy of the estimation is directly linked to the number of executed simulations.

It is difficult to segment Glioma and its internal structure because the Glioma boundaries have edemas and complex internal structures. This paper proposes a new optimized, integrated 3D U-Net network to achieve accurate segmentation of Glioma and internal subareas. The contribution of this paper is twofold, it studies the clinical path of patients with Glioma and constructs an optimized 3D U-Net deep learning algorithm by combining them with the radiologic feature set. The proposed model was validated in the published Glioma operation data set of multi-modal MRI resonance images and clinicians manual segmentation data. The model can accurately segment the MRI multi-modality images of Glioma and intra-tumour nodes and achieve the multi-modality prediction of the overall survival period of patients. The experimental results further indicated that the segmentation accuracy of the proposed method was higher than other sophisticated methods. The Dice similarity coefficients of the whole tumor (WT) region, the core tumor (CT) region, and the augmentation / enhanced tumor (ET) region, were 0.9632, 0.8763, and 0.8421, respectively, which are better than the clinical experts’ manual segmentation results. Hence, this research can effectively promote the development of deep learning clinical precise diagnosis and medical technology for Glioma.

Aiming at the difference in the segmentation performance of the three segmentation target regions in the glioma image segmentation task based on the fully convolutional neural network, we propose a comprehensive evaluation method of neural network performance based on four evaluation indices. In addition, we analyze the performance and characteristics of neural network in the segmentation task of glioma, study the segmentation performance of neural network in the whole tumor (WT), tumor core (TC) and enhanced tumor (ET) regions, and propose a deep learning algorithm based on multiple networks in parallel. In this paper, the input image of the two-dimensional neural network is sliced, and the input of the three-dimensional neural network is processed in two ways: overlapping and non-overlapping, and in the image post-processing part, the three-dimensional image is reconstructed before the evaluation index is calculated. This article uses four evaluation indexes, which are Dice, Sensitivity, PPV, and Hausdorff, for the three segmentation target regions, and performs RSR* weight calculation, and finally performs a comprehensive evaluation. Experimental results show that Vnet has the best comprehensive segmentation performance, FCN-8s has the best segmentation performance in the TC area, Unet++ has the best segmentation performance in the ET area, and Vnet has the best segmentation performance in the WT area. Based on this, we propose a FUV multi-network parallel algorithm, combined with a reverse attention mechanism to improve the segmentation accuracy of the three segmentation target regions.

This paper analyzes the impact of executives' hometown identification on corporate environmental social responsibility (CESR) using a sample of Chinese A-share-listed companies from 2007 to 2018. It finds that: the CESR scores of companies are higher when executives work in their hometowns, indicating that executives' hometown identification significantly improves the fulfillment of CESR; mechanism tests show that the above relationship is more significant in regions with superior environmental quality, indicating that executives take CESR more seriously in their hometowns more due to social pressure; further tests found that executive characteristics, such as executive type and age, have a regulating effect on this relationship. In addition, the nature of property rights of listed companies also affects executives' hometown identification. Executives of state-owned enterprises have a stronger hometown identification, which enhances the fulfillment of CESR to a higher extent. In the context of the micro level of the enterprise, this paper provides positive evidence that an informal system, named as "hometown identity", can enhance the performance of CESR and the pressure effect implicitly behind the social network, which enriches and expands the research related to CESR fulfillment.

Variable annuities are long-term insurance products that offer a large variety of investment-linked benefits, which have gained much popularity in the last decade. Accurate valuation of large variable annuity portfolios is an essential task for insurers. However, these products often have complicated payoffs that depend on both of the policyholder's mortality risk and the financial market risk. Consequently, their values are usually estimated by computationally intensive Monte Carlo simulation. Simulating large numbers of sample paths from complex dynamic asset models is often a computational bottleneck. In this study, we propose and analyze three Quasi-Monte Carlo path generation methods, Cholesky decomposition, Brownian Bridge, and Principal Component Analysis, for the valuation of large VA portfolios. Our numerical results indicate that all three PGMs produce more accurate estimates than the standard Monte Carlo simulation at both the contract and portfolio levels.

This article studied the social network of fund managers based on their historical working relationships, and investigated how these relations influenced their information sharing as well as the fund performance. We found that (i) An investment portfolio with a high degree of information sharing had a high average return and Sharpe ratio. (ii) Fund managers' degree of centrality in social networks had a significant positive effect on their information sharing and trading behaviours.

Momentum as elaborated under a recent novel definition has been shown quantitatively to have a significant impact on basketball game outcomes. This paper makes two contributions to the analytical literature on sports momentum: (1) two aspects of the new definition are operationalized so that its practicality becomes evident; and (2) through a dimension-reduction technique (elastic net), key factors associated with momentum are identified. Both technical variables such as field goals, assists, rebounds, etc. and environmental variables such as the spectator attendance rate and player salary dispersion are considered, and the potential for useful real-time analyzes is illustrated.

The valuation of large variable annuity (VA) portfolios is an important problem of interest, not only because of its practical relevance but also because of its theoretical significance. This is prompted by the phenomenon that many existing sophisticated algorithms are typically efficient at valuing a single VA policy but they are not scalable to valuing large VA portfolios consisting of hundreds of thousands of policies. As a result, this sparks a new research direction exploiting machine learning methods (such as data clustering, nearest neighbor kriging, neural network) on providing more efficient algorithms to estimate the market values and sensitivities of large VA portfolios. The idea underlying these approximation methods is to first determine a set of VA policies that is "representative" of the entire large VA portfolio. Then the values from these representative VA policies are used to estimate the respective values of the entire large VA portfolio. A substantial reduction in computation time is possible because we only need to value the representative set of VA policies, which typically is a much smaller subset of the entire large VA portfolio. Ideally the large VA portfolio valuation method should adequately address issues such as (1) the complexity of the proposed algorithm; (2) the cost of finding representative VA policies; (3) the cost of the initial training set, if any; (4) the cost of estimating the entire large VA portfolio from the representative VA policies; (5) the computer memory constraint; and (6) the portability to other large VA portfolio valuation. Most of the existing large VA portfolio valuation methods do not necessary reflect all of these issues, particularly the property of portability, which ensures that we only need to incur the start-up time once and the same representative VA policies can be recycled to valuing other large portfolios of VA policies. Motivated by their limitations and by exploiting the greater uniformity of the randomized low discrepancy sequence and the Taylor expansion, we show that our proposed method, a green mesh method, addresses all of the above issues. The numerical experiment further highlights its simplicity, efficiency, portability, and, more important, its real-time valuation application.