Annals of Operations Research最新文献

In healthcare, especially within intensive care units (ICU), informed decision-making by medical professionals is crucial due to the complexity of medical data. Healthcare analytics seeks to support these decisions by generating accurate predictions through advanced machine learning (ML) models, such as boosted decision trees and random forests. While these models frequently exhibit accurate predictions across various medical tasks, they often lack interpretability. To address this challenge, researchers have developed interpretable ML models that balance accuracy and interpretability. In this study, we evaluate the performance gap between interpretable and black-box models in two healthcare prediction tasks, mortality and length-of-stay prediction in ICU settings. We focus specifically on the family of generalized additive models (GAMs) as powerful interpretable ML models. Our assessment uses the publicly available Medical Information Mart for Intensive Care dataset, and we analyze the models based on (i) predictive performance, (ii) the influence of compact feature sets (i.e., only few features) on predictive performance, and (iii) interpretability and consistency with medical knowledge. Our results show that interpretable models achieve competitive performance, with a minor decrease of 0.2–0.9 percentage points in area under the receiver operating characteristic relative to state-of-the-art black-box models, while preserving complete interpretability. This remains true even for parsimonious models that use only 2.2 % of patient features. Our study highlights the potential of interpretable models to improve decision-making in ICUs by providing medical professionals with easily understandable and verifiable predictions.

In this work, we propose a new variant of construct, merge, solve, and adapt (CMSA), which is a recently introduced hybrid metaheuristic for combinatorial optimization. Our newly proposed variant, named reinforcement learning CMSA (RL-CMSA), makes use of a reinforcement learning (RL) mechanism trained online with data gathered during the search process. In addition to generally outperforming standard CMSA, this new variant proves to be more flexible as it does not require a greedy function for the evaluation of solution components at each solution construction step. We present RL-CMSA as a general framework for enhancing CMSA by leveraging a simple RL learning process. Moreover, we study a range of specific designs for the employed learning mechanism. The advantages of the introduced CMSA variant are demonstrated in the context of the far from most string and minimum dominating set problems, showing the improvement in performance and simplicity with respect to standard CMSA. In particular, the best performing RL-CMSA variant proposed is statistically significantly better than the standard algorithm for both problems, obtaining 1.28% and 0.69% better results on average respectively.

Over recent years, scholarly interest in universities’ allocation and effective utilisation of financial resources has been growing. When used efficiently, financial resources may improve universities’ quality of research and teaching, and therefore their positions in world university rankings. However, despite the relevance of financial efficiency to university placement in academic rankings, universities’ total available financial resources appear much more significant. In the present study, we propose an innovative methodology to determine realistic ranking targets for individual universities, based on their available financial resources. In particular, we combine data envelopment analysis, as developed by Banker et al. (Manag Sci 30(9):1078–1092, 1984), and a directed Louvain community detection algorithm to examine 318 universities across five countries, considering their ARWU scores alongside key financial indicators (i.e., long-term physical capital, total operating revenues). We identify clusters of universities with similar financial profiles and corresponding ARWU scores, as well as universities that have optimised their use of financial resources, representing benchmarks for similar universities to emulate. The approach is subsequently applied to Italian universities, as a specific national case. The findings may be useful for policy makers and university managers seeking reliable strategies for climbing academic rankings, particularly in countries with limited public investment in higher education.

As one source of user-generated content, online reviews embed vast quantities of important business information, significantly affecting consumer demand. In this study, we aim to propose a new forecasting approach to predict the demand for new energy vehicles (NEVs) by incorporating perceived quality measures extracted from online reviews into the traditional Bass model. To this end, we consider three crucial dimensions (i.e., emotional experience, defect perception, and brand/product image) and adopt text analysis techniques to mine perceived quality information from online reviews for NEVs comprehensively. Coping with the limited datasets, we further dynamically incorporate the mined perceived quality into the Bass model to improve the accuracy of new energy vehicle (NEV) demand forecasting. Finally, we meticulously conduct a series of experiments with crawled online reviews and historical sales of distinct NEV models. The experimental results demonstrate that the perceived quality measures identified from online reviews jointly affect the consumers’ purchasing decisions, and effectively enhance the performance of the NEV demand forecasting. Furthermore, some interesting and important findings are achieved based on the proposed methodology, including the time-lag effect of perceived quality on consumers’ purchasing decisions and the formulation of specific product strategies based on demand trends.

The manufacturer-rebate where the manufacturer directly provides consumer rebates and the channel-rebate where the manufacturer stimulates downstream retailers to sell more products with channel rebates are two typical rebate strategies for manufacturers to increase sales. Considering the prevalence of rebate promotions and the downstream retailer’s optimism on the effect of rebating consumers, we incorporate rebate promotions, retailers’ optimism, and the corresponding information asymmetry issue into the manufacturer’s wholesale-ordering contract design problem. We find that with rebate promotions, the wholesale price and the order quantity should be higher than without rebate promotion. Specially, under the manufacturer-rebate strategy, the wholesale price and the order quantity should increase with the manufacturer’s rebate and the retailer’s optimism. However, under the channel-rebate strategy, the order quantity (the wholesale price) is no longer affected by the manufacturer’s rebate (the retailer’s optimism). Moreover, when the retailer’s optimism is private information, the retailer’s information distortion behaviors under the two rebate strategies are similar, but the manufacturer should accept different kinds of wholesale price-order quantity contract menus to reveal the retailer’s information and maximize profits. Specifically, under the manufacturer-rebate strategy (the channel-rebate strategy), differentiated contracts (a pooling contract) to different types of retailers are optimal. Besides, under the manufacturer-rebate strategy, the value of the rebate can play a moderating role in contract design. Hence, when the rebate is exogenously decided, the manufacturer can punish the retailer who is likely to distort information by decreasing the rebate.