Expert Systems with Applications最新文献

{"title":"Frequency domain transfer learning for remote sensing visual question answering","authors":"Ze Zhang ,&nbsp;Enyuan Zhao ,&nbsp;Ziyi Wan ,&nbsp;Xinyue Liang ,&nbsp;Min Ye ,&nbsp;Jie Nie ,&nbsp;Lei Huang","doi":"10.1016/j.eswa.2025.128395","DOIUrl":"10.1016/j.eswa.2025.128395","url":null,"abstract":"<div><div>Remote Sensing Visual Question Answering (RSVQA) aims to parse the content of remote sensing images through multimodal interaction to accurately extract scientific knowledge. Current RSVQA methods typically fine-tune pre-trained models on specific datasets, which overlooks the mining of complex structured information in remote sensing images that are highly coupled with color, scale, and semantics. Additionally, these methods lack adequate handling of overfitting issues caused by the high complexity and noise attributes of remote sensing data, resulting in predictions that are neither comprehensive nor accurate. To mitigate this issue, this paper proposes a Parameter-Efficient Transfer Learning (PETL) method based on the frequency domain. By leveraging Fourier Transform, it captures the intricate structural information of complex remote sensing and enhances the generalizability across domains, data, and models. The main contributions of this paper are as follows: 1) We introduce an efficient and stable X-PFA framework for Remote Sensing Visual Question Answering (RSVQA). Here, ‘X’ denotes pretrained VLP models, and ‘PFA’ stands for Primary Frequency Adapter, which performs a Fast Fourier Transform (FFT) over the intermediate spatial domain features at each layer to produce the corresponding frequency representation, including an amplitude component (encoding scene-perceptual style such as texture, color, scene contrast) and a phase component (encoding rich semantics). The PFA adapts to the specific dataset distribution by learning salient features in the frequency domain. 2) Our proposed framework demonstrates stable and excellent performance across various pre-trained models, significantly mitigating overfitting issues on small datasets. On average, accuracy improves by 0.62 %, and stability increases by 28.5 %. 3) The PFA contains only 17.2 million trainable parameters. Compared to full-parameter fine-tuning, our approach reduces training time by 54.9 %, resulting in substantial training cost savings.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"291 ","pages":"Article 128395"},"PeriodicalIF":7.5,"publicationDate":"2025-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144280396","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Hybrid contextual and sentiment-based machine learning model for identifying depression risk in social media","authors":"Nha Tran ,&nbsp;Phi Ta ,&nbsp;Hung Nguyen ,&nbsp;Hien D. Nguyen ,&nbsp;Anh-Cuong Le","doi":"10.1016/j.eswa.2025.128505","DOIUrl":"10.1016/j.eswa.2025.128505","url":null,"abstract":"<div><div>Depression is a dangerous and widespread mental disorder globally, often leading to feelings of low self-esteem, hopelessness, and suicide. With the rapid development of social media platforms, they have become spaces for people to share experiences and emotions and relieve stress and fatigue. Consequently, detecting depression on social media has become meaningful and consistent with development trends. However, it faces significant challenges due to the unstructured nature of social media data and the complex interaction of linguistic signals, context, and sentiment. In this paper, a novel model for detecting depressive posts on social media is proposed, called <em>CLSDepDet</em>. This model leverages effective feature extraction techniques, combining context, language, and sentiment features to enhance classification performance. We employ the Long Short-Term Memory (LSTM) architecture to capture linguistic and sentiment characteristics, augmented by the Hierarchical Contextual Attention Network (HCAN) to capture contextual information at both the word and sentence levels. Experimental results on a Reddit dataset demonstrate that CLSDepDet outperforms advanced methods, achieving an accuracy of 93 % and an F1 score of 95 %. The proposed model underscores the importance of integrating diverse features to improve classification accuracy and opens avenues for further research in developing efficient deep learning models for mental health applications. CLSDepDet not only provides a novel approach to detecting depressive posts on social media but also contributes to the development of early detection and diagnosis systems for depression, thereby improving the quality of life for affected individuals.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"291 ","pages":"Article 128505"},"PeriodicalIF":7.5,"publicationDate":"2025-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144270679","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Deciphering circRNA-drug sensitivity associations via global-local heterogeneous matrix factorization and hypergraph contrastive learning","authors":"Tiyao Liu ,&nbsp;Shudong Wang ,&nbsp;Yuanyuan Zhang ,&nbsp;Shanchen Pang ,&nbsp;Wenjing Yin ,&nbsp;Wenhao Wu ,&nbsp;Yingye Liu","doi":"10.1016/j.eswa.2025.128548","DOIUrl":"10.1016/j.eswa.2025.128548","url":null,"abstract":"<div><div>Growing evidence highlights the critical role of circular RNAs (circRNAs) as regulators of cellular drug sensitivity, significantly influencing drug efficacy. While matrix factorization has proven feasible in uncovering circRNA-drug sensitivity associations, existing methods rely solely on decomposing the association matrix and fail to efficiently incorporate richer biological information. Moreover, current predictive models are limited in representing multi-perspective relationships and higher-order relationships in circRNA-drug sensitivity associations. To address these limitations, we propose a novel model based on global-local heterogeneous matrix factorization and hypergraph contrastive learning (HMFHCL). HMFHCL first calculates the global and local similarities of circRNAs and drugs, then constructs global and local circRNA-drug heterogeneous networks and performs matrix factorization of the adjacency matrices of these networks to extract information-rich feature representations. By utilizing multi-source information, HMFHCL effectively captures richer structural features and reveals potential connections in heterogeneous networks. Next, we constructed multiple circRNA and drug hypergraphs using global and local association matrices to capture higher-order interactions between nodes via hypergraph convolution. To enhance feature learning, comparative learning is applied to both global and local views of circRNA/drugs, effectively mining the similarities and differences between global structures and local details, improving the model’s ability to perceive underlying patterns and the consistency of representation learning. Finally, HMFHCL integrates circRNA and drug features from different perspectives to predict circRNA-drug associations effectively. Comprehensive experiments on three benchmark datasets demonstrate that HMFHCL outperforms state-of-the-art models, highlighting its superior ability in uncovering complex circRNA-drug sensitivity associations.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"292 ","pages":"Article 128548"},"PeriodicalIF":7.5,"publicationDate":"2025-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144297251","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A novel method based on wavelet transform and prototypical network for gearbox detection in few-shot learning","authors":"Xianhua Chen ,&nbsp;Zhigang Tian ,&nbsp;Yuejian Chen","doi":"10.1016/j.eswa.2025.128601","DOIUrl":"10.1016/j.eswa.2025.128601","url":null,"abstract":"<div><div>Fault diagnosis is crucial for industrial systems, with traditional methods such as CNN heavily reliant on large training datasets to achieve high accuracy. However, such datasets are often-times inaccessible in the real world. Even in few-shot learning models, such as Model-Agnostic Meta-Learning (MAML), the quantity of training data significantly impacts the stability and accuracy of the models, posing challenges for reliable fault diagnosis under limited data conditions. To address these issues, the Wavelet Transform Prototypical Network (WTPN) is proposed, which integrates discrete wavelet transform with prototypical networks for limited training dataset. There are two main structures in WTPN. Firstly, this method transforms one-dimensional vibration signals into two-dimensional distance matrices, enhancing feature extraction and classification accuracy. Secondly, a confidence weighting mechanism assigns weights to decomposed signals based on their classification reliability, thereby improving consistency and reducing performance variability. Then, results from both experimental and publicly available datasets validate that WTPN consistently outperforms existing few-shot learning models in terms of accuracy and stability. Furthermore, the contributions include enhanced feature extraction through DWT, improved stability via confidence weighting, and robust performance in scenarios with limited training data. In conclusion, WTPN represents a significant advancement in fault diagnosis, offering reliable outcomes with minimal training data, making it particularly suitable for applications where data availability is constrained.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"292 ","pages":"Article 128601"},"PeriodicalIF":7.5,"publicationDate":"2025-06-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144297253","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Customization or jailbreaking for bloatware: strategic impacts of consumer-initiated behavior of software products","authors":"Zhitang Li ,&nbsp;Benjamin Lev","doi":"10.1016/j.eswa.2025.128434","DOIUrl":"10.1016/j.eswa.2025.128434","url":null,"abstract":"<div><div>Bloatware refers to software that contains unwanted plug-ins. Jailbreak software aims to eliminate these undesirable components and enhance the flexibility of bloatware, while custom software is specifically tailored to fulfill individual customer requirements. In contrast to previous research, our study addresses the challenge of proposing two strategies to mitigate bloatware inadaptability: jailbreak and customization. Some interesting conclusions were drawn. Firstly, when customization costs are low and the degree of customization is high, users tend to purchase bloatware at a lower price rather than opting for the expected high-priced customized software, indicating that consumers do not prefer the high premiums associated with extensive customization. Moreover, when users’ effort and the positive benefits of jailbreaking are low, they surprisingly choose to pay a higher price for jailbroken software. This phenomenon challenges the conventional view that jailbreaking is typically regarded as a low-cost alternative sought by users. Secondly, due to the larger user base in the jailbreak market compared to the bloatware market, there exists a market vacuum that can be capitalized on by jailbreak software vendors. It is worth noting that the profit margins in the jailbreak market tend to be higher than those in the bloatware market, indicating the disruptive influence of small firms resembling jailbreak vendors on the software industry. Despite the presence of established bloatware companies, jailbreak vendors still manage to capture a significant share of the overall profits. Thirdly, our research findings indicate that user jailbreak behavior leads to the highest consumer surplus. Finally, we conducted a comprehensive case study analyzing the impact of different user behavior strategies on software enterprises’ revenue in the context of bloatware, customized, and jailbroken apps specific to Apple’s platform.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"291 ","pages":"Article 128434"},"PeriodicalIF":7.5,"publicationDate":"2025-06-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144263559","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Knowledge-aware modeling of group commonality for group recommendation","authors":"Wen Wu ,&nbsp;Guangze Ye ,&nbsp;Hui Yu ,&nbsp;Wenxin Hu ,&nbsp;Xi Chen ,&nbsp;Liang He","doi":"10.1016/j.eswa.2025.128543","DOIUrl":"10.1016/j.eswa.2025.128543","url":null,"abstract":"<div><div>Group Recommendation (GR) aims to offer recommendations that satisfy the entire group. Due to the inherent sparsity of group-item interactions in GR, relying solely on individual-level preference aggregation is often insufficient for producing high-quality recommendations. In contrast, mining group-level commonality that reflects shared behavioral patterns can help mitigate this challenge. Some existing methods attempt to model group commonality based on the number of overlapping users across groups. However, this approach often fails in sparse settings where shared users between groups are absent, leaving the data sparsity issue unresolved. To tackle these issues, we propose a novel model based on Knowledge-Aware Modeling of Group <u><strong>Com</strong></u>monality for Group <u><strong>Rec</strong></u>ommendation (ComRec). ComRec eliminates the reliance on overlapping users by modeling fine-grained commonality from the item side. Specifically, we construct a Group Collaborative Knowledge Graph (G-CKG) by integrating group members’ interactions, membership relations, and item knowledge, enabling the capture of multi-hop relational paths for each member. We then extract fine-grained commonality by fusing multiple relational representations with an orthogonal constraint to ensure signal independence. A novel commonality attention mechanism further aggregates member entity representations to derive the overall group-level commonality representation. Beyond modeling group commonality, we further consider the specific group composition by introducing a user-based fine-tuning module that refines the group representation through member-level differences. The results show that our model significantly outperforms existing methods in terms of classification accuracy and interpretability on Yelp and MovieLens-20M datasets, while effectively addressing the data sparsity issue in GR.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"291 ","pages":"Article 128543"},"PeriodicalIF":7.5,"publicationDate":"2025-06-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144296934","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"H2-MARL: Multi-agent reinforcement learning for Pareto optimality in hospital capacity strain and human mobility during epidemic","authors":"Xueting Luo ,&nbsp;Hao Deng ,&nbsp;Jihong Yang ,&nbsp;Yao Shen ,&nbsp;Huanhuan Guo ,&nbsp;Zhiyuan Sun ,&nbsp;Mingqing Liu ,&nbsp;Jiming Wei ,&nbsp;Shengjie Zhao","doi":"10.1016/j.eswa.2025.128432","DOIUrl":"10.1016/j.eswa.2025.128432","url":null,"abstract":"<div><div>Effectively balancing the losses from mobility restrictions and hospital capacity strain has drawn significant attention in the aftermath of COVID-19. Reinforcement learning (RL)-based strategies for human mobility management have recently advanced in addressing the dynamic evolution of cities and epidemics; however, they still face challenges in achieving coordinated control at the township level and adapting to cities of varying scales. To address the above issues, we propose a multi-agent RL approach that achieves Pareto optimality in managing hospital capacity and human mobility (H2-MARL), applicable across cities of different population scales. We first develop a township-level infection model with online-updatable parameters to simulate disease transmission and construct a city-wide dynamic spatiotemporal epidemic simulator. On this basis, H2-MARL is designed to treat each division as an agent, with a trade-off dual-objective reward function formulated and an experience replay buffer enriched with expert knowledge built. To evaluate the effectiveness of the model, we construct a township-level human mobility dataset containing over one billion records from four representative cities of varying scales. Extensive experiments demonstrate that H2-MARL has the optimal dual-objective trade-off capability, which can simultaneously minimize hospital capacity strain and human mobility restriction loss. Meanwhile, the applicability of the proposed model to epidemic control in cities of varying scales is verified, which showcases its feasibility and versatility in practical applications.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"291 ","pages":"Article 128432"},"PeriodicalIF":7.5,"publicationDate":"2025-06-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144263615","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Exploiting open data for CO2 estimation via artificial intelligence and eXplainable AI","authors":"Stefano Bilotta ,&nbsp;Luciano Alessandro Ipsaro Palesi ,&nbsp;Paolo Nesi","doi":"10.1016/j.eswa.2025.128598","DOIUrl":"10.1016/j.eswa.2025.128598","url":null,"abstract":"<div><div>Climate change is a challenge of global relevance that requires in-depth understanding and immediate response. Urban CO₂ emissions are one of the main causes of climate change, and their estimation is crucial for planning actions aimed at creating more sustainable cities. Currently, models for estimating CO₂ emissions mainly focus on traffic patterns, energy consumption, or a limited set of socio-economic factors, often overlooking the increasing role of urban services in today’s cities. Moreover, these data are rarely accessible, which limits their usefulness for policy design. In the present paper, a general CO₂ emission estimation approach, based on a large range of often accessible open data as predictors, is presented. Such open data are related to human activity regarding services and (socio-economic) conditions arising in a given (urban) area. The proposed model focuses on fine-scale prediction to better understand the dynamics of emissions via machine learning approaches, while taking into account an innovative study based on open data sources including city services. The resulting best models have been based on XgBoost and GCN (graph convolutional network). The outcomes provided better precision (MAPE in the order of 8%) with respect to the state-of-the-art solutions. The goal has been to understand how specific predictors can contribute to or mitigate CO₂ emissions in the observed area. To this end, the impact of several features has been analyzed in order to identify the related key factors influencing emissions. A formal study has been conducted to perform feature relevance analysis by using eXplainable AI (XAI) approach. The proposed model is useful to define targeted policies reducing the pollutant impact of cities, promote a more ecologically sustainable urban lifestyle and improve sustainable urban planning. Both solutions and models have been assessed and improved, so as to be more flexible by using some transfer learning techniques. This research and its related results have been produced and validated by exploiting the Snap4City framework for smart city, mobility and transport and data analytics on CN MOST, national center on sustainable mobility. In particular, such results deal with the metropolitan cities of Florence and Bologna.</div><div>© 2017 Elsevier Inc. All rights reserved.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"291 ","pages":"Article 128598"},"PeriodicalIF":7.5,"publicationDate":"2025-06-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144270682","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Jujube-YOLO: a precise jujube fruit recognition model in unstructured environments","authors":"Lijun Wang ,&nbsp;Shuheng Wang ,&nbsp;Bo Wang ,&nbsp;Zhilei Yang ,&nbsp;Yanyu Zhang","doi":"10.1016/j.eswa.2025.128530","DOIUrl":"10.1016/j.eswa.2025.128530","url":null,"abstract":"<div><div>Accurate and efficient detection of jujube fruits in unstructured environments is considered a key challenge. The distinction between cracked and crack-free jujube is considered crucial for enabling selective harvesting by robots. The issues of low detection performance and difficulty distinguishing cracked from crack-free jujube fruits are addressed by the proposed Jujube-YOLO model, based on an improved YOLOv11. A double convolution squeeze-and-excitation (DCSE) module is integrated into the backbone network, and a rectangular self-calibration module (RCM) with the C3k2 module is introduced to enhance the expression of initial features and the extraction of multi-scale contextual information from fresh jujube fruits. A multi-branch channel attention (MBCA) module is designed to replace the standard convolution in the neck network, enabling effective fusion of shallow detail, deep semantic, and multi-scale information. The experimental results show that Jujube-YOLO achieves precision, recall, [email protected], and F1 of 98.37 %, 94.96 %, 97.65 %, and 96.63 %, respectively, with performance shown to be superior to that of Faster R-CNN, YOLOv3, YOLOv3-tiny, YOLOv5n, YOLOv6n, YOLOv8n, and YOLOv11n. At the same time, a practical analysis of lighting conditions, occlusions, speed, sample sources, and model size is performed, and it is concluded that Jujube-YOLO is capable of completing the recognition task in unstructured environment The Jujube-YOLO model is designed for recognizing fresh jujube fruits in orchards, offering theoretical insights for quality assessment, growth monitoring, and selective harvesting robots. The code will be released on GitHub. (<span><span>https://github.com/wangshuheng000210/Jujube-YOLO.git</span><svg><path></path></svg></span>).</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"291 ","pages":"Article 128530"},"PeriodicalIF":7.5,"publicationDate":"2025-06-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144280410","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Cardiac index adaptive physiological control system for continuous-flow left ventricular assist device","authors":"Hongtao Liu ,&nbsp;Aili Guan ,&nbsp;Fayu Xing ,&nbsp;Yunpeng Zhang ,&nbsp;Wencheng Wang ,&nbsp;Guoxu Liu","doi":"10.1016/j.eswa.2025.128536","DOIUrl":"10.1016/j.eswa.2025.128536","url":null,"abstract":"<div><div>A personalized continuous-flow left ventricular assist device (CFLVAD) therapy is designed to provide a precise treatment strategy based on the specific circumstances and needs of the individual patient. The aim is to improve treatment outcomes and the patient’s quality of life. Given this background, this study proposes a cardiac index adaptive physiological control (CIAPC) strategy to personalize the CFLVAD therapy. The CFLVAD physiological control system was constructed based on the cardiovascular coupling system model. A cardiac index regulator was designed by deriving a functional relationship between cardiac index and heart rate, height, and weight. The cardiac index adaptive controller consisted of the cardiac index regulator combined with a fuzzy controller and a ventricular suction prevention controller. The rotary pump flow and left ventricular pressure were measured indirectly by the model estimation method. The computer results show that the CIAPC system could maintain hemodynamic stability under extreme physiological conditions, effectively preventing ventricular suction and pump regurgitation. In a resting state, the system could automatically adjust and maintain personalized cardiac output (4 L/min, 5 L/min, and 7 L/min) based on patient body type differences, while ensuring that the cardiac index for all patients remains stable at around 3 L/min/m², which is within the normal physiological range. Furthermore, following an approximate 30 s period of transitional adjustment, the CIAPC system demonstrated the capacity to dynamically modulate cardiac output in response to patient needs. Specifically, it was able to increase the cardiac output of a patient with a normal body type from 5 L/min at rest to 8 L/min during exercise, or decrease it to 4 L/min during sleep. The CIAPC can control the rotary pump speed of the CFLVAD based on individual characteristics, enabling the delivery of personalized therapy to patients.</div></div>","PeriodicalId":50461,"journal":{"name":"Expert Systems with Applications","volume":"291 ","pages":"Article 128536"},"PeriodicalIF":7.5,"publicationDate":"2025-06-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144296783","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}