Smart Health最新文献

{"title":"Classification of body mass index levels using breast thermography: A preliminary proof-of-concept analysis with convolutional neural networks","authors":"Rodrigo M. Carrillo-Larco","doi":"10.1016/j.smhl.2025.100597","DOIUrl":"10.1016/j.smhl.2025.100597","url":null,"abstract":"<div><h3>Background</h3><div>Thermal imaging has shown promise in distinguishing between obese and non-obese individuals, yet its potential in stratifying thinner body mass index (BMI) categories remains largely unexplored.</div></div><div><h3>Methods</h3><div>We utilized thermal images depicting the upper abdomen to the neck of women (aged 18–81) with benign breast pathology, each comprising anterior, oblique left, and oblique right views. Employing transfer learning and convolutional neural networks (CNN), we classified images into normal weight, overweight, and obesity categories. GradCAM activation maps identified influential areas in the images.</div></div><div><h3>Results</h3><div>84 women were included in the analysis, with a mean age of 45 years (standard deviation (SD): 12.2). The average BMI was 28.6 kg/m² (SD: 6.5), and BMI categories were evenly distributed. The overall accuracy was 87.0 %. The model performed best for the obesity category (precision: 100 %, recall: 93 %, and F1 score: 97 %). The lowest precision was observed for the normal weight category (75 %), while the overweight category had the lowest recall (67 %) and F1 score (76 %). The confusion matrix showed that misclassifications were predominantly between the normal weight and overweight categories. Activation maps showed that in the normal weight group, the sternum area was most influential; for the obesity group, regions above the armpits were emphasized; and for the overweight group, key areas included the upper abdomen, below the breasts, and upper chest.</div></div><div><h3>Conclusions</h3><div>Thermal imaging effectively differentiated between BMI categories. Further validation of thermal imaging's predictive capabilities is warranted considering the limited and women-only sample herein analyzed.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"37 ","pages":"Article 100597"},"PeriodicalIF":0.0,"publicationDate":"2025-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144523367","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"ML-predicted surgical site infections: An epidemiological study utilizing machine learning on routinely collected healthcare data to predict infection risk","authors":"Davide Golinelli ,&nbsp;Simona Rosa ,&nbsp;Paola Rucci ,&nbsp;Francesco Sanmarchi ,&nbsp;Dario Tedesco ,&nbsp;Carlo Biagetti ,&nbsp;Alessio Gili ,&nbsp;Andrea Bucci ,&nbsp;Luca Romeo ,&nbsp;Roberto Grilli","doi":"10.1016/j.smhl.2025.100596","DOIUrl":"10.1016/j.smhl.2025.100596","url":null,"abstract":"<div><h3>Background</h3><div>Surgical site infections (SSIs) are a major public health issue, causing increased morbidity, longer hospital stays, and higher healthcare costs. Despite progress in infection control, predicting and preventing SSIs remain crucial for improving patient outcomes. This study examines the use of machine learning (ML) on routinely collected healthcare data (RCD) to predict SSIs in orthopaedic surgery, aiming to improve risk stratification and guide interventions.</div></div><div><h3>Objectives</h3><div>To develop, test, and validate an ML predictive model using RCD to assess SSI risk in orthopaedic surgery patients.</div></div><div><h3>Methods</h3><div>A retrospective study was carried out using RCD from a 1.2 million population in an Italian Local Health Authority, covering surgeries from 2017 to 2021. The population included patients undergoing hip or knee arthroplasty and open reduction of fractures. Several ML algorithms, including eXtreme Gradient Boosting (XGBoost), were used for model development. The models’ performance was assessed by recall, accuracy, and area under the receiver operating characteristic curve (AUC). A feature importance analysis identified key SSI risk predictors.</div></div><div><h3>Results</h3><div>The XGBoost model demonstrated superior performance, with a recall exceeding 70% and an AUC&gt;0.70, overcoming other methods. Significant predictors included the ASA classification, opioid use, priority class of the surgery operation, and length of hospital stay.</div></div><div><h3>Conclusions</h3><div>ML models, particularly XGBoost, effectively predicted SSI risk in orthopaedic patients, offering a new approach to infection control and prevention. Incorporating ML and RCD highlights the potential for scalable, data-driven personalized medicine interventions. Future research will focus on model validation and integration of these tools into healthcare systems for enhanced patient management.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"37 ","pages":"Article 100596"},"PeriodicalIF":0.0,"publicationDate":"2025-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144518492","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A non-immersive virtual reality serious game to assess distractor inhibition and spatial attention in post-stroke individuals","authors":"Gregorio Sorrentino ,&nbsp;Gauthier Everard ,&nbsp;Florence Vanhoof ,&nbsp;Thierry Lejeune ,&nbsp;Martin Gareth Edwards","doi":"10.1016/j.smhl.2025.100595","DOIUrl":"10.1016/j.smhl.2025.100595","url":null,"abstract":"<div><h3>Background</h3><div>REAsmash, a non-immersive virtual reality serious game based on Feature Integration Theory (FIT) was used to evaluate distractor inhibition and spatial attention.</div></div><div><h3>Methods</h3><div>Participants (15 post-stroke and 15 age matched healthy controls) performed the visual search REAsmash task with manipulations of high and low target-distractor salience and distractor number. Dependent variables included frequency of errors and omissions and mean correct response time.</div></div><div><h3>Results</h3><div>Post-stroke participants made more errors (χ²(1) = 19.452, p &lt; 0.001) and omissions (χ²(1) = 55.108, p &lt; 0.001), and responded slower (F(1,28) = 17.957, p &lt; 0.001, η² = 0.391) than controls. FIT significant effects showed salience (F(1,28) = 497.626, p &lt; 0.001, η² = 0.947), distractor number (F(2,56) = 24.968, p &lt; 0.001, η² = 0.471), and a salience and distractor number interaction (F(2,56) = 26.616, p &lt; 0.001, η² = 0.487), with increased distractor number slowing response time in the low salience condition. A salience-by-group interaction (F(1,28) = 7.794, p = 0.009, η² = 0.218) indicated greater post-stroke difficulties in low salience conditions. A target-hand congruency (Simon) effect was observed only in post-stroke participants (F(1,28) = 4.499, p = 0.043, η² = 0.138).</div></div><div><h3>Conclusion</h3><div>REAsmash-niVR replicated FIT results, confirming its validity for assessing distractor inhibition and spatial attention in post-stroke populations. Future research will compare individuals with and without visual neglect and explore rehabilitation applications to improve spatial attention and motor responses.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"37 ","pages":"Article 100595"},"PeriodicalIF":0.0,"publicationDate":"2025-06-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144469934","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Privacy-enhanced heart stroke detection using Federated Learning and Homomorphic Encryption","authors":"Vankamamidi S. Naresh ,&nbsp;Gadhiraju Tej Varma","doi":"10.1016/j.smhl.2025.100594","DOIUrl":"10.1016/j.smhl.2025.100594","url":null,"abstract":"<div><div>Heart stroke detection plays a vital role in early diagnosis and intervention leveraging machine learning (ML) models to predict stroke events from medical data. However, the use of such models often faces significant challenges related to data privacy and security, especially when sensitive health information is involved. To tackle these concerns, we present an innovative privacy-preserving approach for heart stroke detection using Federated Learning (FL) combined with Homomorphic Encryption (HE) utilizing the Cheon-Kim-Kim-Song (CKKS) scheme. FL is employed to enable edge nodes to locally train a Feed Forward Neural Network (FFNN) model on heart stroke data, without transferring any raw patient data to a central server, thus ensuring the data remains decentralized and private. During the communication between edge nodes and central server, CKKS encryption ensures that model updates remain encrypted throughout the aggregation process, allowing computations to be performed without decryption, thus enhancing privacy and security. Furthermore, the FedAvg is incorporated to aggregate model updates efficiently, ensuring robust model training while maintaining decentralized data integrity. Our model achieved an average accuracy of over 95 % in predicting heart stroke events, demonstrating the effectiveness of combining FL with an FFNN in terms of both performance and privacy. This approach allows the utilization of rich medical data for training while maintaining the security and confidentiality of sensitive health information, making it an effective option for real-world healthcare applications in contexts where privacy and security of data is of the utmost importance.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"37 ","pages":"Article 100594"},"PeriodicalIF":0.0,"publicationDate":"2025-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144184264","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An interpretable deep learning framework for predicting hospital readmissions from electronic health records","authors":"Fabio Azzalini ,&nbsp;Tommaso Dolci ,&nbsp;Marco Vagaggini","doi":"10.1016/j.smhl.2025.100581","DOIUrl":"10.1016/j.smhl.2025.100581","url":null,"abstract":"<div><div>With the increasing availability of patient data, modern medicine is shifting towards prospective healthcare. Electronic health records offer a variety of information useful for clinical patient characterization and the development of predictive models, given that similar medical histories often lead to analogous health progressions. One application is the prediction of unplanned hospital readmissions, an essential task for reducing healthcare costs and improving patient outcomes. While predictive models demonstrate strong performances especially with deep learning approaches, they are often criticized for their lack of interpretability, a critical requirement in the medical domain where incorrect predictions may have severe consequences for patient safety. In this paper, we propose a novel and interpretable deep learning framework for predicting unplanned hospital readmissions, supported by NLP findings on word embeddings and by ConvLSTM neural networks for better handling temporal data. We validate the framework on two predictive tasks for hospital readmission within 30 and 180 days, using real-world data. Additionally, we introduce and evaluate a model-dependent technique designed to enhance result interpretability for medical professionals. Our solution outperforms traditional machine learning models in prediction accuracy while simultaneously providing more interpretable results.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"37 ","pages":"Article 100581"},"PeriodicalIF":0.0,"publicationDate":"2025-05-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144147537","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Technoethics and recommendations for technological interventions to reduce loneliness in older adults","authors":"Emma Cho,&nbsp;Connie M. Ulrich","doi":"10.1016/j.smhl.2025.100583","DOIUrl":"10.1016/j.smhl.2025.100583","url":null,"abstract":"","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"37 ","pages":"Article 100583"},"PeriodicalIF":0.0,"publicationDate":"2025-05-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144116780","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Smart implantable devices for cardiac health: A novel self-powered wireless ECG monitoring system using energy harvesting and machine learning-driven anomaly detection","authors":"Khadija Pervez ,&nbsp;Muhammad Izhar ,&nbsp;Adeel Ahmed ,&nbsp;Nazik Alturki ,&nbsp;Saima Abdullah","doi":"10.1016/j.smhl.2025.100582","DOIUrl":"10.1016/j.smhl.2025.100582","url":null,"abstract":"<div><div>Introduction of smart implantable devices is changing the face of cardiac health monitoring through continuous and real time ECG monitoring useful in early diagnosis of cardiac pathologies. This paper describes CardioHarvest-Net, a newly developed self-powered wireless ECG monitoring system that employs, physiological movements for its power in order to reduce the probability of frequent power replenishment. This self-powered capability eliminates dependency on conventional batteries, thereby offering a viable solution for continuous, long-term cardiac monitoring in real-world conditions. CardioHarvest-Net enables an enhanced machine learning (ML)-based anomaly detection model that learns and adapt to each patient's cardiac behavior to provide high sensitivity in abnormal ECG signs related to diseases like arrhythmia, myocardial infarction, and other diseases of the heart. The CardioHarvest-Net model applies CNN for feature extraction of vital signs such as ECG and uses LSTM for temporal feature extraction for accurate anomaly detection in real-world settings. Evaluation results reveal that gain scores of cardio health phenomena via CardioHarvest-Net is a detection accuracy of 97.2 % and the anomaly recall rate of 95.3 % that qualifies the proposed system as an effective and timely monitoring tool of putting up a signal and cautionary measure on possible event of cardiac occurrences. The average response time for an entire system to detect an anomaly is 10 ms, which makes the system's intervention capacity rather fast. Moreover, they use a power build-up efficiency of 78 % in otherwise low power, real-life in-vivo conditions ranging from acute circumstances to chronic conditions requiring prolonged operation. This ML model is running on an energy-efficient microcontroller suitable for wearable and implantable medical devices along with a feedback adaptation that enhances the accuracy of the predictions based on data that changes over time concerning an individual patient. The outcomes of this study further state the viability of CardioHarvest-Net to transform sustainable cardiac niche by addressing limitations into power independence and facilitating real-time tracking. This development is a breakthrough in moving towards the preventive, long-term approach to cardiac reliability enhancing our method in a manner that offers a solid framework for constant, individualized cardiac monitoring, and timely action in cases of essential occurrences in the heart.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"37 ","pages":"Article 100582"},"PeriodicalIF":0.0,"publicationDate":"2025-05-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144068558","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Clinically relevant predictive modeling for personalized ACL reconstruction classification","authors":"Xishi Zhu ,&nbsp;Ryan Henry ,&nbsp;Emily Jackson ,&nbsp;Joe M. Hart ,&nbsp;Jiaqi Gong","doi":"10.1016/j.smhl.2025.100575","DOIUrl":"10.1016/j.smhl.2025.100575","url":null,"abstract":"<div><div>Anterior Cruciate Ligament (ACL) reconstruction outcomes and return-to-sport readiness vary significantly among patients, yet current classification methods often lack interpretability and personalization. We propose an explainable predictive model for ACL reconstruction classification through multi-modal analysis of gait dynamics and patient characteristics. Using inertial measurement unit (IMU) sensors on participants’ wrists, ankles, and sacrum, we collected gait data during walking and jogging tasks, alongside patient-specific survey information. For gait dynamics, we employed Phase Slope Index to quantify inter-sensor relationships and trained classifiers for different ACL reconstruction outcomes(left vs right injury, healthy vs injured), achieving high classification performance (96.37% accuracy). Model explanations using heatmaps and permutation importance revealed that paired body movements are crucial in classification, with more distinct patterns in jogging than walking. For patient characteristics, t-SNE visualization demonstrated that model confidence correlated strongly with recovery duration. While longer recovery typically leads to more normal gait patterns, our approach provides a quantitative method to visualize this process transparently. This explainable, personalized approach can improve rehabilitation strategies and inform more accurate return-to-sport decisions in sports medicine.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"36 ","pages":"Article 100575"},"PeriodicalIF":0.0,"publicationDate":"2025-04-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143895599","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Editorial – Elsevier smart health special issue: Advancing ICT for health, accessibility, and wellbeing","authors":"Achilleas Achilleos ,&nbsp;Edwige Pissaloux ,&nbsp;George A. Papadopoulos ,&nbsp;Ramiro Velazquez","doi":"10.1016/j.smhl.2025.100580","DOIUrl":"10.1016/j.smhl.2025.100580","url":null,"abstract":"","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"36 ","pages":"Article 100580"},"PeriodicalIF":0.0,"publicationDate":"2025-04-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143947103","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Enhancing polyp detection in endoscopy with cross-channel self-attention fusion","authors":"Xiaolong Liang ,&nbsp;Shuijiao Chen ,&nbsp;Linfeng Shu ,&nbsp;Dechun Wang ,&nbsp;Qilei Chen ,&nbsp;Yu Cao ,&nbsp;Benyuan Liu ,&nbsp;Honggang Zhang ,&nbsp;Xiaowei Liu","doi":"10.1016/j.smhl.2025.100578","DOIUrl":"10.1016/j.smhl.2025.100578","url":null,"abstract":"<div><div>Colorectal cancer (CRC) poses a significant global health challenge, ranking as a leading cause of cancer-related mortality. Colonoscopy, the most effective means of preventing CRC, is utilized for early detection and removal of precancerous growths. However, while there have been many efforts that utilize deep learning based approaches for automatic polyp detection, false positive rates in polyp detection during colonoscopy remain high due to the diverse characteristics of polyps and the presence of various artifacts. This paper introduces an innovative technique aimed at improving polyp detection accuracy in colonoscopy video frames. The proposed method introduces a novel framework incorporating a cross-channel self-attention fusion unit, aimed at enhancing polyp detection accuracy in endoscopic procedures. The integration of this unit proves to play an important role in refining prediction quality, resulting in more precise detection outcomes in complex medical imaging scenarios. To substantiate the effectiveness of our framework, we create an extensive private dataset comprising complete endoscopy videos, captured from diverse equipment from different manufacturers. This dataset represents realistic and intricate application scenarios, offering an authentic and effective foundation for both training and evaluating our framework. Thorough experiments and ablation studies are conducted to assess the performance of our proposed approach. The results demonstrate that our framework, featuring key technical innovations, significantly reduces false detections and achieves a higher recall rate. This underscores the remarkable effectiveness of our framework in upgrading polyp detection accuracy in real-world endoscopy procedures.</div></div>","PeriodicalId":37151,"journal":{"name":"Smart Health","volume":"36 ","pages":"Article 100578"},"PeriodicalIF":0.0,"publicationDate":"2025-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143850205","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}