Current healthcare data systems face major challenges in preventing unauthorized access, ensuring compliance with data privacy regulations, and enabling intelligent secondary use of patient information. To address these issues, we introduce cluster-based analysis with machine learning for enhanced healthcare data security (CAML-EHDS), a unified framework that combines homomorphic encryption, attribute-based elliptic curve cryptography (ECC), and semantic clustering with machine learning. CAML-EHDS improves upon existing models by offering fine-grained access control, adaptive threat detection, and data-driven insights while preserving privacy. Experimental results show that CAML-EHDS achieves up to 98% classification accuracy with low node count, and maintains 94% accuracy even at high node distribution levels, while ensuring encryption time under 24 seconds and acceptable data loss below 29%. Moreover, in comparative analysis with state-of-the-art models (support vector machine (SVM), random forest (RF), and decision tree (DT)), CAML-EHDS outperforms all in key metrics with an accuracy of 0.96. These results demonstrate CAML-EHDS’s potential for real-world deployment in secure, scalable, and intelligent healthcare environments, including privacy-aware digital marketing integration.

Synthetic data, synthetic trust: navigating data challenges in the digital revolution.

Artificial intelligence methods and approaches to improve data quality in healthcare data

Balancing palliative care needs and medication appropriateness: Initiation and reinitiation of medications at the end of life.

The Future of the Future: Artificial Intelligence in Transforming Primary and Health Care.

This review explores state-of-the-art natural language processing (NLP) methods applied to electronic medical records (EMRs) for key tasks such as expanding medical abbreviations, automated diagnosis generation, international classification of diseases (ICD) classification, and explaining model outcomes. With the growing digitization of healthcare data, the complexity of EMR analysis presents a significant challenge for accurate and interpretable results. This paper evaluates various methodologies, highlighting their strengths, limitations, and potential for improving clinical decision-making. Special attention is given to abbreviation expansion as a crucial step for disambiguating terms in the clinical text, followed by an exploration of auto-diagnosis models and ICD code assignment techniques. Finally, interpretability methods like integrated gradients and attention-based approaches are reviewed to understand model predictions and their applicability in healthcare. This review aims to provide a comprehensive guide for researchers and practitioners interested in leveraging NLP for clinical text analysis.

An analytical framework for improving healthcare data management and organizational performance

SecureNet: A deep learning inspired security framework for healthcare data

Machine learning models to predict 6-month mortality risk in home-based hospice patients with advanced cancer.

Since the population is rising worldwide, a vast need arises to deliver appropriate medical care services. The sensor is an effective technology primarily employed to enable the Internet of Things (IoT)- based healthcare monitoring method. The IoT is transporting a novel revolution in research and academia. It has powerful roots, which are producing amazing variations in numerous areas, especially healthcare. IoT healthcare methods enable patients to receive personalized care by remotely monitoring their conditions. IoT applications are primarily beneficial for delivering healthcare, as they allow secure and real-time remote patient monitoring. In recent times, the traditional linear method has been replaced by innovative techniques of Artificial Intelligence (AI) and Machine Learning (ML). Whereas, Deep Learning (DL) is a sub-field of ML, which is much more trustworthy and stronger to certainly manage and study from a vast quantity of intricate healthcare data, and provides actionable visions and solutions to complex issues. This study proposes an Efficient Health Care Monitoring System using Advanced Metaheuristic Optimisation Algorithms and Spiking Neural Network Method for Smart Diagnosis (EHCMS-MOASNN) model for Smart Diagnosis in IoT. The primary objective of the EHCMS-MOASNN technique is to develop a smart healthcare monitoring system for the medical sector utilizing advanced models. Initially, the data pre-processing applies the min-max scaling method to convert input data into an appropriate format. Furthermore, the feature selection process is implemented using the Binary Grouper and Moray Eel (BGME) optimization approach to detect and select the most relevant and significant features in the input data. For the classification process, the EHCMS-MOASNN technique implements the Spiking Neural Network (SNN) approach. Additionally, the Mountain Gazelle Optimiser (MGO)-based hyperparameter tuning is performed. The comparison analysis of the EHCMS-MOASNN method demonstrated a superior accuracy value of 99.12% over recent techniques under the Healthcare IoT dataset.

A new method known as Lionized Remora optimization based Recurrent Neural Network (LRObRNN) is recommended to enhance the safety of medical information stored on cloud servers to tackle these issues. To safeguard patient data, healthcare organizations must thoughtfully choose reliable and compliant cloud service providers while implementing robust security measures. Storing patient information in cloud systems raises issues with illegal access and data breaches. The LRObRNN generates a secret key using Lionized Remora optimization and employs cryptography to encrypt sensitive healthcare data. Continuous monitoring ensures the security of data transmission by identifying irregularities. Leveraging Recurrent Neural Networks the system analyzes sequential data, enabling the detection of patterns and potential security breaches during data transmission. The performance evaluation includes metrics such as encryption and decryption time, confidentiality rate, processing time, resource usage and efficiency, which are compared with other existing models.

The healthcare institutions have started inculcating practices towards personalized care to its patients. Genomic data has now become an inevitable component of healthcare data allowing realization of personalize care of the patients. Medical practitioners are utilizing different technologies for efficient clinical interpretations of genomic data. The genomic data poses many challenges related to its management, amongst which data size and sensitivity are critical. The blockchain technology, a recent and still evolving technology, provides a single integrated solution for the various challenges encountered in managing healthcare data, both genomic and non-genomic. But still there are many challenges yet to be resolved in the domain of genomics and connected technological solutions. This paper presents a systematic literature review covering the research done on genomic data management using blockchain technology in healthcare. The review has incorporated 44 research and 43 review papers. We have designed our study based on 4 research questions targeted to cover efforts on genomic data management via blockchain technology. To our knowledge, majority of the research has focused on data sharing, privacy and security of genomic data. This systematic literature review (SLR) will contribute in identifying research gaps and directions for untouched areas.

Background The optimal sequencing of CDK4/6 inhibitors combined with endocrine therapy for advanced hormone receptor-positive, HER2-negative (HR+/HER2-) breast cancer remains uncertain, particularly in resource-limited settings such as China. This study evaluated the cost-effectiveness of first-line versus second-line CDK4/6 inhibitor use based on the SONIA trial. Methods A partitioned survival model was developed to compare costs and effectiveness of first-line (CDK4/6i-first) versus second-line (CDK4/6i-second) CDK4/6 inhibitor strategies among Chinese women with advanced HR+/HER2- breast cancer. Model inputs were derived from the SONIA trial and Chinese healthcare data. Outcomes included total costs, life years (LYs), quality-adjusted life years (QALYs), and incremental cost-effectiveness ratios (ICERs). Both deterministic and probabilistic sensitivity analyses were performed. Scenario analyses incorporated generic drug pricing. Results The base-case analysis showed that the CDK4/6i-first strategy yielded 3.07 QALYs at a lifetime cost of CNY 372420.21, compared to 2.86 QALYs and CNY 366445.93 for the CDK4/6i-second strategy. The ICER for first-line CDK4/6 inhibitor use was CNY 28126.33 per QALY, well below the willingness-to-pay (WTP) threshold of CNY 287,247/QALY. Scenario analysis with generics showed an ICER of CNY 198439.62 per QALY. Sensitivity analyses confirmed the robustness of these results. Conclusion This study supports the early use of CDK4/6 inhibitors combined with endocrine therapy as a cost-effective strategy for advanced HR+/HER2- breast cancer in China. Continued real-world monitoring is needed to adapt to changes in drug pricing and clinical practice.

The public perception of artificial intelligence (AI) in healthcare is key to its large-scale acceptance and implementation. This study investigated how exposure to ChatGPT changed public perception of AI in healthcare, using baseline and follow-up data from 5899 survey participants reporting their perception of AI in 2022 (before ChatGPT's launch) and 2024, and ChatGPT use in 2024. Multinomial multivariate logistic regression was used to model how exposure to ChatGPT use affected changes in perception of AI. At follow-up, 1195 individuals (20%) had been exposed to ChatGPT use, which was associated with higher odds of changing perception of AI to beneficial (OR 3.21 [95% CI: 2.34-4.40]) among individuals who were unsure at baseline, and lower odds of changing to uncertainty from more defined baseline perceptions. This study demonstrates the potential for reducing uncertainty and improving public perception of AI in healthcare through exposure to AI tools.

This paper addresses the critical challenge of fraud detection in medical insurance claims-a pervasive issue causing significant financial losses in healthcare-using Graph Neural Networks (GNNs). Given the intricate nature of healthcare data, traditional fraud detection methods do not inherently capture the complex relationships and patterns among different entities. We explore the potential of GNNs to effectively identify fraudulent claims by modeling the interactions among various entities-such as patients, healthcare providers, diagnoses, and services-as a heterogeneous graph. We employ two state-of-the-art heterogeneous GNN architectures, HINormer (Heterogeneous Information Network Transformer) and HybridGNN, along with a modified homogeneous GNN, RE-GraphSAGE (GraphSAGE "Graph Sample and Aggregate" with relation embeddings), adapted to handle the heterogeneity of healthcare data. The models are evaluated on real-world claims datasets of different sizes, comprising millions of medical activities. For the small-size claims dataset, the HINormer architecture followed by the RE-GraphSAGE architecture achieved the highest F-score (84% and 83%, respectively). For the medium-sized claims dataset, RE-GraphSAGE followed by HINormer achieved the highest F-score (84% and 81%, respectively), and for the large-size claims dataset, HINormer followed by RE-GraphSAGE achieved the highest F-score (82% and 79%, respectively). Additionally, we apply explainability techniques, namely GNNExplainer and PGExplainer, to provide insights into the models' decision-making processes and to examine their medical significance.

Abstract Internet of Medical Things (IoMT) is built with various medical equipment to improve healthcare technology, including smart devices, hardware infrastructure, and software applications. The network experiences massive data traffic due to the data generated by these medical devices. Controlling this data flow while meeting user expectations becomes difficult. Information Centric Network (ICN) networks are employed to overcome data management problems and effectively handle data transfer in a network. This work aims to develop a patient-centric approach for IoMT to optimize healthcare data access using ICN in-network caching by prioritizing the content and categorizing the content based on the patient’s disease ranking. A two-queue technique with dynamic caching is presented in this paper for effective caching at edge devices. Every content item is divided into four categories, and each edge router maintains two queues to store the content based on priority. Age and disease ranking fields are added to interest and data packets to identify the type of content. Based on the frequency of access, the content in the edge routers is dynamically updated. Least Recently Used (LRU) based prioritized queue cache replacement algorithm is proposed to replace content in each queue by prioritizing emergency content. The proposed work is evaluated in terms of cache hit ratio, content latency, and the stretch ratio in a Java-based Java Information Centric Cache Network Simulator (JICCNS) simulator. The performance of the proposed work shows better results than existing strategies and ensures optimal cache utilization and data retrieval efficiency.

In the contemporary era of healthcare informatics, preserving the privacy of EHD while ensuring its accessibility for legitimate purposes poses a significant challenge. Conventional methods of protecting sensitive medical information have frequently proven inadequate, exposing vulnerabilities that jeopardize patient privacy. Blockchain technology has emerged as a promising solution due to its inherent characteristics of decentralization, immutability, and transparency. The objective of this paper is to propose a novel Blockchain-Assisted Optimized Privacy Preservation Model (BAOPPM) for Electronic Health Records (EHRs) that enhances the privacy of sensitive healthcare data while ensuring its accessibility for legitimate purposes. The model leverages blockchain technology for secure data storage and integrates advanced algorithms for efficient privacy key generation and data sanitization. At first, the HCK Algorithm is used to generate sensitive data clusters and non-sensitive data clusters. Then, the sensitive data are sanitized by adopting the Kronecker product between the sensitive data and the optimal key. Here, the optimal key is generated via the BMACO algorithm that considers the level of privacy by including privacy and privacy information ratio. Finally, the sanitized data are stored in the blockchain. On the contrary, the data restoration process is done, which is the reverse operation of data sanitization to retrieve the sensitive data from the blockchain. The BMACO approach achieves the lowest cost rate of 0.2451 at the 50th iteration, showcasing its superior convergence efficiency. In contrast, conventional methods such as COOT, BMO, KOA, COATI, LBO, EHO-OBL, and ECSO register higher cost ratings of 0.2463, 0.2498, 0.2512, 0.2484, 0.2473, 0.2536, and 0.2469, respectively. The sanitized data were stored in the blockchain and restored securely using the developed model.

Type 2 diabetes mellitus (T2DM) follows a social gradient; however, access to health-related services and amenities in urban areas may also impact its prevalence. This study investigates the association between characteristics of the built environment and T2DM prevalence in Milan, Italy. Utilising administrative health care data and open access territorial information, we examined the relationship between neighbourhood-level features and T2DM among the Milanese population in 2019. The analysis focused on six domains: food environment, physical activity infrastructure, cultural facilities, neighbourhood disorder, health care accessibility, and public transportation. We employed spatial analysis techniques based on road network buffers to evaluate potential access to health-related resources. Multilevel Poisson regression models, stratified by age group and sex, were applied to explore associations while adjusting for average neighbourhood educational attainment. Several built environment indicators demonstrated significant associations with T2DM; however, these associations lost statistical significance after controlling for area-level education. The influence of the built environment on health outcomes may differ in Milan's unique context compared to those settings where previous research has yielded consistent findings, particularly in the American context, which involves larger metropolises with greater urban heterogeneity and low- and middle-income countries characterised by notable urban/rural differences.

Glucagon-like peptide-1 (GLP-1) agonists are increasingly prescribed for obesity and type 2 diabetes. GLP-1 agonists influence body composition through effects on both fat mass and fat-free mass. Given that critically ill patients experience severe protein catabolism and commonly develop intensive care unit (ICU)-acquired weakness, questions arise about outcomes when metabolic demands are high during critical illness. The objective of this study is to examine the relationship between prior GLP-1 agonist use and critical care outcomes. We conducted a retrospective cohort study using Stanford Health Care data from January 2015 to July 2024. Adults aged 18-89 years admitted to intensive care with body mass index (BMI) 20-60 kg/m2 were included. Of 15,191 eligible ICU patients, 468 (3.1%) received GLP-1 agonist prescriptions within 12 months before hospitalization. Using high-dimensional propensity score matching with lasso regression, we created 452 matched pairs and compared in-hospital mortality, hospital length of stay, and ICU length of stay between groups. Baseline characteristics were well-balanced. The matched GLP-1 agonist and comparison groups showed similar in-hospital mortality (5.1% vs. 4.9%, odds ratio [OR]: 1.05, 95% confidence interval [CI]: 0.58 to 1.91, p = .88), mean hospital length of stay (13.7 ± 21.3 vs. 13.4 ± 18.1 days, mean difference [MD]: 0.38, 95% CI: -2.21 to 3.05, p = .77), and ICU length of stay (5.9 ± 9.0 vs. 5.4 ± 6.6 days, MD: 0.51, 95% CI: -0.52 to 1.50, p = .33). In this first study examining the relationship between prior GLP-1 agonist use and critical care outcomes, we found no significant associations with in-hospital mortality, hospital length of stay, or ICU length of stay.

Researchers typically identify pregnancies in healthcare data based on observed outcomes. This approach misses pregnancies that received prenatal care but whose outcomes were not recorded, potentially inducing selection bias in prenatal effect estimates. Alternatively, prenatal encounters can be used to identify pregnancies with unobserved outcomes, but this requires addressing loss to follow-up (LTFU). We simulated 10,000,000 pregnancies and estimated the total effect of treatment on preeclampsia. Across 36 scenarios, we varied the treatment effect on miscarriage and/or preeclampsia; percent LTFU (5% or 20%); and cause of LTFU: (1) measured covariates, (2) unobserved miscarriage, and (3) both. We created analytic samples to address LTFU-observed deliveries, observed deliveries and miscarriages, and all pregnancies-and estimated treatment effects using non-parametric direct standardization. Risk differences (RDs) and risk ratios (RRs) from the samples were similarly biased when LTFU was due to miscarriage (log-transformed RR bias: -0.12-0.33 among observed deliveries; -0.11-0.32 among observed deliveries and miscarriages; and -0.11-0.32 among all pregnancies). When predictors of LTFU were measured, only estimates among all pregnancies were unbiased (-0.27-0.33; -0.29-0.03; and -0.02-0.01, respectively). While including all pregnancies does not prevent bias, it quantifies the extent of selection, enabling direct assessment of its potential impact on findings.