Liver cancer has a high incidence and mortality rate globally, particularly in patients requiring intensive care unit (ICU) admission. Early prediction of in-hospital mortality for these patients is crucial, yet lacking reliable tools. This study aims to develop and evaluate machine learning (ML) models for predicting in-hospital mortality in critically ill liver cancer patients admitted to the ICU. This retrospective study used data from the MIMIC-III and MIMIC-IV databases, including 862 patients from MIMIC-III (training cohort) and 692 patients from MIMIC-IV (validation cohort). The study focused on patients diagnosed with liver cancer, identified by specific ICD codes. Four ML algorithms, namely logistic regression, random forest, XGBoost, and LightGBM, were used to predict in-hospital mortality based on clinical characteristics, laboratory results, and severity scores. Performance was evaluated using accuracy, AUROC, AUPRC, F1 score, and Kaplan-Meier curves. A total of 1,554 patients were included. The random forest model demonstrated the best performance, with an AUROC of 0.911 (95% CI: 0.855–0.956) and an AUPRC of 0.823 (95% CI: 0.718–0.905) in the internal test set, and an AUROC of 0.857 (95% CI: 0.826–0.889) and an accuracy of 0.828 (95% CI: 0.802–0.857) in the external validation set. Kaplan-Meier curves showed that all four models effectively stratified high-risk and low-risk groups. Key features influencing the prediction included APSIII, SAPSII, LODS, OASIS, and vital signs such as heart rate, temperature, and oxygen saturation. Feature importance analysis revealed that clinical severity scores played a major role in predicting mortality. This study demonstrates the potential of machine learning algorithms, particularly random forest, in predicting in-hospital mortality for critically ill liver cancer patients in the ICU. The identified clinical features provide valuable insights for clinicians in assessing patient risk and making timely interventions.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-24369-x.

Influence of HIV status, age and ethnicity on survival in patients with burkitt lymphoma (BL) and diffuse large B cell lymphoma (DLBCL): A real-world nationwide database study

Real-world step-up dosing of patients receiving elranatamab for multiple myeloma in Japan

Repeat TURBT for Ta and T1 bladder cancer: An updated review.

The growing prevalence of environmental issues globally has intensified the demand for greater transparency and accountability in corporate Environmental Disclosure (ED). In Nigeria, however, ED practices remain largely discretional, with no formal regulatory framework mandating such disclosures. The study thus investigates the combined effect of the latent constructs of board and audit committee monitoring mechanisms on the extent of ED among firms listed in Nigeria. Adopting a survey research design, the study analysed audited annual reports of 95 listed firms on the Nigerian Exchange Group (NGX) from 2012 to 2021, The Global Reporting Index (GRI) was employed to assess the level of ED across firms providing a standardised basis for measurement. Structural Equation Modelling (SEM) technique was utilised to estimate the relationships between the identified monitoring mechanisms and ED practices. The findings reveal that both board and audit committee monitoring mechanisms have a significant positive impact on ED. Based on the findings, the study recommends that Nigerian firms should pay more attention to fostering diverse boards and including experienced members. In addition, ED and regulators should consider enacting policies that will mandate minimum qualifications or certifications for audit committee members. Additionally, industry specific governance codes could be considered to ensure that audit practices align with firms’ specific attributes.

Abstract Background and Aims The analysis of US medication patterns by ATC code in dialysis treatment using a NDC (National Drug Code) to ATC/DDD Index (Anatomical Therapeutic Chemical Classification System/Defined Daily Dose) mapping has been rarely conducted so far. This project aims to examine the representation of selected drug classes based on ATC codes. We applied a publicly available NDC to ATC mapping algorithm and verified the mapping from a pharmaceutical perspective. The focus is placed on home medication in the US patient population from the ApolloDialDb™ database. Method The presented multi-center, retrospective analysis uses real life data from adult dialysis patients in the US captured in the first version of the ApolloDialDb dataset. It contains anonymized dialysis data (longitudinal observation level data) on more than 540,000 unique dialysis patients in 40 countries from Jan 2018–Mar 2021. The NDC to ATC mapping was applied to US home medication data including number of patients and frequencies per NDC as well as official product information from the FDA. This analysis was focusing on the frequencies of NDC prescriptions regardless of the duration of the prescription. Six ATC codes on a 4–5-digit level (A10A, A10BJ, A10BK, B01AA, C10AA, H05BX) were analyzed. As the NDC to ATC mapping algorithm provides a one-to-multiple matching, only the first three assigned ATC codes per NDC were included in the manual pharmaceutical plausibility checks. Initially, it was checked how many prescriptions occurred for NDCs that have been mapped to the specific ATC code (e.g., A10BJ). As some NDCs have not been assigned to ATC codes, a manual search was performed for substances belonging to the ATC code (e.g., A10BJ: semaglutide). This ensured that the number of prescriptions for NDCs were recorded even though the mapping has not been successful. The numbers for the two procedures have been combined. All identified drug names have been double-checked for pharmaceutical compatibility with the mapped ATC code. Particular attention was paid to active substance combinations and the area of application specified by the ATC classification. Drug combinations were excluded if a separate ATC code exists in the ATC/DDD Index (e.g., A10AE: insulin degludec/liraglutide). Results 8,931,154 home medication prescriptions in US patients were analyzed. The following results are obtained from the NDC to ATC mapping including manual verification and completion for the ATC codes (outlined in Figure 1). The total numbers and percentages of prescriptions for each ATC code are presented. Thus, the graph illustrates that there are e.g., 320,678 prescriptions for drugs classified in A10A, accounting for 3.59 % of all prescriptions. The totality of prescriptions results from the sum of all NDCs recorded in the dataset. Percentages are calculated by n/8,931,154 × 100. Conclusion The mapping provides an initial insight into the use of home medication in the US based on ATC codes. The ATC/DDD Index provides a meaningful combination of drug classes with regard to the area of application. Figure 1 demonstrates the possibility of applying the concept to data based on NDC level. It has been demonstrated that important medications such as insulins, HMG CoA reductase inhibitors and other anti-parathyroid agents are represented with 816,316 prescriptions, i.e. 9.14 % of the total home medication. Using ATC codes, medication in specific therapeutic areas (e.g., cardiac medication) could be highlighted to facilitate a comparison of the use of different active substance classes within one therapeutic area. Analyzing medication patterns while using NDC to ATC mapping will be advantageous for future cross-national analyses based on the clear definition of the ATC/DDD Index to ensure consistency and comparability. In this way, it can be ensured that the same active substances are included for the desired therapeutic area in the appropriate dosage form.

As technology continues to advance, it becomes increasingly integrated into daily life facilitating complex tasks across a range of environments. While some applications such as smartphones and smartwatches are less critical, others like healthcare devices and autonomous vehicles demand bug-free performance to prevent financial loss or harm. Traditionally, simulation-based testing and formal verification played a major role in ensuring a bug-free device. However, the simulation of bigger systems is limited to a definite number of scenarios on the Design under Verification (DUV). Hence, it is unable to explore all possible inputs that can occur. Formal verification, on the other hand, offers a higher level of assurance through mathematical proofs but is both time-consuming and suffers from scalability issues, especially as designs grow in complexity. Recently, Large Language Models (LLMs) have shown promise in tasks previously limited to human expertise. Their natural language processing capabilities can assist in handling extensive specifications and source code, particularly in debugging hardware descriptions and analyzing security and functionality. The utilization of Retrieval Augmented Generation (RAG) has further enhanced LLMs by incorporating large specification or source code bases, thereby improving their bug-identification and correction capabilities. While recent advancements in LLMs, particularly with RAG, have yielded promising results in bug identification and correction for a small class of hardware bugs, significant gaps remain in their full potential for systematically addressing a wide range of hardware bugs. For instance, existing LLM methodologies struggle to detect bugs involving incorrect constant values, i.e., the use of wrong constants in source code. This limitation underscores the need for further exploration in utilizing LLMs to fully optimize the verification process. To bridge this gap, we propose a 3-phased 4-stage LLM-assisted systematic bug closure methodology that focuses on functional bugs in Verilog HDL rather than structural or syntactic issues. Our approach extracts functional properties of the DUV and systematically breaks down complex expressions into smaller sub-expressions to facilitate bug detection and correction. By employing RAG, the LLM is guided using the functional specifications and source code to identify and correct bugs. If the initial guidance through RAG is insufficient, our methodology initiates an iterative bug closure process. This includes incorporating more extensive information from the specifications, fetching additional lines of code for bug localization, and breaking down complex Verilog HDL expressions. In our comprehensive evaluation, we assess the LLM’s capabilities using 9 different categories of bugs. As benchmarks, we use 5 OpenTitan Intellectual Property (IP) cores to demonstrate the scalability and effectiveness of our bug closure methodology where ≈ 60% of the bugs were corrected. Specifically, we evaluate OpenAI’s GPT-4 in its ability to identify and correct functional bugs in Verilog HDL code.

Widespread use of the 'unspecified coccidioidomycosis' code (B38.9) may negatively impact provider reimbursement and complicate study of disease burden. We sought to determine the frequency of B38.9 use in routine practice and assess how often it could be changed to a more specific code, what code that might be, and reasons for its initial use. To estimate the proportion of all coccidioidomycosis cases that were classified using the International Classification of Diseases-Tenth-Revision unspecified diagnosis code (B38.9), three real-world datasets were queried. Further, in January 2025, providers in coccidioidomycosis-endemic areas were invited to participate in an online electronic health record (EHR) audit to study their three most recent patients coded as B38.9 over the prior 12 months. The proportion of patients that could have received a more specific coccidioidomycosis code by provider was determined, as were the recommended alternative code(s), and reasons for the B38.9's initial use. Across queried datasets, 17.8-49.5% of coccidioidomycosis cases were coded as unspecified. We recruited 19 providers to audit EHRs of 53 patients, in which B38.9 was used. Thirty-six (67.9%) patients could have been assigned a more specific coccidioidomycosis code, including 14 (38.9%) to disseminated disease (B38.7). Common reasons for using B38.9 included evolving clinical assessment (37.0%), lack of coding expertise (22.2%), and entry errors (22.2%). In conclusion, a substantial proportion of coccidioidomycosis diagnoses are assigned to B38.9. Over two-thirds of these could have been better described using a more specific code. There is a need for educational efforts to promote more precise coding.

Abstract This study presents the design, development, and performance analysis of three types of Z-shaped metamaterial unit cell arrangements operating at terahertz frequencies. These unit cells were developed to evaluate their electromagnetic responses, and corresponding coding sequences were designed using the 1-bit coding method. In this method, 0-bit and 1-bit unit cells are defined by comparing the phase responses between different unit cells. All unit cell designs, numerical analyses of S-parameters, and radar cross-section (RCS) evaluations of the coding metamaterials were carried out using the Finite Integration Technique (FIT). The S-parameter results were further validated with the Finite Element Method (FEM). The investigation was conducted over the frequency range of 0–5 THz. Additionally, the transmission coefficient and RCS values of 6 × 6 coding sequences were analyzed to compare the performance of the different Z-shaped unit cells. The results showed slight variations depending on the specific unit cell type and coding sequence. Notably, coding sequence 2, which incorporates a Z-shaped resonator inside a square-ring element, achieved an RCS reduction of –48.65 dBm2. Overall, the findings demonstrate that 1-bit coding metamaterials based on optimized Z-shaped unit cells with carefully designed coding sequences can effectively manipulate electromagnetic waves and provide significant RCS reduction.

BackgroundIn 2018, electronic death certification was gradually introduced in the Norwegian Cause of Death Registry (NCoDR) and made mandatory from 2022. This may have affected the use and pattern of ill-defined death codes, so-called garbage codes, reported as the underlying cause of death. This study aimed to assess whether the transition to electronic certification changed the magnitude and pattern of garbage codes in NCoDR.MethodsThe analyses included all deaths with a registered underlying cause of death in the period 2018-2023 (n = 248 225). A garbage code was defined according to the definition used by the Global Burden of Disease study. Trends and patterns of garbage codes were assessed and its association with type of certificate (paper/electronic) estimated using a modified Poisson regression model adjusting for sex, age and place of death.ResultsIn the complete study period, 22.6% of the deaths were registered with a garbage code as the underlying cause of death. The probability of garbage was 21% lower in electronic compared to paper certificates (adjusted relative risk 0.79, 95% CI 0.78-0.80). This difference was stable over time and led to an overall decrease in the proportion of garbage codes with increasing use of electronic certifications, mainly driven by more specific coding. The five most common garbage codes were I50 (Heart failure), J18 (Pneumonia, organism unspecified), I64 (Unspecified stroke), R96 (Other sudden death, cause unknown) and X59 (Accidental exposure to unspecified factor), accounting for more than one third of the garbage codes in the study period.ConclusionsThe transition from paper to electronic certification led to an improvement in cause of death registration through a reduction in the use of garbage codes, although garbage codes remain a considerable problem. An electronic system can provide guidance and feedback to the certifying physician, thereby providing opportunities for further improvement in data quality.Key messages• The shift from paper-based to electronic certification in the Norwegian Cause of Death registry enhanced data quality through a reduction in the use of garbage codes.• Electronic certification provides opportunities for further improvement in data quality.

Wide Variation in Mitral Valve Repair Rates Among U.S. Surgeons: Analysis of Medicare Claims Data

Trauma is an important cause of hospitalizations and surgeries in emergency services, with high morbidity and mortality rates and in Brazil, it presents relevant statistics, with variations in mechanisms and topographies depending on the context. This study analyzed hospital data on vascular traumas by anatomical region in the Unified Health System (SUS) over the past 20 years, seeking to highlight relevant epidemiological information according to the particularities of vascular injuries in the Brazilian context. Ecological, descriptive, time-series study with a quantitative approach. Data were obtained through SIH/SUS, via TABWIN/DATASUS, between February and April 2024, referring to hospital admission records for the surgical treatment of vascular trauma, according to specific coding. There was a higher concentration of hospitalizations between 2012 and 2016, peaking in 2012, and a slight declining trend until 2024, with the highest number of hospitalizations (32%) in the Southeast. Men accounted for 76.6% of hospitalizations, but the relative risk of death was higher among women (RR=1.36). Upper limbs were the most affected, while abdominal traumas presented the highest lethality (22.1%) and lower limbs the longest hospital stay (6.6 days). Most procedures were performed under emergency conditions (88.4%), representing the highest mortality rate (6.3%). Although the majority of hospitalizations occurred among the population aged 15 to 49 years (60%), the highest mortality rates were observed at the age extremes: infants (11%) and the group over 50 years (10%).

This study stems from the growing need to uphold ethical standards within the Indonesian judicial system, particularly concerning prosecutors, who play a pivotal role in law enforcement and justice. The lack of a specific and binding code of ethics for prosecutors has raised concerns about accountability, professionalism, and integrity in the exercise of their duties. This research aims to analyze the urgency of establishing a code of ethics for prosecutors and examine the standards of ethical behavior in accordance with the 1945 Constitution of the Republic of Indonesia. This is a normative legal study employing a legislative and conceptual approach. Data were collected through an extensive literature review and analyzed using qualitative methods. The findings indicate that a dedicated code of ethics is essential for guiding prosecutors in carrying out their duties, functions, and authorities with integrity, professionalism, and wisdom. Ethical integrity emphasizes loyalty to Pancasila and the 1945 Constitution, while professionalism demands adherence to legal and institutional regulations. Wisdom reflects the prosecutor’s responsibility as a public servant who respects prevailing religious, moral, and societal norms. Ethical standards are an extension of the rule of law principle as enshrined in Article 1 paragraph (3) of the 1945 Constitution.

Adequacy of the Diagnostic Workup of Suspected Physical Abuse in Young Children: An Audit in the Paris Metropolitan Area.

ObjectiveInjection laryngoplasty (IL) is performed to correct glottic insufficiency. There has been a purported shift away from operative techniques in favor of awake, in‐office procedures, but no studies comparing utilization include updated current procedural terminology (CPT) coding. We analyzed the usage of operative versus awake injections CPT codes over 2 decades, recognizing that these encompass a broad array of injection procedures.Study DesignRetrospective database study.SettingUnited States Medicare Population from 2001 to 2022.MethodsUtilization and reimbursement data compiled by the US Centers for Medicare & Medicaid Services were queried for CPT codes encompassing awake injections (31513, 31573, 31574), operative ILs (31570, 31571), and operative medialization laryngoplasties (31588, 31591).Results10,186 injections were performed in 2022, a 195% increase from 2001. Awake injections grew by 412.77%, while operative injections grew by 134.75%. Operative injections are still more common (79.4% in 2001; 64.24% in 2022) and population‐adjusted use of 31571 grew between 2001 and 2021, while 31570 decreased. Since its adoption in 2017, awake IL code 31574 increased by 66.6%. From 2001 to 2006, the annual growth rate of awake injections was significantly lower than that of operative IL (P < .0001). From 2017 onwards, the awake IL growth rate was significantly higher than operative injections (P = .020).ConclusionsAwake injection utilization increased over the 22‐year period, as introduction of code 31574 coincided with a relative decline in operative and an increase in awake IL. Otolaryngologists may be implementing awake injections due to reduced recovery time and introduction of more specific CPT codes.

Retrospective cohort study. To evaluate the accuracy of ICD-10 codes in identifying isolated degenerative cervical radiculopathy versus concurrent myelopathy in surgical patients across an 11-hospital health care system (2016-2023). ICD-10 codes are commonly used for patient stratification in database studies, yet their accuracy in differentiating cervical radiculopathy from concurrent myelopathy remains poorly characterized. This distinction is clinically critical as treatment approaches and outcomes differ significantly between these conditions. We reviewed 830 surgical patients identified using ICD-10 codes for cervical radiculopathy (M50.10-50.13, M54.11-M54.13, or M25.78). Patients with codes for myelopathy, traumatic, oncologic, or infectious etiologies were excluded. Primary diagnoses were determined through standardized chart review by 3 independent clinical reviewers blinded to ICD-10 codes. Patients were classified using specific criteria: radiculopathy (dermatomal symptoms, positive provocative tests, and foraminal stenosis) and myelopathy (upper motor neuron signs, gait disturbances, fine motor deficits, and cord signal changes on MRI). Chart review revealed 637 patients (76.7%) had isolated cervical radiculopathy, confirming ICD-10 coding. However, 151 patients (18.2%) presented with concurrent myelopathic symptoms (myeloradiculopathy), and 35 patients (4.2%) demonstrated isolated cervical myelopathy, despite being coded only for radiculopathy. Nondegenerative etiologies comprised 0.8% of cases. The nonspecific code M25.78 showed higher error rates (26%) compared with specific codes, though M54.12 still demonstrated a 22% error rate. Code position significantly influenced accuracy, with error rates increasing from 14% (primary position) to 43% (fourth position). A significant proportion of patients coded for isolated cervical radiculopathy have concurrent myelopathy that goes unrecognized with ICD-10 coding alone. Code specificity and diagnostic position significantly impact accuracy. Future strategies, including decision support tools and natural language processing, may improve coding accuracy. Level III.

National trends in laryngectomy and the influence of hospital volume on short-term outcomes in Brazil: A 16-year cross-sectional analysis

Modern imaging techniques such as optical coherence tomography (OCT), lateral-confocal OCT, and confocal laser microscopy (KLM) are now well established in dermatology, enhancing diagnostic precision and therapeutic decisions. Currently, billing is handled via analogous codes in the German fee schedule (GOÄ), leading to legal and reimbursement challenges. Although the upcoming revised GOÄ will include specific codes, this may result in lower compensation. Public health insurance does not yet cover these services. Legal and reimbursement frameworks must be urgently adapted to ensure sustainable integration of these technologies into clinical practice.

The Clinical Practice Research Datalink (CPRD) is a large and widely used resource of electronic health records from the UK, linking primary care data to hospital data, death registration data, cancer registry data, deprivation data and mental health services data. Extraction and management of CPRD data is a computationally demanding process and requires a significant amount of work, in particular when using R. The rcprd package simplifies the process of extracting and processing CPRD data in order to build datasets ready for statistical analysis. Raw CPRD data is provided in thousands of.txt files, making querying this data cumbersome and inefficient. rcprd saves the relevant information into an SQLite database stored on the hard drive which can then be queried efficiently to extract required information about individuals. rcprd follows a four-stage process: 1) Definition of a cohort, 2) Read in medical/prescription data and save into an SQLite database, 3) Query this SQLite database for specific codes and tests to create variables for each individual in the cohort, 4) Combine extracted variables into a dataset ready for statistical analysis. Functions are available to extract common variable types (e.g., history of a condition, or time until an event occurs, relative to an index date), and more general functions for database queries, allowing users to define their own variables for extraction. The entire process can be done from within R, with no knowledge of SQL required. This manuscript showcases the functionality of rcprd by running through an example using simulated CPRD Aurum data. rcprd will reduce the duplication of time and effort among those using CPRD data for research, allowing more time to be focused on other aspects of research projects.

Healthcare administrative data often rely on the International Classification of Diseases (ICD) system, which lacks specific codes to identify etiological subgroups of epilepsy. Combining indicators for epilepsy and potential etiologies is possible, but such approaches require validation. This study aimed to validate methods for identifying poststroke epilepsy (PSE) in Swedish administrative data. The algorithms were based on combinations of ICD-10 codes for stroke and seizures, with some also incorporating antiseizure medication prescriptions. We focused on positive predictive values (PPVs), using medical records as the reference standard. We identified individuals in the National Patient Register with a primary inpatient diagnostic code for stroke (I61 or I63) during 2005-2010 and a first-ever seizure-related code (G40, G41, or R56.8), occurring more than seven days post-stroke. To facilitate access to medical records, only patients who were deceased at data extraction (Jan 16, 2021) were eligible. A nationwide random sample of 500 patients was selected, with the intended sample for medical record review being 250. Medical records were reviewed before processing the administrative data. Records were obtained for 321 patients (median age 78; 56% males), with no significant differences in characteristics between those included and the rest of the sample. Across different algorithms, PPVs ranged from 84.1% (95% CI: 79.2-88.3) to 92.5% (95% CI: 87.3-96.1). Relative coverage ranged from 60% to 89% compared to the most inclusive algorithm. Our findings demonstrate the potential of administrative data to reliably identify PSE cases, supporting the use of these algorithms for large-scale studies of treatment and outcomes. Stricter algorithms, limited to G40 codes for epilepsy or requiring ASM prescriptions, improve accuracy but at the cost of missing more cases. Limitations include the inability to calculate sensitivity due to study design, and the need for local validation before use in other healthcare systems.