Federated machine learning for predicting acute kidney injury in critically ill patients: a multicenter study in Taiwan.

Purpose: To address the contentious data sharing across hospitals, this study adopted a novel approach, federated learning (FL), to establish an aggregate model for acute kidney injury (AKI) prediction in critically ill patients in Taiwan. Methods: This study used data from the Critical Care Database of Taichung Veterans General Hospital (TCVGH) from 2015 to 2020 and electrical medical records of the intensive care units (ICUs) between 2018 and 2020 of four referral centers in different areas across Taiwan. AKI prediction models were trained and validated thereupon. An FL-based prediction model across hospitals was then established. Results: The study included 16,732 ICU admissions from the TCVGH and 38,424 ICU admissions from the other four hospitals. The complete model with 60 features and the parsimonious model with 21 features demonstrated comparable accuracies using extreme gradient boosting, neural network (NN), and random forest, with an area under the receiver-operating characteristic (AUROC) curve of approximately 0.90. The Shapley Additive Explanations plot demonstrated that the selected features were the key clinical components of AKI for critically ill patients. The AUROC curve of the established parsimonious model for external validation at the four hospitals ranged from 0.760 to 0.865. NN-based FL slightly improved the model performance at the four centers. Conclusion: A reliable prediction model for AKI in ICU patients was developed with a lead time of 24 h, and it performed better when the novel FL platform across hospitals was implemented. Supplementary Information: The online version contains supplementary material available at 10.1007/s13755-023-00248-5.

The association between the severity of periodontitis and osteonecrosis of the jaw in patients with different cancer locations: a nationwide population-based study.

OBJECTIVES: To determine the relation between the severity of periodontitis and osteonecrosis of the jaw (ONJ) occurrence among different cancer locations and estimate the effect of dental care on ONJ prevention in cancer patients. MATERIALS AND METHODS: This population-based cross-sectional study was conducted through the Longitudinal Health Insurance Database, Taiwan. Patients with malignancies were collected and subdivided into groups according to their different cancer locations, the severity of periodontitis, and dental care. Multivariable logistic regression analysis was performed to assess the associations between ONJ and ONJ-related factors. RESULTS: A total of 8,234 ONJ patients and 32,912 control patients were investigated. Lip, oral cavity, and pharynx malignancies had the highest ONJ risk among all cancer locations (OR from 3.07 to 9.56, P < 0.01). There is a linear relationship between different severities of periodontitis and ONJ. Patients with radiotherapy and severe periodontitis had the highest ONJ risk (adjusted OR, 9.56; 95% CI, 5.34-17.1). Patients with good dental care had a lower ONJ risk. CONCLUSIONS: The periodontal condition and cancer location showed a significant impact on the risk of developing ONJ after adjusting for bisphosphonate use. Good dental care could decrease the risk of ONJ in cancer patients. The severity of periodontitis might be a target to predict the potency of ONJ. CLINICAL RELEVANCE: Dentists must be vigilant about the increased risk of ONJ in cancer patients with periodontitis, especially in the head and neck cancer population. Good dental care is advised for cancer patients with severe periodontitis.

Tracking Hospital Visitors/Chaperones during the COVID-19 Pandemic.

OBJECTIVE: The coronavirus disease (COVID-19) is an emerging infectious disease with strong infectious power and fatality rate. To protect national health, government agencies have regulations on hospital chaperoning and visiting. This article presents the development and implementation of a monitoring system for hospital visiting and chaperoning during the COVID-19 pandemic. The study aimed to create a hospital visiting and chaperoning monitor system that uses nation-wide data sources to more accurately screen hospital visitors and chaperones, assist contract tracing, and prevent transmission of severe acute respiratory syndrome coronavirus 2. METHODS: This project was implemented in 57 ward units of an academic medical center. The system was connected to the National Health Insurance (NHI) system and Hospital Information System (HIS), and built on the data of everyone who accessed either the hospital or ward using an NHI smart card or national identification card. To shorten the time for manual identification, we also developed a new system of "app for appointment visits and chaperones" to make appointments online. RESULTS: After the implementation of the system, data from visitors and chaperones in the nursing information system could be accessed. Given that all data were registered in the HIS visiting/chaperoning monitor system, an epidemic investigation could be performed whenever there was a confirmed case. CONCLUSION: Through the establishment of this system, people entering the ward can be accurately controlled, and all the contacts of potential cases can be traced.

Molecular survey of selected viral pathogens in wild leopard cats (Prionailurus bengalensis) in Taiwan with an emphasis on the spatial and temporal dynamics of carnivore protoparvovirus 1.

The leopard cat (Prionailurus bengalensis) was listed as an endangered species under the Wildlife Conservation Act in Taiwan in 2009. However, no study has evaluated the possible direct or indirect effects of pathogens on the Taiwanese leopard cat population. Here, we targeted viral pathogens, including carnivore protoparvovirus 1 (genus Protoparvovirus), feline leukemia virus (FeLV), feline immunodeficiency virus (FIV), coronaviruses (CoVs), and canine distemper virus (CDV), through molecular screening. The spatial and temporal dynamics of the target pathogens were evaluated. Through sequencing and phylogenetic analysis, we clarified the phylogenetic relationship of viral pathogens isolated from leopard cats and domestic carnivores. Samples from 23 live-trapped leopard cats and 29 that were found dead were collected from 2015 to 2019 in Miaoli County in northwestern Taiwan. Protoparvoviruses and CoVs were detected in leopard cats, and their prevalence (95% confidence interval) was 63.5% (50.4%-76.6%) and 8.8% (0%-18.4%), respectively. Most of the protoparvovirus sequences amplified from Taiwanese leopard cats and domestic carnivores were identical. All of the CoV sequences amplified from leopard cats were identified as feline CoV. No spatial or temporal aggregation of protoparvovirus infection in leopard cats was found in the sampling area, indicating a wide distribution of protoparvoviruses in the leopard cat habitat. We consider sympatric domestic carnivores to be the probable primary reservoir for the identified pathogens. We strongly recommend management of protoparvoviruses and feline CoV in the leopard cat habitat, particularly vaccination programs and population control measures for free-roaming dogs and cats.