Patient-Centered Care Through Nurse Practitioner-Led Integrated Behavioral Health: A Case Study.

Integrated behavioral health can improve primary care and mental health outcomes. Access to behavioral health and primary care services in Texas is in crisis because of high uninsurance rates, regulatory restrictions, and lack of workforce. To address gaps in access to care, a partnership formed among a large local mental health authority in central Texas, a federally designated rural health clinic, and the Texas A&M University School of Nursing to create an interprofessional team-based health care delivery model led by nurse practitioners in rural and medically underserved areas of central Texas. Academic-practice partners identified 5 clinics for an integrated behavioral health care delivery model. From July 1, 2020, through December 31, 2021, a total of 3183 patient visits were completed. Patients were predominantly female (n = 1719, 54%) and Hispanic (n = 1750, 55%); 1050 (33%) were living at or below the federal poverty level; and 1400 (44%) were uninsured. The purpose of this case study was to describe the first year of implementation of the integrated health care delivery model, barriers to implementation, challenges to sustainability, and successes. We analyzed data from multiple sources, including meeting minutes and agendas, grant reports, direct observations of clinic flow, and interviews with clinic staff, and identified common qualitative themes (eg, challenges to integration, sustainability of integration, outcome successes). Results revealed implementation challenges with the electronic health record, service integration, low staffing levels during a global pandemic, and effective communication. We also examined 2 patient cases to illustrate the success of integrated behavioral health and highlighted lessons learned from the implementation process, including the need for a robust electronic health record and organizational flexibility.

Number and timing of primary cleft lip and palate repair surgeries in England: whole nation study of electronic health records before and during the COVID-19 pandemic.

OBJECTIVE: To quantify differences in number and timing of first primary cleft lip and palate (CLP) repair procedures during the first year of the COVID-19 pandemic (1 April 2020 to 31 March 2021; 2020/2021) compared with the preceding year (1 April 2019 to 31 March 2020; 2019/2021). DESIGN: National observational study of administrative hospital data. SETTING: National Health Service hospitals in England. STUDY POPULATION: Children <5 years undergoing primary repair for an orofacial cleft Population Consensus and Surveys Classification of Interventions and Procedures-fourth revisions (OPCS-4) codes F031, F291). MAIN EXPOSURE: Procedure date (2020/2021 vs 2019/2020). MAIN OUTCOMES: Numbers and timing (age in months) of first primary CLP procedures. RESULTS: 1716 CLP primary repair procedures were included in the analysis. In 2020/2021, 774 CLP procedures were carried out compared with 942 in 2019/2020, a reduction of 17.8% (95% CI 9.5% to 25.4%). The reduction varied over time in 2020/2021, with no surgeries at all during the first 2 months (April and May 2020). Compared with 2019/2020, first primary lip repair procedures performed in 2020/2021 were delayed by 1.6 months on average (95% CI 0.9 to 2.2 months). Delays in primary palate repairs were smaller on average but varied across the nine geographical regions. CONCLUSION: There were significant reductions in the number and delays in timing of first primary CLP repair procedures in England during the first year of the pandemic, which may affect long-term outcomes.

Constructing a disease database and using natural language processing to capture and standardize free text clinical information.

The ability to extract critical information about an infectious disease in a timely manner is critical for population health research. The lack of procedures for mining large amounts of health data is a major impediment. The goal of this research is to use natural language processing (NLP) to extract key information (clinical factors, social determinants of health) from free text. The proposed framework describes database construction, NLP modules for locating clinical and non-clinical (social determinants) information, and a detailed evaluation protocol for evaluating results and demonstrating the effectiveness of the proposed framework. The use of COVID-19 case reports is demonstrated for data construction and pandemic surveillance. The proposed approach outperforms benchmark methods in F1-score by about 1-3%. A thorough examination reveals the disease's presence as well as the frequency of symptoms in patients. The findings suggest that prior knowledge gained through transfer learning can be useful when researching infectious diseases with similar presentations in order to accurately predict patient outcomes.

Implementation of medication reconciliation in outpatient cancer care.

BACKGROUND: Medication reconciliation (MedRec) is a process where providers work with patients to document and communicate comprehensive medication information by creating a complete medication list (best possible medication history (BPMH)) then reconciling it against what patient is actually taking to identify potential issues such as drug-drug interactions. We undertook an environmental scan of current MedRec practices in outpatient cancer care to inform a quality improvement project at our centre with the aim of 30% of patients having a BPMH or MedRec within 30 days of initiating treatment with systemic therapy. METHODS: We conducted semi-structured interviews with key stakeholders from 21 cancer centres across Canada, probing on current policies, and barriers and facilitators to MedRec. Guided by the findings of the scan, we then undertook a quality improvement project at our cancer centre, comprising six iterative improvement cycles. RESULTS: Most institutions interviewed had a process in place for collecting a BPMH (81%) and targeted patients initiating systemic therapy (59%); however, considerable practice variation was noted and completion of full MedRec was uncommon. Lack of resources, high patient volumes, lack of a common medical record spanning institutions and settings which limits access to medication records from external institutions and community pharmacies were identified as significant barriers. Despite navigating challenges related to the COVID-19 pandemic, we achieved 26.6% of eligible patients with a documented BPMH. However, uptake of full MedRec remained low whereby 4.7% of patients had a documented MedRec. CONCLUSIONS: Realising improvements to completion of MedRec in outpatient cancer care is possible but takes considerable time and iteration as the process is complex. Resource allocation and information sharing remain major barriers which need to be addressed in order to observe meaningful improvements in MedRec.

An increase in cesarean section rate during the first wave of COVID-19 pandemic in Iran.

BACKGROUND: The COVID-19 pandemic and its impact on healthcare services is likely to affect birth outcomes including the delivery mode. However, recent evidence has been conflicting in this regard. The study aimed to assess changes to C-section rate during the COVID-19 pandemic in Iran. METHODS: This is a retrospective analysis of electronic medical records of women delivered in the maternity department of hospitals in all provinces of Iran before the COVID-19 pandemic (February-August 30, 2019) and during the pandemic (February-August 30, 2020). Data were collected through the Iranian Maternal and Neonatal Network (IMAN), a country-wide electronic health record database management system for maternal and neonatal information. A total of 1,208,671 medical records were analyzed using the SPSS software version 22. The differences in C-section rates according to the studied variables were tested using the χ2 test. A logistic regression analysis was conducted to determine the factors associated with C-section. RESULTS: A significant rise was observed in the rates of C-section during the pandemic compared to the pre-pandemic (52.9% vs 50.8%; p = .001). The rates for preeclampsia (3.0% vs 1.3%), gestational diabetes (6.1% vs 3.0%), preterm birth (11.6% vs 6.9%), IUGR (1.2% vs 0.4%), LBW (11.2% vs 7.8%), and low Apgar score at first minute (4.2% vs 3.2%) were higher in women who delivered by C-section compared to those with normal delivery (P = .001). CONCLUSIONS: The overall C-section rate during the first wave of COVID-19 pandemic was significantly higher than the pre-pandemic period. C-section was associated with adverse maternal and neonatal outcomes. Thus, preventing the overuse of C-section especially during pandemic becomes an urgent need for maternal and neonatal health in Iran.

De-black-boxing health AI: demonstrating reproducible machine learning computable phenotypes using the N3C-RECOVER Long COVID model in the All of Us data repository.

Machine learning (ML)-driven computable phenotypes are among the most challenging to share and reproduce. Despite this difficulty, the urgent public health considerations around Long COVID make it especially important to ensure the rigor and reproducibility of Long COVID phenotyping algorithms such that they can be made available to a broad audience of researchers. As part of the NIH Researching COVID to Enhance Recovery (RECOVER) Initiative, researchers with the National COVID Cohort Collaborative (N3C) devised and trained an ML-based phenotype to identify patients highly probable to have Long COVID. Supported by RECOVER, N3C and NIH's All of Us study partnered to reproduce the output of N3C's trained model in the All of Us data enclave, demonstrating model extensibility in multiple environments. This case study in ML-based phenotype reuse illustrates how open-source software best practices and cross-site collaboration can de-black-box phenotyping algorithms, prevent unnecessary rework, and promote open science in informatics.

No Transfer Without Validation: A Data Sharing Framework Use Case.

Availability and accessibility are important preconditions for using real-world patient data across organizations. To facilitate and enable the analysis of data collected at a large number of independent healthcare providers, syntactic- and semantic uniformity need to be achieved and verified. With this paper, we present a data transfer process implemented using the Data Sharing Framework to ensure only valid and pseudonymized data is transferred to a central research repository and feedback on success or failure is provided. Our implementation is used within the CODEX project of the German Network University Medicine to validate COVID-19 datasets at patient enrolling organizations and securely transfer them as FHIR resources to a central repository.

PDSS: A Pharmacological Decision Support System for Diabetics Patients with COVID-19.

With the advent of SARS-CoV-2, several studies have shown that there is a higher mortality rate in patients with diabetes and, in some cases, it is one of the side effects of overcoming the disease. However, there is no clinical decision support tool or specific treatment protocols for these patients. To tackle this issue, in this paper we present a Pharmacological Decision Support System (PDSS) providing intelligent decision support for COVID-19 diabetic patient treatment selection, based on an analysis of risk factors with data from electronic medical records using Cox regression. The goal of the system is to create real world evidence including the ability to continuously learn to improve clinical practice and outcomes of diabetic patients with COVID-19.

Study protocol for the use of time series forecasting and risk analyses to investigate the effect of the COVID-19 pandemic on hospital admissions associated with new-onset disability and frailty in a national, linked electronic health data setting.

INTRODUCTION: Older people were at particular risk of morbidity and mortality during COVID-19. Consequently, they experienced formal (externally imposed) and informal (self-imposed) periods of social isolation and quarantine. This is hypothesised to have led to physical deconditioning, new-onset disability and frailty. Disability and frailty are not routinely collated at population level but are associated with increased risk of falls and fractures, which result in hospital admissions. First, we will examine incidence of falls and fractures during COVID-19 (January 2020-March 2022), focusing on differences between incidence over time against expected rates based on historical data, to determine whether there is evidence of new-onset disability and frailty. Second, we will examine whether those with reported SARS-CoV-2 were at higher risk of falls and fractures. METHODS AND ANALYSIS: This study uses the Office for National Statistics (ONS) Public Health Data Asset, a linked population-level dataset combining administrative health records with sociodemographic data of the 2011 Census and National Immunisation Management System COVID-19 vaccination data for England. Administrative hospital records will be extracted based on specific fracture-centric International Classification of Diseases-10 codes in years preceding COVID-19 (2011-2020). Historical episode frequency will be used to predict expected admissions during pandemic years using time series modelling, if COVID-19 had not occurred. Those predicted admission figures will be compared with actual admissions to assess changes in hospital admissions due to public health measures comprising the pandemic response. Hospital admissions in prepandemic years will be stratified by age and geographical characteristics and averaged, then compared with pandemic year admissions to assess more granular changes. Risk modelling will assess risk of experiencing a fall, fracture or frail fall and fracture, if they have reported a positive case of COVID-19. The combination of these techniques will provide insight into changes in hospital admissions from the COVID-19 pandemic. ETHICS AND DISSEMINATION: This study has approval from the National Statistician's Data Ethics Advisory Committee (NSDEC(20)12). Results will be made available to other researchers via academic publication and shared via the ONS website.

Deep learning approach for early prediction of COVID-19 mortality using chest X-ray and electronic health records.

BACKGROUND: An artificial-intelligence (AI) model for predicting the prognosis or mortality of coronavirus disease 2019 (COVID-19) patients will allow efficient allocation of limited medical resources. We developed an early mortality prediction ensemble model for COVID-19 using AI models with initial chest X-ray and electronic health record (EHR) data. RESULTS: We used convolutional neural network (CNN) models (Inception-ResNet-V2 and EfficientNet) for chest X-ray analysis and multilayer perceptron (MLP), Extreme Gradient Boosting (XGBoost), and random forest (RF) models for EHR data analysis. The Gradient-weighted Class Activation Mapping and Shapley Additive Explanations (SHAP) methods were used to determine the effects of these features on COVID-19. We developed an ensemble model (Area under the receiver operating characteristic curve of 0.8698) using a soft voting method with weight differences for CNN, XGBoost, MLP, and RF models. To resolve the data imbalance, we conducted F1-score optimization by adjusting the cutoff values to optimize the model performance (F1 score of 0.77). CONCLUSIONS: Our study is meaningful in that we developed an early mortality prediction model using only the initial chest X-ray and EHR data of COVID-19 patients. Early prediction of the clinical courses of patients is helpful for not only treatment but also bed management. Our results confirmed the performance improvement of the ensemble model achieved by combining AI models. Through the SHAP method, laboratory tests that indicate the factors affecting COVID-19 mortality were discovered, highlighting the importance of these tests in managing COVID-19 patients.

Frequency of Parkinson disease following COVID-19 infection: A two-year retrospective cohort study.

INTRODUCTION: COVID-19 infection is known to cause various neurological symptoms, and potentially increases the risk of developing subsequent neurodegenerative conditions including parkinsonism. To our knowledge, no study to date has used a large data set in the United States to ascertain the risk of developing incident Parkinson disease in patients with history of COVID-19 infection compared to the risk amongst those without prior COVID-19 infection. METHODS: We utilized data from TriNetX electronic health records network which includes 73 healthcare organizations and over 107 million patients. We compared adult patients with and without COVID-19 infection, with health records from January 1, 2020 through July 26, 2022, to determine the relative risk of developing Parkinson disease stratified by 3-month intervals. We used propensity score matching to control for patients' age, sex, and smoking history. RESULTS: We collected data on 27,614,510 patients meeting our study criteria: 2,036,930 patients with a positive COVID-19 infection (COVID-19) and 25,577,580 without a positive COVID-19 infection (non-COVID-19). After propensity score matching, age, sex, and smoking history differences became non-significant, with 2,036,930 patients in each cohort. After propensity score matching, we found significantly increased odds of new onset Parkinson disease in the COVID-19 cohort at three, six, nine, and twelve months from the index event, with peak odds ratio at six months. After twelve months there is no significant difference between the COVID-19 group and non-COVID-19 group. CONCLUSIONS: There may be a transiently increased risk of developing Parkinson disease in the first year following COVID-19 infection.

Cyberthreats: A primer for healthcare professionals.

INTRODUCTION: Cyberattacks are one of the most widespread, damaging, and disruptive forms of action against healthcare entities. Data breaches, ransomware attacks, and other intrusions can lead to significant cost both in monetary and personal harm to those affected and may result in large payouts to cyber criminals, crashes of information technology systems, leaks of protected health and personal information, as well as fines and lawsuits. This study is a descriptive analysis of healthcare-related cyber breaches affecting 500 or more individuals in the past decade in the United States. METHODS: The publicly available U.S. breach report database was downloaded in the Microsoft Excel (Microsoft, Redmond, Washington, USA) format and searched for all reported breaches occurring between January 1, 2011 - December 31, 2021 (10 years). Breaches were subdivided by category and analyzed by states, breach submission dates, types of breach, location of breached information, entity type, and individuals affected. All subcategories were predefined by the breach report. RESULTS: There were a total of 3822 PHI breaches that affected 283,335,803 people in the United States from January 1, 2011 to December 31, 2021. Of the 3822 PHI breaches, 1593 (41.7%) were hacking/ IT related, 1055 (27.6%) were listed as unknown, 819 (21.4%) were theft related, 194 (5.1%) were loss related, 97 (2.5%) were related to improper disposal and 64 (1.7%) were listed as "others". Year 2020 saw the most breaches with 631 and California was the state with the highest number of breaches at 403. CONCLUSION: Cyberattacks and healthcare breaches are one of the most costly and disruptive situations facing healthcare today. A total of 3822 breaches affecting 283,335,803 people in the United States were recorded from January 1, 2011 to December 31, 2021. By understanding the extent of cyberthreats this will better prepare healthcare organizations and providers to mitigate, respond, and recover from these devastating attacks.

Comparison of mRNA vaccine effectiveness against COVID-19-associated hospitalization by vaccination source: Immunization information systems, electronic medical records, and self-report-IVY Network, February 1-August 31, 2022.

BACKGROUND: Accurate determination of COVID-19 vaccination status is necessary to produce reliable COVID-19 vaccine effectiveness (VE) estimates. Data comparing differences in COVID-19 VE by vaccination sources (i.e., immunization information systems [IIS], electronic medical records [EMR], and self-report) are limited. We compared the number of mRNA COVID-19 vaccine doses identified by each of these sources to assess agreement as well as differences in VE estimates using vaccination data from each individual source and vaccination data adjudicated from all sources combined. METHODS: Adults aged ≥18 years who were hospitalized with COVID-like illness at 21 hospitals in 18 U.S. states participating in the IVY Network during February 1-August 31, 2022, were enrolled. Numbers of COVID-19 vaccine doses identified by IIS, EMR, and self-report were compared in kappa agreement analyses. Effectiveness of mRNA COVID-19 vaccines against COVID-19-associated hospitalization was estimated using multivariable logistic regression models to compare the odds of COVID-19 vaccination between SARS-CoV-2-positive case-patients and SARS-CoV-2-negative control-patients. VE was estimated using each source of vaccination data separately and all sources combined. RESULTS: A total of 4499 patients were included. Patients with ≥1 mRNA COVID-19 vaccine dose were identified most frequently by self-report (n = 3570, 79 %), followed by IIS (n = 3272, 73 %) and EMR (n = 3057, 68 %). Agreement was highest between IIS and self-report for 4 doses with a kappa of 0.77 (95 % CI = 0.73-0.81). VE point estimates of 3 doses against COVID-19 hospitalization were substantially lower when using vaccination data from EMR only (VE = 31 %, 95 % CI = 16 %-43 %) than when using all sources combined (VE = 53 %, 95 % CI = 41 %-62%). CONCLUSION: Vaccination data from EMR only may substantially underestimate COVID-19 VE.

Multimodal representation learning for predicting molecule-disease relations.

MOTIVATION: Predicting molecule-disease indications and side effects is important for drug development and pharmacovigilance. Comprehensively mining molecule-molecule, molecule-disease and disease-disease semantic dependencies can potentially improve prediction performance. METHODS: We introduce a Multi-Modal REpresentation Mapping Approach to Predicting molecular-disease relations (M2REMAP) by incorporating clinical semantics learned from electronic health records (EHR) of 12.6 million patients. Specifically, M2REMAP first learns a multimodal molecule representation that synthesizes chemical property and clinical semantic information by mapping molecule chemicals via a deep neural network onto the clinical semantic embedding space shared by drugs, diseases and other common clinical concepts. To infer molecule-disease relations, M2REMAP combines multimodal molecule representation and disease semantic embedding to jointly infer indications and side effects. RESULTS: We extensively evaluate M2REMAP on molecule indications, side effects and interactions. Results show that incorporating EHR embeddings improves performance significantly, for example, attaining an improvement over the baseline models by 23.6% in PRC-AUC on indications and 23.9% on side effects. Further, M2REMAP overcomes the limitation of existing methods and effectively predicts drugs for novel diseases and emerging pathogens. AVAILABILITY AND IMPLEMENTATION: The code is available at https://github.com/celehs/M2REMAP, and prediction results are provided at https://shiny.parse-health.org/drugs-diseases-dev/. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Racial differences in COVID-19 severity associated with history of substance use disorders and overdose: Findings from multi-site electronic health records in New York City.

Substance use disorders (SUD) are associated with increased risk of worse COVID-19 outcomes. Likewise, racial/ethnic minority patients experience greater risk of severe COVID-19 disease compared to white patients. Providers should understand the role of race and ethnicity as an effect modifier on COVID-19 severity among individuals with SUD. This retrospective cohort study assessed patient race/ethnicity as an effect modifier of the risk of severe COVID-19 disease among patients with histories of SUD and overdose. We used merged electronic health record data from 116,471 adult patients with a COVID-19 encounter between March 2020 and February 2021 across five healthcare systems in New York City. Exposures were patient histories of SUD and overdose. Outcomes were risk of COVID-19 hospitalization and subsequent COVID-19-related ventilation, acute kidney failure, sepsis, and mortality. Risk factors included patient age, sex, and race/ethnicity, as well as medical comorbidities associated with COVID-19 severity. We tested for interaction between SUD and patient race/ethnicity on COVID-19 outcomes. Findings showed that Non-Hispanic Black, Hispanic/Latino, and Asian/Pacific Islander patients experienced a higher prevalence of all adverse COVID-19 outcomes compared to non-Hispanic white patients. Past-year alcohol (OR 1.24 [1.01-1.53]) and opioid use disorders (OR 1.91 [1.46-2.49]), as well as overdose history (OR 4.45 [3.62-5.46]), were predictive of COVID-19 mortality, as well as other adverse COVID-19 outcomes. Among patients with SUD, significant differences in outcome risk were detected between patients of different race/ethnicity groups. Findings indicate that providers should consider multiple dimensions of vulnerability to adequately manage COVID-19 disease among populations with SUDs.

Clinical encounter heterogeneity and methods for resolving in networked EHR data: a study from N3C and RECOVER programs.

OBJECTIVE: Clinical encounter data are heterogeneous and vary greatly from institution to institution. These problems of variance affect interpretability and usability of clinical encounter data for analysis. These problems are magnified when multisite electronic health record (EHR) data are networked together. This article presents a novel, generalizable method for resolving encounter heterogeneity for analysis by combining related atomic encounters into composite "macrovisits." MATERIALS AND METHODS: Encounters were composed of data from 75 partner sites harmonized to a common data model as part of the NIH Researching COVID to Enhance Recovery Initiative, a project of the National Covid Cohort Collaborative. Summary statistics were computed for overall and site-level data to assess issues and identify modifications. Two algorithms were developed to refine atomic encounters into cleaner, analyzable longitudinal clinical visits. RESULTS: Atomic inpatient encounters data were found to be widely disparate between sites in terms of length-of-stay (LOS) and numbers of OMOP CDM measurements per encounter. After aggregating encounters to macrovisits, LOS and measurement variance decreased. A subsequent algorithm to identify hospitalized macrovisits further reduced data variability. DISCUSSION: Encounters are a complex and heterogeneous component of EHR data and native data issues are not addressed by existing methods. These types of complex and poorly studied issues contribute to the difficulty of deriving value from EHR data, and these types of foundational, large-scale explorations, and developments are necessary to realize the full potential of modern real-world data. CONCLUSION: This article presents method developments to manipulate and resolve EHR encounter data issues in a generalizable way as a foundation for future research and analysis.

Nirmatrelvir and risk of hospital admission or death in adults with covid-19: emulation of a randomized target trial using electronic health records.

OBJECTIVE: To estimate the effectiveness of nirmatrelvir, compared with no treatment, in reducing admission to hospital or death at 30 days in people infected with the SARS-CoV-2 virus and at risk of developing severe disease, according to vaccination status and history of previous SARS-CoV-2 infection. DESIGN: Emulation of a randomized target trial with electronic health records. SETTING: Healthcare databases of the US Department of Veterans Affairs PARTICIPANTS: 256 288 participants with a SARS-CoV-2 positive test result and at least one risk factor for developing severe covid-19 disease, between 3 January and 30 November 2022. 31 524 were treated with nirmatrelvir within five days of testing positive for SARS-CoV-2 and 224 764 received no treatment. MAIN OUTCOME MEASURES: The effectiveness of starting nirmatrelvir within five days of a positive SARS-CoV-2 test result versus no treatment in reducing the risk of admission to hospital or death at 30 days was estimated in those who were not vaccinated, in those who received one or two doses of vaccine, and those who received a vaccine booster and, separately, in participants with a primary SARS-CoV-2 infection or reinfection. The inverse probability weighting method was used to balance personal and health characteristics between the groups. Relative risk and absolute risk reduction were computed from cumulative incidence at 30 days, estimated by weighted Kaplan-Meier estimator. RESULTS: Among people who were not vaccinated (n=76 763; 5338 nirmatrelvir and 71 425 no treatment), compared with no treatment, the relative risk of nirmatrelvir in reducing admission to hospital or death at 30 days was 0.60 (95% confidence interval 0.50 to 0.71); the absolute risk reduction was 1.83% (95% confidence interval 1.29% to 2.49%). The relative risk and absolute risk reduction, compared with no treatment, were 0.65 (0.57 to 0.74) and 1.27% (0.90% to 1.61%), respectively, in people who received one or two doses of vaccine (n=84 620; 7989 nirmatrelvir and 76 631 no treatment); 0.64 (0.58 to 0.71) and 1.05% (0.85% to 1.27%) in individuals who received a booster dose of vaccine (n=94 905; 18 197 nirmatrelvir and 76 708 no treatment); 0.61 (0.57 to 0.65) and 1.36% (1.19% to 1.53%) in participants with a primary SARS-CoV-2 infection (n=228 081; 26 350 nirmatrelvir and 201 731 no treatment); and 0.74 (0.63 to 0.87) and 0.79% (0.36% to 1.18%) in participants who were reinfected with the SARS-CoV-2 virus (n=28 207; 5174 nirmatrelvir and 23 033 no treatment). Nirmatrelvir was associated with a reduced risk of admission to hospital or death in those aged ≤65 years and > 65 years; in men and women; in black and white participants; in those with 1-2, 3-4, and ≥5 risk factors for progression to severe covid-19 illness; and in those infected during the omicron BA.1 or BA.2 predominant era, and the BA.5 predominant era. CONCLUSIONS: In people with SARS-CoV-2 infection who were at risk of developing severe disease, compared with no treatment, nirmatrelvir was associated with a reduced risk of admission to hospital or death at 30 days in people who were not vaccinated, vaccinated, and had received a booster vaccine, and in those with a primary SARS-CoV-2 infection and reinfection.

Integrating machine learning with linguistic features: A universal method for extraction and normalization of temporal expressions in Chinese texts.

BACKGROUND AND OBJECTIVE: With the rapid development of information dissemination technology, the amount of events information contained in massive texts now far exceeds the intuitive cognition of humans, and it is hard to understand the progress of events in order of time. Temporal information runs through the whole process of beginning, proceeding, and ending of events, and plays an important role in many natural language processing applications, such as information extraction, question answering, and text summary. Accurately extracting temporal information from Chinese texts and automatically mapping the temporal expressions in natural language to the time axis are crucial to understanding the development of events and dynamic changes in them. METHODS: This study proposes a method integrating machine learning with linguistic features (IMLLF) for extraction and normalization of temporal expressions in Chinese texts to achieve the above objectives. Linguistic features are constructed by analyzing the expression rules of temporal information, and are combined with machine learning to map the natural language form of time onto a one-dimensional timeline. The web text dataset we build is divided into five parts for five-fold cross-validation, to compare the influence of different combinations of linguistic features and different methods. In the open medical dialog dataset, based on the training model obtained from the web text dataset, 200 disease descriptions are randomly selected each time for three rounds of experiments. RESULTS: The F1 of multi-feature fusion is 95.2%, which is better than the single-feature and double-feature combination. The results of experiments showed that the proposed IMLLF method can improve the accuracy of recognition of temporal information in Chinese to a greater extent than classical methods, with an F1-score of over 95% on the web text dataset and medical conversation dataset. In terms of the normalization of time expressions, the accuracy of the IMLLF method is higher than 93%. CONCLUSIONS: IMLLF has better results in extracting and normalizing time expressions on the web text dataset and the medical conversation dataset, which verifies the universality of IMLLF to identify and quantify temporal information. IMLLF method can accurately map the time information to the time axis, which is convenient for doctors to intuitively see when and what happened to the patient, and helps to make better medical decisions.