AE-GPT: Using Large Language Models to extract adverse events from surveillance reports-A use case with influenza vaccine adverse events.
Ontology highlight
ABSTRACT:
SUBMITTER: Li Y
PROVIDER: S-EPMC10956752 | biostudies-literature | 2024
REPOSITORIES: biostudies-literature
Publications
AE-GPT: Using Large Language Models to extract adverse events from surveillance reports-A use case with influenza vaccine adverse events.
Li Yiming Y Li Jianfu J He Jianping J Tao Cui C
PloS one 20240321 3
Though Vaccines are instrumental in global health, mitigating infectious diseases and pandemic outbreaks, they can occasionally lead to adverse events (AEs). Recently, Large Language Models (LLMs) have shown promise in effectively identifying and cataloging AEs within clinical reports. Utilizing data from the Vaccine Adverse Event Reporting System (VAERS) from 1990 to 2016, this study particularly focuses on AEs to evaluate LLMs' capability for AE extraction. A variety of prevalent LLMs, includi ...[more]
Similar Datasets
Project description:ImportanceAntivaccine sentiment is increasingly associated with conservative political positions. Republican-inclined states exhibit lower COVID-19 vaccination rates, but the association between political inclination and reported vaccine adverse events (AEs) is unexplored.ObjectiveTo assess whether there is an association between state political inclination and the reporting rates of COVID-19 vaccine AEs.Design, setting, and participantsThis cross-sectional study used the AE reports after COVID-19 vaccination from the Vaccine Adverse Event Reporting System (VAERS) database from 2020 to 2022, with reports after influenza vaccines from 2019 to 2022 used as a reference. These reports were examined against state-level percentage of Republican votes in the 2020 US presidential election.ExposureState-level percentage of Republican votes in the 2020 US presidential election.Main outcomes and measuresRates of any AE among COVID-19 vaccine recipients, rates of any severe AE among vaccine recipients, and the proportion of AEs reported as severe.ResultsA total of 620 456 AE reports (mean [SD] age of vaccine recipients, 51.8 [17.6] years; 435 797 reports from women [70.2%]; a vaccine recipient could potentially file more than 1 report, so reports are not necessarily from unique individuals) for COVID-19 vaccination were identified from the VAERS database. Significant associations between state political inclination and state AE reporting were observed for all 3 outcomes: a 10% increase in Republican voting was associated with increased odds of AE reports (odds ratio [OR], 1.05; 95% CI, 1.05-1.05; P < .001), severe AE reports (OR, 1.25; 95% CI, 1.24-1.26; P < .001), and the proportion of AEs reported as severe (OR, 1.21; 95% CI, 1.20-1.22; P < .001). These associations were seen across all age strata in stratified analyses and were more pronounced among older subpopulations.Conclusions and relevanceThis cross-sectional study found that the more states were inclined to vote Republican, the more likely their vaccine recipients or their clinicians reported COVID-19 vaccine AEs. These results suggest that either the perception of vaccine AEs or the motivation to report them was associated with political inclination.
| S-EPMC10980960 | biostudies-literature
Project description:Post licensure surveillance of adverse events following immunization (AEFI) is a fundamental activity to improve safety and maintain public confidence in vaccines. Since 2011, the Liguria Region has been involved in the inter-regional project of post-marketing surveillance of AEFI, coordinated by the Italian Medicine Agency and the Veneto region. The main objectives of the project are: (1) to coordinate the surveillance activities in the 8 Italian Regions included in the project; (2) to encourage the signal of AEFI by healthcare workers and patients; (3) to organize education activities addressed to health care workers, and, finally; (4) to establish vaccination counseling services in each Region. In particular, the Ligurian multidisciplinary team, composed by physicians expert in the field of vaccination and pharmacists, is involved in the causality assessment between vaccines and all adverse events signaled within the Liguria Region and in the analysis of all adverse events signaled in Italy as possibly related to influenza vaccines. During 2013, the team has organized 4 courses, addressed to healthcare personnel of vaccination outpatient clinics, focused on European and Italian legislation on pharmaco-vigilance and vaccine-vigilance and aimed at promoting signal of AEFI. Since October 2013, the Liguria Region has been participating to the inter-regional project of active surveillance of adverse events aimed at promoting the signal of AEFI by parents of vaccinated infants. After two years of implementation of the project both the number of reported AEFI and the reporting rate per 100 000 administered doses of vaccine increased. The activities need to be consolidated in the next years in order to guarantee high standard of vaccine safety, maintain the confidence in current immunization programs and reach optimal vaccination coverage rate.
| S-EPMC4514437 | biostudies-other
Project description:ObjectiveAutomated analysis of vaccine postmarketing surveillance narrative reports is important to understand the progression of rare but severe vaccine adverse events (AEs). This study implemented and evaluated state-of-the-art deep learning algorithms for named entity recognition to extract nervous system disorder-related events from vaccine safety reports.Materials and methodsWe collected Guillain-Barré syndrome (GBS) related influenza vaccine safety reports from the Vaccine Adverse Event Reporting System (VAERS) from 1990 to 2016. VAERS reports were selected and manually annotated with major entities related to nervous system disorders, including, investigation, nervous_AE, other_AE, procedure, social_circumstance, and temporal_expression. A variety of conventional machine learning and deep learning algorithms were then evaluated for the extraction of the above entities. We further pretrained domain-specific BERT (Bidirectional Encoder Representations from Transformers) using VAERS reports (VAERS BERT) and compared its performance with existing models.Results and conclusionsNinety-one VAERS reports were annotated, resulting in 2512 entities. The corpus was made publicly available to promote community efforts on vaccine AEs identification. Deep learning-based methods (eg, bi-long short-term memory and BERT models) outperformed conventional machine learning-based methods (ie, conditional random fields with extensive features). The BioBERT large model achieved the highest exact match F-1 scores on nervous_AE, procedure, social_circumstance, and temporal_expression; while VAERS BERT large models achieved the highest exact match F-1 scores on investigation and other_AE. An ensemble of these 2 models achieved the highest exact match microaveraged F-1 score at 0.6802 and the second highest lenient match microaveraged F-1 score at 0.8078 among peer models.
| S-EPMC8279785 | biostudies-literature
Project description:Both cancer patients and the elderly are at high risk of developing flu complications, so influenza vaccination is recommended. We aimed to evaluate potential adverse events (AEs) following influenza vaccination in elderly cancer patients using the self-controlled tree-temporal scan statistic method. From a large linked database of Korea Disease Control and Prevention Agency vaccination data and the National Health Insurance Service claims data, we identified cancer patients aged over 65 who received flu vaccines during the 2016/2017 and 2017/2018 seasons. We included all the outcomes occurring on 1-84 days post-vaccination and evaluated all temporal risk windows, which started 1-28 days and ended 2-42 days. Patients who were diagnosed with the same disease during a year prior to vaccination were excluded. We used the hierarchy of ICD-10 to identify statistically significant clustering. This study included 431,276 doses of flu vaccine. We detected signals for 1 set: other dorsopathies on 1-15 days (attributable risk 16.5 per 100,000, P = 0.017). Dorsopathy is a known AE of influenza vaccine. No statistically significant clusters were found when analyzed by flu season. Therefore, influenza vaccination is more recommended for elderly patients with cancer and weakened immune systems.
| S-EPMC10432531 | biostudies-literature
Project description:Since the beginning of the COVID-19 pandemic, vaccines have been developed to mitigate the spread of SARS-CoV-2, the virus that causes COVID-19. These vaccines have been effective in reducing the rate and severity of COVID-19 infection but also have been associated with various adverse events (AEs). In this study, data from the Vaccine Adverse Event Reporting System (VAERS) was queried and analyzed via the Cov19VaxKB vaccine safety statistical analysis tool to identify statistically significant (i.e., enriched) AEs for the three currently FDA-authorized or approved COVID-19 vaccines. An ontology-based classification and literature review were conducted for these enriched AEs. Using VAERS data as of 31 December 2021, 96 AEs were found to be statistically significantly associated with the Pfizer-BioNTech, Moderna, and/or Janssen COVID-19 vaccines. The Janssen COVID-19 vaccine had a higher crude reporting rate of AEs compared to the Moderna and Pfizer COVID-19 vaccines. Females appeared to have a higher case report frequency for top adverse events compared to males. Using the Ontology of Adverse Event (OAE), these 96 adverse events were classified to different categories such as behavioral and neurological AEs, cardiovascular AEs, female reproductive system AEs, and immune system AEs. Further statistical comparison between different ages, doses, and sexes was also performed for three notable AEs: myocarditis, GBS, and thrombosis. The Pfizer vaccine was found to have a closer association with myocarditis than the other two COVID-19 vaccines in VAERS, while the Janssen vaccine was more likely to be associated with thrombosis and GBS AEs. To support standard AE representation and study, we have also modeled and classified the newly identified thrombosis with thrombocytopenia syndrome (TTS) AE and its subclasses in the OAE by incorporating the Brighton Collaboration definition. Notably, severe COVID-19 vaccine AEs (including myocarditis, GBS, and TTS) rarely occur in comparison to the large number of COVID-19 vaccinations administered in the United States, affirming the overall safety of these COVID-19 vaccines.
| S-EPMC9263450 | biostudies-literature
Project description:BackgroundInfluenza vaccination recently has been recommended for children 6-23 months old, but is not currently recommended for routine use in non-high-risk older children. Information on disease impact, costs, benefits, risks, and community preferences could help guide decisions about which age and risk groups should be vaccinated and strategies for improving coverage. The objective of this study was to measure preferences and willingness-to-pay for changes in health-related quality of life associated with uncomplicated influenza and two rarely-occurring vaccination-related adverse events (anaphylaxis and Guillain-Barré syndrome) in children.MethodsWe conducted telephone interviews with adult members selected at random from a large New England HMO (n = 112). Respondents were given descriptions of four health outcomes: uncomplicated influenza in a hypothetical 1-year-old child of their own, uncomplicated influenza in a hypothetical 14-year-old child of their own, anaphylaxis following vaccination, and Guillain-Barré syndrome. "Uncomplicated influenza" did not require a physician's visit or hospitalization. Preferences (values) for these health outcomes were measured using time-tradeoff and willingness-to-pay questions. Time-tradeoff questions asked the adult to assume they had a child and to consider how much time from the end of their own life they would be willing to surrender to avoid the health outcome in the child.ResultsRespondents said they would give a median of zero days of their lives to prevent an episode of uncomplicated influenza in either their (hypothetical) 1-year-old or 14-year-old, 30 days to prevent an episode of vaccination-related anaphylaxis, and 3 years to prevent a vaccination-related case of Guillain-Barré syndrome. Median willingness-to-pay to prevent uncomplicated influenza in a 1-year-old was $175, uncomplicated influenza in a 14-year-old was $100, anaphylaxis $400, and Guillain-Barré syndrome $4000. The median willingness-to-pay for an influenza vaccination for their children with no risk of anaphylaxis or Guillain-Barré syndrome was $50 and $100, respectively.ConclusionMost respondents said they would not be willing to trade any time from their own lives to prevent uncomplicated influenza in a child of their own, and the time traded did not vary by the age of the hypothetical affected child. However, adults did indicate a willingness-to-pay to prevent uncomplicated influenza in children, and that they would give more money to prevent the illness in a 1-year-old than in a 14-year-old. Respondents also indicated a willingness to pay a premium for a vaccine without any risk of severe complications.
| S-EPMC1083419 | biostudies-literature
Project description:Chloroquine (CQ) and hydroxychloroquine (HCQ) are on the World Health Organization's List of Essential Medications for treating non-resistant malaria, rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE). In addition, both drugs are currently used off-label in hospitals worldwide and in numerous clinical trials for the treatment of SARS-CoV-2 infection. However, CQ and HCQ use has been associated with cardiac side effects, which is of concern due to the higher risk of COVID-19 complications in patients with heart related disorders, and increased mortality associated with COVID-19 cardiac complications. In this study we analyzed over thirteen million adverse event reports form the United States Food and Drug Administration Adverse Event Reporting System to confirm and quantify the association of cardiac side effects of CQ and HCQ. Additionally, we identified several confounding factors, including male sex, NSAID coadministration, advanced age, and prior diagnoses contributing to drug related cardiotoxicity. These findings may help guide therapeutic decision making and ethical trial design for COVID-19 treatment.
| S-EPMC7644696 | biostudies-literature
Project description:BackgroundWhen increased rates of adverse events following immunization are detected, regulatory action can be taken by public health agencies. However to be interpreted reports of adverse events must be encoded in a consistent way. Regulatory agencies rely on guidelines to help determine the diagnosis of the adverse events. Manual application of these guidelines is expensive, time consuming, and open to logical errors. Representing these guidelines in a format amenable to automated processing can make this process more efficient.Methods and findingsUsing the Brighton anaphylaxis case definition, we show that existing clinical guidelines used as standards in pharmacovigilance can be logically encoded using a formal representation such as the Adverse Event Reporting Ontology we developed. We validated the classification of vaccine adverse event reports using the ontology against existing rule-based systems and a manually curated subset of the Vaccine Adverse Event Reporting System. However, we encountered a number of critical issues in the formulation and application of the clinical guidelines. We report these issues and the steps being taken to address them in current surveillance systems, and in the terminological standards in use.ConclusionsBy standardizing and improving the reporting process, we were able to automate diagnosis confirmation. By allowing medical experts to prioritize reports such a system can accelerate the identification of adverse reactions to vaccines and the response of regulatory agencies. This approach of combining ontology and semantic technologies can be used to improve other areas of vaccine adverse event reports analysis and should inform both the design of clinical guidelines and how they are used in the future.AvailabilitySufficient material to reproduce our results is available, including documentation, ontology, code and datasets, at http://purl.obolibrary.org/obo/aero.
| S-EPMC3965435 | biostudies-literature
Project description:The Vaccine Adverse Event Reporting System (VAERS) is a passive reporting system, used for monitoring the safety of all US licensed vaccines. In March 2008, ACAM2000(®) replaced Dryvax(®) as the only licensed smallpox vaccine and is administered to all persons entering military service and certain civilian researchers. In 2011, routine data mining of VAERS identified a vaccine safety concern resulting in acute ischemic cardiac events (ICE) following ACAM2000(®).During March 1, 2008 through June 30, 2013, we reviewed all serious reports received following ACAM2000(®)and classified them by diagnostic category. We identified possible ICE cases by searching the Medical Dictionary for Regulatory Affairs (MedDRA(®)) terms for "myocardial ischaemia," "acute myocardial infarction," "myocardial infarction," and "ischaemia," and applied standardized surveillance case definitions.VAERS received 1149 reports following ACAM2000(®) administration; 169 (14.7%) were serious (resulting in permanent disability, hospitalization or prolongation of hospitalization, life-threatening illness or death), including one death. The two most frequent diagnostic categories for serious reports were cardiovascular and other infectious conditions. The MedDRA(®) search found 31 reports of possible ICE after receipt of ACAM2000(®) vaccine. Of a total 30 possible ICE cases with demographic information, all but one was male; the age range was 20-45 years (median 32) and median interval to onset of symptoms was 12 days. On clinical review there were 16 cases of myocarditis/pericarditis and 15 ICE cases.Our review of the data mining signal did not substantiate the concerns about ICE after ACAM2000(®). Our study also suggests that with current pre-vaccination screening, cardiac morbidity in generally healthy vaccinated populations remains uncommon.
| S-EPMC5946322 | biostudies-other
Project description:Increasing evidence has shown that sex differences exist in Adverse Drug Events (ADEs). Identifying those sex differences in ADEs could reduce the experience of ADEs for patients and could be conducive to the development of personalized medicine. In this study, we analyzed a normalized US Food and Drug Administration Adverse Event Reporting System (FAERS). Chi-squared test was conducted to discover which treatment regimens or drugs had sex differences in adverse events. Moreover, reporting odds ratio (ROR) and P value were calculated to quantify the signals of sex differences for specific drug-event combinations. Logistic regression was applied to remove the confounding effect from the baseline sex difference of the events. We detected among 668 drugs of the most frequent 20 treatment regimens in the United States, 307 drugs have sex differences in ADEs. In addition, we identified 736 unique drug-event combinations with significant sex differences. After removing the confounding effect from the baseline sex difference of the events, there are 266 combinations remained. Drug labels or previous studies verified some of them while others warrant further investigation.
| S-EPMC4840306 | biostudies-literature