Explainable detection of adverse drug reaction with imbalanced data distribution

Analysis of health-related texts can be used to detect adverse drug reactions (ADR). The greatest challenge for ADR detection lies in imbalanced data distributions where words related to ADR symptoms are often minority classes. As a result, trained models tend to converge to a point that strongly biases towards the majority class and then ignores the minority class. Since the most used cross-entropy criteria is an approximation to accuracy, the model focuses more readily on the majority class to achieve high accuracy. To address this issue, existing methods apply either oversampling or down-sampling strategies to balance the data distribution and exploit the most difficult samples of the minority class. However, increasing or reducing the number of individual tokens alone in sequence labeling tasks will result in the loss of the syntactic relations of the sentence. This paper proposes a weighted variant of conditional random field (CRF) for data-imbalanced sequence labeling tasks. Such a weighting strategy can alleviate data distribution imbalances between majority and minority classes. Instead of using softmax in the output layer, the CRF can capture the relationship of labels between tokens. The locally interpretable model-agnostic explanations (LIME) algorithm was applied to investigate performance differences between models with and without the weighted loss function. Experimental results on two different ADR tasks show that the proposed model outperforms previously proposed sequence labeling methods.     

Assessing inter-rater reliability when the raters are fixed: Two concepts and two estimates

Intraclass correlation (ICC) is an established tool to assess inter-rater reliability. In a seminal paper published in 1979, Shrout and Fleiss considered three statistical models for inter-rater reliability data with a balanced design. In their first two models, an infinite population of raters was considered, whereas in their third model, the raters in the sample were considered to be the whole population of raters. In the present paper, we show that the two distinct estimates of ICC developed for the first two models can both be applied to the third model and we discuss their different interpretations in this context.     

Intra- and inter-rater reliability of electrical impedance myography using adhesive electrodes in healthy volunteers

In spite of the growing use of the electrical impedance myography (EIM) measures for clinical assessment and follow-up of diseased muscle tissue, reliability studies are scarce. We evaluate the reliability of the (EIM) technique using four adhesive electrodes over the muscle of interest.Intra- and inter-rater reliability was studied within the same session and between sessions. Thirty-one healthy and volunteer subjects aged between 20 and 26 years were recruited. Phase angle, reactance and resistance were assessed for each EIM measurement. Intraclass correlation coefficient (ICC) was used to determine the relative reliability. Absolute reliability was expressed as the standard error of measurement and the minimum detectable change.Relative reliability within the same session and between sessions for the EIM technique was excellent (ICCs > 0.9) concerning both intra- and inter-rater reliability, except for the component reactance. The absolute reliability was very high for the three EIM components.EIM measures using four adhesive electrodes over the area of interest is a reliable technique to assess muscle tissue status. This study confirms that these measurement results barely vary depending on the examiner and the moment. The present study also confirms phase angle as the least affected EIM component by examiner and evaluation moment.     

Percentage of patients with preventable adverse drug reactions and preventability of adverse drug reactions--a meta-analysis

Background

Numerous observational studies suggest that preventable adverse drug reactions are a significant burden in healthcare, but no meta-analysis using a standardised definition for adverse drug reactions exists. The aim of the study was to estimate the percentage of patients with preventable adverse drug reactions and the preventability of adverse drug reactions in adult outpatients and inpatients.

Methods

Studies were identified through searching Cochrane, CINAHL, EMBASE, IPA, Medline, PsycINFO and Web of Science in September 2010, and by hand searching the reference lists of identified papers. Original peer-reviewed research articles in English that defined adverse drug reactions according to WHO’s or similar definition and assessed preventability were included. Disease or treatment specific studies were excluded. Meta-analysis on the percentage of patients with preventable adverse drug reactions and the preventability of adverse drug reactions was conducted.

Results

Data were analysed from 16 original studies on outpatients with 48797 emergency visits or hospital admissions and from 8 studies involving 24128 inpatients. No studies in primary care were identified. Among adult outpatients, 2.0% (95% confidence interval (CI): 1.2–3.2%) had preventable adverse drug reactions and 52% (95% CI: 42–62%) of adverse drug reactions were preventable. Among inpatients, 1.6% (95% CI: 0.1–51%) had preventable adverse drug reactions and 45% (95% CI: 33–58%) of adverse drug reactions were preventable.

Conclusions

This meta-analysis corroborates that preventable adverse drug reactions are a significant burden to healthcare among adult outpatients. Among both outpatients and inpatients, approximately half of adverse drug reactions are preventable, demonstrating that further evidence on prevention strategies is required. The percentage of patients with preventable adverse drug reactions among inpatients and in primary care is largely unknown and should be investigated in future research.     

Perceptions of doctors to adverse drug reaction reporting in a teaching hospital in Lagos, Nigeria

Background  

Spontaneous adverse drug reaction (ADR) reporting is the cornerstone of pharmacovigilance. ADR reporting with Yellow Cards has tremendously improved pharmacovigilance of drugs in many developed countries and its use is advocated by the World Health Organization (WHO). This study was aimed at investigating the knowledge and attitude of doctors in a teaching hospital in Lagos, Nigeria on spontaneous ADR reporting and to suggest possible ways of improving this method of reporting.     

Recognizing and reporting adverse drug reactions.

Although physicians in practice are most likely to see patients with adverse drug reactions, they may fail to recognize an adverse effect or to attribute it to a drug effect and, when recognized, they may fail to report serious reactions to the US Food and Drug Administration (FDA). To recognize and attribute an adverse event to a drug effect, physicians should review the patient''s clinical course, looking at patient risk factors, the known adverse reactions to the suspected drug, and the likelihood of a causal relationship between the drug and the adverse event-based on the temporal relationship, response to stopping or restarting the drug, and whether other factors could explain the reaction. Once an adverse drug reaction has been identified, the patient should be informed and appropriate documentation made in the patient''s medical record. Serious known reactions and all reactions to newly released drugs or those not previously known to occur (even if the certainty is low) should be reported to the FDA.