Summary Background Bacterial meningitis, which is caused mainly by Neisseria meningitidis, Haemophilus influenzae, and Streptococcus pneumoniae, inflicts a substantial burden of disease worldwide. Yet, the temporal dynamics of this disease are poorly characterised and many questions remain about the ecology of the disease. We aimed to comprehensively assess seasonal trends in bacterial meningitis on a global scale. Methods We developed the first bacterial meningitis global database by compiling monthly incidence data as reported by country-level surveillance systems. Using country-level wavelet analysis, we identified whether a 12 month periodic component (annual seasonality) was detected in time-series that had at least 5 years of data with at least 40 cases reported per year. We estimated the mean timing of disease activity by computing the centre of gravity of the distribution of cases and investigated whether synchrony exists between the three pathogens responsible for most cases of bacterial meningitis. Findings We used country-level data from 66 countries, including from 47 countries outside the meningitis belt in sub-Saharan Africa. A persistent seasonality was detected in 49 (96%) of the 51 time-series from 38 countries eligible for inclusion in the wavelet analyses. The mean timing of disease activity had a latitudinal trend, with bacterial meningitis seasons peaking during the winter months in countries in both the northern and southern hemispheres. The three pathogens shared similar seasonality, but time-shifts differed slightly by country. Interpretation Our findings provide key insight into the seasonal dynamics of bacterial meningitis and add to knowledge about the global epidemiology of meningitis and the host, environment, and pathogen characteristics driving these patterns. Comprehensive understanding of global seasonal trends in meningitis could be used to design more effective prevention and control strategies. Funding Princeton University Health Grand Challenge, US National Institutes of Health (NIH), NIH Fogarty International Center Research and Policy for Infectious Disease Dynamics programme (RAPIDD), Bill & Melinda Gates Foundation.

It is well documented that NLP models learn social biases, but little work has been done on how these biases manifest in model outputs for applied tasks like question answering (QA). We introduce the Bias Benchmark for QA (BBQ), a dataset of question sets constructed by the authors that highlight attested social biases against people belonging to protected classes along nine social dimensions relevant for U.S. English-speaking contexts. Our task evaluates model responses at two levels: (i) given an under-informative context, we test how strongly responses refect social biases, and (ii) given an adequately informative context, we test whether the model's biases override a correct answer choice. We fnd that models often rely on stereotypes when the context is under-informative, meaning the model's outputs consistently reproduce harmful biases in this setting. Though models are more accurate when the context provides an informative answer, they still rely on stereotypes and average up to 3.4 percentage points higher accuracy when the correct answer aligns with a social bias than when it conficts, with this difference widening to over 5 points on examples targeting gender for most models tested.

Structured information about entities is critical for many semantic parsing tasks. We present an approach that uses a Graph Neural Network (GNN) architecture to incorporate information about relevant entities and their relations during parsing. Combined with a decoder copy mechanism, this approach provides a conceptually simple mechanism to generate logical forms with entities. We demonstrate that this approach is competitive with the stateof-the-art across several tasks without pretraining, and outperforms existing approaches when combined with BERT pre-training.

No abstract

No abstract