Mapping influenza activity in emergency departments in France using Bayesian model-based geostatistics

Pattern of encounters to emergency departments for suicidal attempts in France: Identification of high-risk days, months and holiday periods

INTRODUCTION

Seasonal change in patterns of suicidal attempts is not well known in France and may differ from other western countries. We aimed to determine the peak times (days, months and holiday periods) of suicidal attempts in France.

METHODS

We carried out a multicentre retrospective epidemiological study, using data from the Organization for Coordinated Monitoring of Emergencies (OSCOUR®) network. We aggregated daily data from January 1, 2010, to December 31, 2019. Variations in suicidal attempts on specific days were investigated by comparing their frequencies (ad hoc Z-scores).

RESULTS

114,805,488 ED encounters were recorded including 233,242 ED encounters regarding suicidal attempts. Men accounted for 45.7%. A significantly higher frequency of ED encounters for suicidal acts were found on Sundays in the months of May-June for both sexes and on New Year's Day for all genders and age groups. An increased risk was also noted on July 14th (National Day) and June 22nd (Summer Solstice). A protective effect was noted on the day after Valentine's Day, on Christmas Day and Christmas time (in particular December 24 and 26).

CONCLUSION

Sundays, June, New Year's Day were at increased risk of suicidal attempts in France requiring a strengthening of prevention.

Deep Learning Transformer Models for Building a Comprehensive and Real-time Trauma Observatory: Development and Validation Study

BACKGROUND

Public health surveillance relies on the collection of data, often in near-real time. Recent advances in natural language processing make it possible to envisage an automated system for extracting information from electronic health records.

OBJECTIVE

To study the feasibility of setting up a national trauma observatory in France, we compared the performance of several automatic language processing methods in a multiclass classification task of unstructured clinical notes.

METHODS

A total of 69,110 free-text clinical notes related to visits to the emergency departments of the University Hospital of Bordeaux, France, between 2012 and 2019 were manually annotated. Among these clinical notes, 32.5% (22,481/69,110) were traumas. We trained 4 transformer models (deep learning models that encompass attention mechanism) and compared them with the term frequency-inverse document frequency associated with the support vector machine method.

RESULTS

The transformer models consistently performed better than the term frequency-inverse document frequency and a support vector machine. Among the transformers, the GPTanam model pretrained with a French corpus with an additional autosupervised learning step on 306,368 unlabeled clinical notes showed the best performance with a micro F1-score of 0.969.

CONCLUSIONS

The transformers proved efficient at the multiclass classification of narrative and medical data. Further steps for improvement should focus on the expansion of abbreviations and multioutput multiclass classification.

How Spatial Epidemiology Helps Understand Infectious Human Disease Transmission

Both directly and indirectly transmitted infectious diseases in humans are spatial-related. Spatial dimensions include: distances between susceptible humans and the environments shared by people, contaminated materials, and infectious animal species. Therefore, spatial concepts in managing and understanding emerging infectious diseases are crucial. Recently, due to the improvements in computing performance and statistical approaches, there are new possibilities regarding the visualization and analysis of disease spatial data. This review provides commonly used spatial or spatial-temporal approaches in managing infectious diseases. It covers four sections, namely: visualization, overall clustering, hot spot detection, and risk factor identification. The first three sections provide methods and epidemiological applications for both point data (i.e., individual data) and aggregate data (i.e., summaries of individual points). The last section focuses on the spatial regression methods adjusted for neighbour effects or spatial heterogeneity and their implementation. Understanding spatial-temporal variations in the spread of infectious diseases have three positive impacts on the management of diseases. These are: surveillance system improvements, the generation of hypotheses and approvals, and the establishment of prevention and control strategies. Notably, ethics and data quality have to be considered before applying spatial-temporal methods. Developing differential global positioning system methods and optimizing Bayesian estimations are future directions.

The epidemiology of hepatitis delta virus infection in Cameroon

OBJECTIVE

To investigate the distribution and risk factors of hepatitis delta virus (HDV) infection in Cameroon.

DESIGN

We tested for hepatitis B virus (HBV) surface antigen (HBsAg) and anti-HDV antibody 14 150 samples collected during a survey whose participants were representative of the Cameroonian adult population. The samples had already been tested for hepatitis C virus and HIV antibodies.

RESULTS

Overall, 1621/14 150 (weighted prevalence=11.9%) participants were HBsAg positive, among whom 224/1621 (10.6%) were anti-HDV positive. In 2011, the estimated numbers of HBsAg positive and HDV seropositives were 1 160 799 and 122 910 in the 15-49 years age group, respectively. There were substantial regional variations in prevalence of chronic HBV infection, but even more so for HDV (from 1% to 54%). In multivariable analysis, HDV seropositivity was independently associated with living with an HDV-seropositive person (OR=8.80; 95% CI: 3.23 to 24.0), being HIV infected (OR=2.82; 95% CI: 1.32 to 6.02) and living in the South (latitude <4°N) while having rural/outdoor work (OR=15.2; 95% CI: 8.35 to 27.6, when compared with living on latitude ≥4°N and not having rural/outdoor work).

CONCLUSION

We found evidence for effective intra-household transmission of HDV in Cameroon. We also identified large differences in prevalence between regions, with cases concentrated in forested areas close to the Equator, as described in other tropical areas. The reasons underlying these geographical variations in HDV prevalence deserve further investigation.

Mapping influenza activity in emergency departments in France using Bayesian model-based geostatistics

BACKGROUND

Maps of influenza activity are important tools to monitor influenza epidemics and inform policymakers. In France, the availability of a high-quality data set from the Oscour^® surveillance network, covering 92% of hospital emergency department (ED) visits, offers new opportunities for disease mapping. Traditional geostatistical mapping methods such as Kriging ignore underlying population sizes, are not suited to non-Gaussian data and do not account for uncertainty in parameter estimates.

OBJECTIVE

Our objective was to create reliable weekly interpolated maps of influenza activity in the ED setting, to inform Santé publique France (the French national public health agency) and local healthcare authorities.

METHODS

We used Oscour^® data of ED visits covering the 2016-2017 influenza season. We developed a Bayesian model-based geostatistical approach, a class of generalized linear mixed models, with a multivariate normal random field as a spatially autocorrelated random effect. Using R-INLA, we developed an algorithm to create maps of the proportion of influenza-coded cases among all coded visits. We compared our results with maps obtained by Kriging.

RESULTS

Over the study period, 45 565 (0.82%) visits were coded as influenza cases. Maps resulting from the model are presented for each week, displaying the posterior mean of the influenza proportion and its associated uncertainty. Our model performed better than Kriging.

CONCLUSIONS

Our model allows producing smoothed maps where the random noise has been properly removed to reveal the spatial risk surface. The algorithm was incorporated into the national surveillance system to produce maps in real time and could be applied to other diseases.