Negation recognition in clinical natural language processing using a combination of the NegEx algorithm and a convolutional neural network.

BACKGROUND: Important clinical information of patients is present in unstructured free-text fields of Electronic Health Records (EHRs). While this information can be extracted using clinical Natural Language Processing (cNLP), the recognition of negation modifiers represents an important challenge. A wide range of cNLP applications have been developed to detect the negation of medical entities in clinical free-text, however, effective solutions for languages other than English are scarce. This study aimed at developing a solution for negation recognition in Spanish EHRs based on a combination of a customized rule-based NegEx layer and a convolutional neural network (CNN). METHODS: Based on our previous experience in real world evidence (RWE) studies using information embedded in EHRs, negation recognition was simplified into a binary problem ('affirmative' vs. 'non-affirmative' class). For the NegEx layer, negation rules were obtained from a publicly available Spanish corpus and enriched with custom ones, whereby the CNN binary classifier was trained on EHRs annotated for clinical named entities (cNEs) and negation markers by medical doctors. RESULTS: The proposed negation recognition pipeline obtained precision, recall, and F1-score of 0.93, 0.94, and 0.94 for the 'affirmative' class, and 0.86, 0.84, and 0.85 for the 'non-affirmative' class, respectively. To validate the generalization capabilities of our methodology, we applied the negation recognition pipeline on EHRs (6,710 cNEs) from a different data source distribution than the training corpus and obtained consistent performance metrics for the 'affirmative' and 'non-affirmative' class (0.95, 0.97, and 0.96; and 0.90, 0.83, and 0.86 for precision, recall, and F1-score, respectively). Lastly, we evaluated the pipeline against two publicly available Spanish negation corpora, the IULA and NUBes, obtaining state-of-the-art metrics (1.00, 0.99, and 0.99; and 1.00, 0.93, and 0.96 for precision, recall, and F1-score, respectively). CONCLUSION: Negation recognition is a source of low precision in the retrieval of cNEs from EHRs' free-text. Combining a customized rule-based NegEx layer with a CNN binary classifier outperformed many other current approaches. RWE studies highly benefit from the correct recognition of negation as it reduces false positive detections of cNE which otherwise would undoubtedly reduce the credibility of cNLP systems.

Earthworms and cadmium - Heavy metal resistant gut bacteria as indicators for heavy metal pollution in soils?

Preservation of the soil resources stability is of high importance for ecosystems, particularly in the current era of environmental change, which presents a severe pollution burden (e.g. by heavy metals) to soil fauna. Gut microbiomes are becoming recognized as important players in organism health, with comprehension of their perturbations in the polluted environment offering new insights into the nature and extent of heavy metal effects on the health of soil biota. Our aim was to investigate the effect of environmentally relevant heavy metal concentrations of cadmium (Cd) on the earthworm (Lumbricus terrestris) gut microbiota. Our results revealed that Cd exposure led to perturbations of earthworm gut microbiota with an increase in bacteria previously described as heavy metal resistant or able to bind heavy metals, revealing the potential of the earthworm-gut microbiota system in overcoming human-caused heavy metal pollution. Furthermore, an 'indicator species analysis' linked the bacterial genera Paenibacillus, Flavobacterium and Pseudomonas, with Cd treatment, suggesting these bacterial taxa as biomarkers of exposure in earthworms inhabiting Cd-stressed soils. The results of this study help to understand the impact of anthropogenic disturbance on soil fauna health and will have implications for environmental monitoring and protection of soil resources.

Gut microbiomes of free-ranging and captive Namibian cheetahs: Diversity, putative functions and occurrence of potential pathogens.

Although the significance of the gut microbiome for host health is well acknowledged, the impact of host traits and environmental factors on the interindividual variation of gut microbiomes of wildlife species is not well understood. Such information is essential; however, as changes in the composition of these microbial communities beyond the natural range might cause dysbiosis leading to increased susceptibility to infections. We examined the potential influence of sex, age, genetic relatedness, spatial tactics and the environment on the natural range of the gut microbiome diversity in free-ranging Namibian cheetahs (Acinonyx jubatus). We further explored the impact of an altered diet and frequent contact with roaming dogs and cats on the occurrence of potential bacterial pathogens by comparing free-ranging and captive individuals living under the same climatic conditions. Abundance patterns of particular bacterial genera differed between the sexes, and bacterial diversity and richness were higher in older (>3.5 years) than in younger individuals. In contrast, male spatial tactics, which probably influence host exposure to environmental bacteria, had no discernible effect on the gut microbiome. The profound resemblance of the gut microbiome of kin in contrast to nonkin suggests a predominant role of genetics in shaping bacterial community characteristics and functional similarities. We also detected various Operational Taxonomic Units (OTUs) assigned to potential pathogenic bacteria known to cause diseases in humans and wildlife species, such as Helicobacter spp., and Clostridium perfringens. Captive individuals did not differ in their microbial alpha diversity but exhibited higher abundances of OTUs related to potential pathogenic bacteria and shifts in disease-associated functional pathways. Our study emphasizes the need to integrate ecological, genetic and pathogenic aspects to improve our comprehension of the main drivers of natural variation and shifts in gut microbial communities possibly affecting host health. This knowledge is essential for in situ and ex situ conservation management.

Evaluation of Natural Language Processing for the Identification of Crohn Disease-Related Variables in Spanish Electronic Health Records: A Validation Study for the PREMONITION-CD Project.

BACKGROUND: The exploration of clinically relevant information in the free text of electronic health records (EHRs) holds the potential to positively impact clinical practice as well as knowledge regarding Crohn disease (CD), an inflammatory bowel disease that may affect any segment of the gastrointestinal tract. The EHRead technology, a clinical natural language processing (cNLP) system, was designed to detect and extract clinical information from narratives in the clinical notes contained in EHRs. OBJECTIVE: The aim of this study is to validate the performance of the EHRead technology in identifying information of patients with CD. METHODS: We used the EHRead technology to explore and extract CD-related clinical information from EHRs. To validate this tool, we compared the output of the EHRead technology with a manually curated gold standard to assess the quality of our cNLP system in detecting records containing any reference to CD and its related variables. RESULTS: The validation metrics for the main variable (CD) were a precision of 0.88, a recall of 0.98, and an F1 score of 0.93. Regarding the secondary variables, we obtained a precision of 0.91, a recall of 0.71, and an F1 score of 0.80 for CD flare, while for the variable vedolizumab (treatment), a precision, recall, and F1 score of 0.86, 0.94, and 0.90 were obtained, respectively. CONCLUSIONS: This evaluation demonstrates the ability of the EHRead technology to identify patients with CD and their related variables from the free text of EHRs. To the best of our knowledge, this study is the first to use a cNLP system for the identification of CD in EHRs written in Spanish.

Assessing the Performance of Clinical Natural Language Processing Systems: Development of an Evaluation Methodology.

BACKGROUND: Clinical natural language processing (cNLP) systems are of crucial importance due to their increasing capability in extracting clinically important information from free text contained in electronic health records (EHRs). The conversion of a nonstructured representation of a patient's clinical history into a structured format enables medical doctors to generate clinical knowledge at a level that was not possible before. Finally, the interpretation of the insights gained provided by cNLP systems has a great potential in driving decisions about clinical practice. However, carrying out robust evaluations of those cNLP systems is a complex task that is hindered by a lack of standard guidance on how to systematically approach them. OBJECTIVE: Our objective was to offer natural language processing (NLP) experts a methodology for the evaluation of cNLP systems to assist them in carrying out this task. By following the proposed phases, the robustness and representativeness of the performance metrics of their own cNLP systems can be assured. METHODS: The proposed evaluation methodology comprised five phases: (1) the definition of the target population, (2) the statistical document collection, (3) the design of the annotation guidelines and annotation project, (4) the external annotations, and (5) the cNLP system performance evaluation. We presented the application of all phases to evaluate the performance of a cNLP system called "EHRead Technology" (developed by Savana, an international medical company), applied in a study on patients with asthma. As part of the evaluation methodology, we introduced the Sample Size Calculator for Evaluations (SLiCE), a software tool that calculates the number of documents needed to achieve a statistically useful and resourceful gold standard. RESULTS: The application of the proposed evaluation methodology on a real use-case study of patients with asthma revealed the benefit of the different phases for cNLP system evaluations. By using SLiCE to adjust the number of documents needed, a meaningful and resourceful gold standard was created. In the presented use-case, using as little as 519 EHRs, it was possible to evaluate the performance of the cNLP system and obtain performance metrics for the primary variable within the expected CIs. CONCLUSIONS: We showed that our evaluation methodology can offer guidance to NLP experts on how to approach the evaluation of their cNLP systems. By following the five phases, NLP experts can assure the robustness of their evaluation and avoid unnecessary investment of human and financial resources. Besides the theoretical guidance, we offer SLiCE as an easy-to-use, open-source Python library.

Phylogeny- and Abundance-Based Metrics Allow for the Consistent Comparison of Core Gut Microbiome Diversity Indices Across Host Species.

The filtering of gut microbial datasets to retain high prevalence taxa is often performed to identify a common core gut microbiome that may be important for host biological functions. However, prevalence thresholds used to identify a common core are highly variable, and it remains unclear how they affect diversity estimates and whether insights stemming from core microbiomes are comparable across studies. We hypothesized that if macroecological patterns in gut microbiome prevalence and abundance are similar across host species, then we would expect that increasing prevalence thresholds would yield similar changes to alpha diversity and beta dissimilarity scores across host species datasets. We analyzed eight gut microbiome datasets based on 16S rRNA gene amplicon sequencing and collected from different host species to (1) compare macroecological patterns across datasets, including amplicon sequence variant (ASV) detection rate with sequencing depth and sample size, occupancy-abundance curves, and rank-abundance curves; (2) test whether increasing prevalence thresholds generate universal or host-species specific effects on alpha and beta diversity scores; and (3) test whether diversity scores from prevalence-filtered core communities correlate with unfiltered data. We found that gut microbiomes collected from diverse hosts demonstrated similar ASV detection rates with sequencing depth, yet required different sample sizes to sufficiently capture rare ASVs across the host population. This suggests that sample size rather than sequencing depth tends to limit the ability of studies to detect rare ASVs across the host population. Despite differences in the distribution and detection of rare ASVs, microbiomes exhibited similar occupancy-abundance and rank-abundance curves. Consequently, increasing prevalence thresholds generated remarkably similar trends in standardized alpha diversity and beta dissimilarity across species datasets until high thresholds above 70%. At this point, diversity scores tended to become unpredictable for some diversity measures. Moreover, high prevalence thresholds tended to generate diversity scores that correlated poorly with the original unfiltered data. Overall, we recommend that high prevalence thresholds over 70% are avoided, and promote the use of diversity measures that account for phylogeny and abundance (Balance-weighted phylogenetic diversity and Weighted Unifrac for alpha and beta diversity, respectively), because we show that these measures are insensitive to prevalence filtering and therefore allow for the consistent comparison of core gut microbiomes across studies without the need for prevalence filtering.

Offspring Microbiomes Differ Across Breeding Sites in a Panmictic Species.

High dispersal rates are known to homogenize host's population genetic structure in panmictic species and to disrupt host local adaptation to the environment. Long-distance dispersal might also spread micro-organisms across large geographical areas. However, so far, to which extent selection mechanisms that shape host's population genetics are mirrored in the population structure of the enteric microbiome remains unclear. High dispersal rates and horizontal parental transfer may homogenize bacterial communities between breeding sites (homogeneous hypothesis). Alternatively, strong selection from the local environment may differentiate bacterial communities between breeding sites (heterogeneous hypothesis). Furthermore, selection from age-specific environmental or physiological factors may differentiate the microbiome between juveniles and adults. Here, we analyzed the cloacal bacterial 16S rRNA gene of fledgling greater flamingos, Phoenicopterus roseus, across nine western Mediterranean breeding sites and four breeding seasons (n = 731) and adult birds (n = 27) from a single site. We found that fledgling cloacal microbiome, as measured by alpha diversity, beta diversity, the relative abundance of assigned sequence variants (ASVs) belonging to a phylum and genus composition within phylum, varied significantly between sampling sites and across time within site despite high adult dispersal rates. The spatio-temporal effects were stronger on individual ASV absence/presence than on ASV abundance (i.e., than on core microbiome composition). Spatial effects had a stronger effect than temporal effects, particularly on ASV abundance. Our study supports the heterogeneous hypothesis whereby local environmental conditions select and differentiate bacterial communities, thus countering the homogenizing effects of high-dispersing host species. In addition, differences in core microbiome between adult vs. fledgling samples suggests that differences in age-specific environmental and/or physiological factors result in differential selection pressure of core enteric microbiome between age classes, even within the same environment. In particular, the genus Corynebacterium, associated with both seasonal fat uptake and migration in previous studies, was much more abundant in high-dispersing fledglings than in more resident adults. To conclude, selection mechanisms that shape the host's genetic structure cannot be extended to the genetic structure of the enteric microbiome, which has important implications regarding our understanding of both host local adaptation mechanisms and enteric microbiome population genetics.

Home-Made Cost Effective Preservation Buffer Is a Better Alternative to Commercial Preservation Methods for Microbiome Research.

The investigation of wildlife gastrointestinal microbiomes by next-generation sequencing approaches is a growing field in microbial ecology and conservation. Such studies often face difficulties in sample preservation if neither freezing facilities nor liquid nitrogen (LQN) are readily available. Thus, in order to prevent microbial community changes because of bacterial growth after sampling, preservation buffers need to be applied to samples. However, the amount of microbial community variation attributable to the different preservation treatments and potentially affecting biological interpretation is hardly known. Here, we sampled feces of 11 sheep (Ovis aries sp.) by using swabs and analyzed the effect of air-drying, an inexpensive self-made nucleic acid preservation buffer (NAP), DNA/RNA Shield™, and RNAlater®, each together with freezing (for 10 days) or storing at room temperature (for 10 days) prior to 16S rRNA gene high-throughput sequencing to determine bacterial communities. Results revealed that the proportions of operational taxonomic units (OTUs) belonging to a bacterial phylum were affected by the preservation treatments, and that alpha diversities [observed OTUs, Shannon index, and phylogenetic diversity (PD)] were lower in all preservation treatments than in samples taken by forensic swabs and immediately frozen which is considered as the favored preservation treatment in the absence of any logistic constraints. Overall, NAP had better preservation qualities than RNAlater® and DNA/RNA Shield™ making this self-made buffer a valuable solution in wildlife microbiome studies.

Effects of host traits and land-use changes on the gut microbiota of the Namibian black-backed jackal (Canis mesomelas).

Host traits and environmental factors drive the natural variation in gut microbiota, and disruption in homeostasis can cause infections and chronic diseases. African wildlife is increasingly facing human-induced agricultural habitats, which also amplifies the contact probability with livestock with unknown consequences for wildlife gut microbiotas and the risk of transmission of potentially pathogenic bacteria. We applied high-throughput sequencing of bacterial 16S rRNA genes and microsatellite genotyping to investigate the impact of host traits and habitat use on the gut microbiotas of black-backed jackals (Canis mesomelas). This abundant carnivore inhabits livestock and game farms in central Namibia and is often persecuted as pathogen reservoir and vector. We further compared the gut microbiotas of black-backed jackals to other wild and domestic carnivores, herbivores and an omnivore, to disentangle the effects of environment, host species and dietary preference. In black-backed jackals, intrinsic host traits had a stronger impact in shaping the host-bacteria relationship than environmental factors. Nevertheless, the abundance of bacterial operational taxonomic units (OTUs) differed in individuals from livestock and game farms for specific bacterial genera such as Lactobacillus and Clostridium. We found, however, no evidence that black-backed jackals harbour abnormal levels of OTUs related to potential bacterial pathogens or that livestock farming has a negative impact on their health. We present here the first study investigating simultaneously the impact of host traits and environmental factors on gut microbiotas of a wildlife carnivore that occurs in a human-modified habitat.

Oligotyping reveals differences between gut microbiomes of free-ranging sympatric Namibian carnivores (Acinonyx jubatus, Canis mesomelas) on a bacterial species-like level.

Recent gut microbiome studies in model organisms emphasize the effects of intrinsic and extrinsic factors on the variation of the bacterial composition and its impact on the overall health status of the host. Species occurring in the same habitat might share a similar microbiome, especially if they overlap in ecological and behavioral traits. So far, the natural variation in microbiomes of free-ranging wildlife species has not been thoroughly investigated. The few existing studies exploring microbiomes through 16S rRNA gene reads clustered sequencing reads into operational taxonomic units (OTUs) based on a similarity threshold (e.g., 97%). This approach, in combination with the low resolution of target databases, generally limits the level of taxonomic assignments to the genus level. However, distinguishing natural variation of microbiomes in healthy individuals from "abnormal" microbial compositions that affect host health requires knowledge of the "normal" microbial flora at a high taxonomic resolution. This gap can now be addressed using the recently published oligotyping approach, which can resolve closely related organisms into distinct oligotypes by utilizing subtle nucleotide variation. Here, we used Illumina MiSeq to sequence amplicons generated from the V4 region of the 16S rRNA gene to investigate the gut microbiome of two free-ranging sympatric Namibian carnivore species, the cheetah (Acinonyx jubatus) and the black-backed jackal (Canis mesomelas). Bacterial phyla with proportions >0.2% were identical for both species and included Firmicutes, Fusobacteria, Bacteroidetes, Proteobacteria and Actinobacteria. At a finer taxonomic resolution, black-backed jackals exhibited 69 bacterial taxa with proportions ≥0.1%, whereas cheetahs had only 42. Finally, oligotyping revealed that shared bacterial taxa consisted of distinct oligotype profiles. Thus, in contrast to 3% OTUs, oligotyping can detect fine-scale taxonomic differences between microbiomes.

Detection of a related difference in efficacy of azadirachtin treatments for the control of whiteflies on Gerbera jamesonii by testing for interactions in generalised linear models.

BACKGROUND: Within a project cluster, systemical effects of two neem treatments (Neem Azal U and Neem Pellet) on the number of Trialeurodes vaporariorum (Westw.) in Gerbera jamesonii Flori Line Maxi were investigated. In order to minimise adverse ecological effects of a product, a new plant protection product can be declared as superior to a reference if it shows the same or a higher effect at a lower dosage. RESULTS: Based on a negative binomial model, the efficacy of the two treatments is compared by applying a user-defined interaction contrast in a multiple-contrast test framework. The corresponding calculation of simultaneous confidence intervals for the ratio of model parameters provides conclusions about statistical significance and biological relevance. The validity of the approach is shown by a small simulation study. As a result of the experiment, besides a significant difference between two negative control treatments, no significant difference between the efficiency of the two neem treatments (Neem Azal U and Neem Pellet) could be detected. CONCLUSIONS: Simultaneous confidence intervals for linear combinations of parameters of generalised linear models provide a useful way to compare the efficiency of two plant protection products. The results for the azadirachtin experiment indicate that large differences between any application effects of either Neem Azal U or Neem Pellet cannot be expected.