Conditional Probability Joint Extraction of Nested Biomedical Events: Design of a Unified Extraction Framework Based on Neural Networks

Abstract

Background Event extraction is essential for natural language processing. In the biomedical field, the nested event phenomenon (event A as a participating role of event B) makes extracting this event more difficult than extracting a single event. Therefore, the performance of nested biomedical events is always underwhelming. In addition, previous works relied on a pipeline to build an event extraction model, which ignored the dependence between trigger recognition and event argument detection tasks and produced significant cascading errors. Objective This study aims to design a unified framework to jointly train biomedical event triggers and arguments and improve the performance of extracting nested biomedical events. Methods We proposed an end-to-end joint extraction model that considers the probability distribution of triggers to alleviate cascading errors. Moreover, we integrated the syntactic structure into an attention-based gate graph convolutional network to capture potential interrelations between triggers and related entities, which improved the performance of extracting nested biomedical events. Results The experimental results demonstrated that our proposed method achieved the best F1 score on the multilevel event extraction biomedical event extraction corpus and achieved a favorable performance on the biomedical natural language processing shared task 2011 Genia event corpus. Conclusions Our conditional probability joint extraction model is good at extracting nested biomedical events because of the joint extraction mechanism and the syntax graph structure. Moreover, as our model did not rely on external knowledge and specific feature engineering, it had a particular generalization performance.

Child-Sum EATree-LSTMs: enhanced attentive Child-Sum Tree-LSTMs for biomedical event extraction

“… He et al [ 30 ] proposed a fine-grained method with multilevel attention and sentence embeddings. Wang et al [ 31 ] presented an end-to-end model that uses the probability distribution of triggers and the syntactic structure in an attention-based gate GCN. Ahmed et al [ 21 ] encoded decomposable attention framework and the soft attention mechanism inside a Tree-LSTM cell on semantic relatedness tasks.…”

“…Wang et al [ 31 ] presented an end-to-end model that uses the probability distribution of triggers and the syntactic structure in an attention-based gate GCN.…”

Child-Sum (N2e2n)Tree-Lstms: An Interactive Child-Sum Tree-Lstms to Extract Biomedical Event

“…(1) Comparision with other models Fei et al [34] combined RvNNs and CRF to detect the biomedical event trigger. Wang et al [35] computed the event triggers' probability distribution to solve the cascading errors, and incorporated the semantics structure into a gate-based GCN with an attention mechanism to catch the potential semantic associations for the complex biomedical events. He et al [36] decomposed the trigger detection into two steps: detection step and classification step.…”

Use of Machine Learning Tools in Evidence Synthesis of Tobacco Use Among Sexual and Gender Diverse Populations: Algorithm Development and Validation

“…Notably, sexual and gender diverse individuals, often referred to as the LGBTQ+ (lesbian, gay, bisexual, transgender, queer, intersex, asexual, Two Spirit, and other persons who identify as part of this community) populations, are particularly vulnerable to nicotine and tobacco product use [ 3 ]. Both the National Cancer Institute and the Centers for Disease Control and Prevention have recognized the LGBTQ+ populations as a critical target in their efforts to combat tobacco use disparities [ 4 - 10 ].…”

“…NLP is dedicated to deciphering and comprehending how computers interpret human language, equipping them to analyze extensive data sets of natural language [ 14 - 16 ]. While NLP tools have garnered considerable recognition in biomedical research [ 4 - 10 ], aiding in tasks such as disease surveillance (eg, COVID-19) and diagnosing using medical records [ 17 - 23 ], their potential to expedite near real-time synthesis of evidence in tobacco control research remains untapped [ 24 ].…”