Search

As two important textual modalities in electronic health records (EHR), both structured data (clinical codes) and unstructured data (clinical narratives) have recently been increasingly applied to the healthcare domain. Most existing EHR-oriented studies, however, either focus on a particular modality or integrate data from different modalities in a straightforward manner, which usually treats structured and unstructured data as two independent sources of information about patient admission and ignore the intrinsic interactions between them. In fact, the two modalities are documented during the same encounter where structured data inform the documentation of unstructured data and vice versa. In this paper, we proposed a Medical Multimodal Pre-trained Language Model, named MedM-PLM, to learn enhanced EHR representations over structured and unstructured data and explore the interaction of two modalities. In MedM-PLM, two Transformer-based neural network components are firstly adopted to learn representative characteristics from each modality. A cross-modal module is then introduced to model their interactions. We pre-trained MedM-PLM on the MIMIC-III dataset and verified the effectiveness of the model on three downstream clinical tasks, i.e., medication recommendation, 30-day readmission prediction and ICD coding. Extensive experiments demonstrate the power of MedM-PLM compared with state-of-the-art methods. Further analyses and visualizations show the robustness of our model, which could potentially provide more comprehensive interpretations for clinical decision-making., 12 pages, 5 figures accepted for publication in the IEEE Journal of Biomedical and Health Informatics (J-BHI)

Cohort selection is an essential prerequisite for clinical research, determining whether an individual satisfies given selection criteria. Previous works for cohort selection usually treated each selection criterion independently and ignored not only the meaning of each selection criterion but the relations among cohort selection criteria. To solve the problems above, we propose a novel unified machine reading comprehension (MRC) framework. In this MRC framework, we design simple rules to generate questions for each criterion from cohort selection guidelines and treat clues extracted by trigger words from patients' medical records as passages. A series of state-of-the-art MRC models based on BiDAF, BIMPM, BERT, BioBERT, NCBI-BERT, and RoBERTa are deployed to determine which question and passage pairs match. We also introduce a cross-criterion attention mechanism on representations of question and passage pairs to model relations among cohort selection criteria. Results on two datasets, that is, the dataset of the 2018 National NLP Clinical Challenge (N2C2) for cohort selection and a dataset from the MIMIC-III dataset, show that our NCBI-BERT MRC model with cross-criterion attention mechanism achieves the highest micro-averaged F1-score of 0.9070 on the N2C2 dataset and 0.8353 on the MIMIC-III dataset. It is competitive to the best system that relies on a large number of rules defined by medical experts on the N2C2 dataset. Comparing these two models, we find that the NCBI-BERT MRC model mainly performs worse on mathematical logic criteria. When using rules instead of the NCBI-BERT MRC model on some criteria regarding mathematical logic on the N2C2 dataset, we obtain a new benchmark with an F1-score of 0.9163, indicating that it is easy to integrate rules into MRC models for improvement.

BackgroundText Matching (TM) is a fundamental task of natural language processing widely used in many application systems such as information retrieval, automatic question answering, machine translation, dialogue system, reading comprehension, etc. In recent years, a large number of deep learning neural networks have been applied to TM, and have refreshed benchmarks of TM repeatedly. Among the deep learning neural networks, convolutional neural network (CNN) is one of the most popular networks, which suffers from difficulties in dealing with small samples and keeping relative structures of features. In this paper, we propose a novel deep learning architecture based on capsule network for TM, called CapsTM, where capsule network is a new type of neural network architecture proposed to address some of the short comings of CNN and shows great potential in many tasks.MethodsCapsTM is a five-layer neural network, including an input layer, a representation layer, an aggregation layer, a capsule layer and a prediction layer. In CapsTM, two pieces of text are first individually converted into sequences of embeddings and are further transformed by a highway network in the input layer. Then, Bidirectional Long Short-Term Memory (BiLSTM) is used to represent each piece of text and attention-based interaction matrix is used to represent interactive information of the two pieces of text in the representation layer. Subsequently, the two kinds of representations are fused together by BiLSTM in the aggregation layer, and are further represented with capsules (vectors) in the capsule layer. Finally, the prediction layer is a connected network used for classification. CapsTM is an extension of ESIM by adding a capsule layer before the prediction layer.ResultsWe construct a corpus of Chinese medical question matching, which contains 36,360 question pairs. This corpus is randomly split into three parts: a training set of 32,360 question pairs, a development set of 2000 question pairs and a test set of 2000 question pairs. On this corpus, we conduct a series of experiments to evaluate the proposed CapsTM and compare it with other state-of-the-art methods. CapsTM achieves the highest F-score of 0.8666.ConclusionThe experimental results demonstrate that CapsTM is effective for Chinese medical question matching and outperforms other state-of-the-art methods for comparison.

Biomedical named entity normalization (BNEN) is a fundamental natural language processing (NLP) task in the biomedical domain. Many representation learning-based methods have been successfully applied to BNEN in recent years. Most of them encode a given biomedical named entity mention (BNEM) and candidates separately, some of them consider relations between the BNEM and its candidates, however, few consider relations among the candidates, which may be useful for BNEN.In this paper, we propose a novel interaction-based synonym marginalization for BNEN, which can capture both the relations between a given mention and the mention's candidates and that among the candidates, called IA-BIOSYN. In IA-BIOSYN, given a BNEM, a candidate selector is used to obtain the candidates of the BNEM dynamically, then an interaction module is used to model BNEM-candidate relations as well as candidate-candidate relations, and finally a synonym marginalization module is used to determine which candidate(s) the BNEM should be mapped to. To validate the effectiveness of our proposed method, we compare it with other state-of-the-art (SOTA) methods on three public BNEN datasets: NCBI-Disease, BC5CDR-Disease and BC5CDR-Chemical.Our proposed method achieves Acc@1 of 0.9333, 0.9379 and 0.9693 on NCBI-Disease, BC5CDR-Disease and BC5CDR-Chemical, respectively, significantly better than other SOTA methods.Both the relations between a given BNEM and its candidates, and the relations among the candidates are useful for BNEN, and the proposed IA-BIOSYN can capture the two types of relations effectively.

Electronic health records (EHRs) have been widely used to help physicians to make decisions by predicting medical events such as diseases, prescriptions, outcomes, and so on. How to represent patient longitudinal medical data is the key to making these predictions. Recurrent neural network (RNN) is a popular model for patient longitudinal medical data representation from the view of patient status sequences, but it cannot represent complex interactions among different types of medical information, i.e., temporal medical event graphs, which can be represented by graph neural network (GNN). In this paper, we propose a hybrid method of RNN and GNN, called RGNN, for next-period prescription prediction from two views, where RNN is used to represent patient status sequences, and GNN is used to represent temporal medical event graphs. Experiments conducted on the public MIMIC-III ICU data show that the proposed method is effective for next-period prescription prediction, and RNN and GNN are mutually complementary.

This article proposes a novel sentence semantic equivalence identification (SSEI) method by using the semantic difference features between sentences. The lexical differences of a sentence pair are first extracted, and the bidirectional long short term memory (BiLSTM) network is then applied on them to generate the semantic difference representations. Finally, an efficient gate mechanism is proposed to integrate the semantic differences with existing models (called base model) to enhance their encoding capability in the SSEI task. Exhaustive experiments conducted on the standard Quora corpus, and the Large-scale Chinese Question Matching Corpus (LCQMC) show that the proposed gated semantic difference (GSD) method brings significant improvement for different existing state-of-the-art models. When the bidirectional encoder representations from transformers model (BERT) is used as the base model, the accuracy for SSEI on Quora is improved from 90.63% to 91.98%, and the F1 score on the LCQMC is improved from 87.0% to 87.7%, which outperforms the best-published results.

Predicting a comprehensive set of relevant labels on chest X-ray images faces great challenges towards bridging visual and textual modalities. Despite the success of Graph Convolutional Networks (GCN) on modeling label dependencies using co-occurrence matrix generated from dataset, they still suffer from inherent label imbalance in dataset and ignore the explicit relations among labels presented in external medical knowledge graph (KG). We argue that jointly exploiting both the label co-occurrence matrix in dataset and the label relations in external knowledge graph facilitates multi-label lesion annotation. To model relevant lesion labels more comprehensively, we propose a KG-augmented model via Aggregating Explicit Relations for multi-label lesion annotation, called AER-GCN. The KG-augmented model employs GCN to learn the explicit label relations in external medical KG, and aggregates the explicit relations into statistical graph built from label co-occurrence information. Specially, we present three approaches on modeling the explicit label correlations in external knowledge, and two approaches on incorporating the explicit relations into co-occurrence relations for lesion annotation. We exploit SNOMED CT as the source of external knowledge and evaluate the performance of AER-GCN on the ChestX-ray and IU X-ray datasets. Extensive experiments demonstrate that our model outperforms other state-of-the-art models.

Question matching (QM) is a fundamental task of information retrieval (IR)-based question-answering (QA) systems, which can be formulated as a paraphrase identification (PI) problem and relies on large-scale labeled data, which is not easy to be obtained, especially in specific domains such as the medical domain. In this paper, we investigate transfer learning for QM in the medical domain, aiming to adapt the shared knowledge learnt from questions for one disease to other diseases. We first compare six state-of-the-art deep learning methods for QM in Chinese on different diseases, and then apply a transfer learning framework to these deep learning methods. Experiments on a Chinese medical corpus of three diseases about children show that the proposed transfer learning framework is efficient and can bring stable performance improvement for most deep learning based QM models on small-scale medical corpora of different diseases.