skip to main content

Title: Collaborative Graph Learning with Auxiliary Text for Temporal Event Prediction in Healthcare

Accurate and explainable health event predictions are becoming crucial for healthcare providers to develop care plans for patients. The availability of electronic health records (EHR) has enabled machine learning advances in providing these predictions. However, many deep-learning-based methods are not satisfactory in solving several key challenges: 1) effectively utilizing disease domain knowledge; 2) collaboratively learning representations of patients and diseases; and 3) incorporating unstructured features. To address these issues, we propose a collaborative graph learning model to explore patient-disease interactions and medical domain knowledge. Our solution is able to capture structural features of both patients and diseases. The proposed model also utilizes unstructured text data by employing an attention manipulating strategy and then integrates attentive text features into a sequential learning process. We conduct extensive experiments on two important healthcare problems to show the competitive prediction performance of the proposed method compared with various state-of-the-art models. We also confirm the effectiveness of learned representations and model interpretability by a set of ablation and case studies.

; ; ; ;
Award ID(s):
Publication Date:
Journal Name:
Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence
Sponsoring Org:
National Science Foundation
More Like this
  1. Electronic health records (EHRs) have been heavily used in modern healthcare systems for recording patients' admission information to health facilities. Many data-driven approaches employ temporal features in EHR for predicting specific diseases, readmission times, and diagnoses of patients. However, most existing predictive models cannot fully utilize EHR data, due to an inherent lack of labels in supervised training for some temporal events. Moreover, it is hard for the existing methods to simultaneously provide generic and personalized interpretability. To address these challenges, we propose Sherbet, a self-supervised graph learning framework with hyperbolic embeddings for temporal health event prediction. We first propose a hyperbolic embedding method with information flow to pretrain medical code representations in a hierarchical structure. We incorporate these pretrained representations into a graph neural network (GNN) to detect disease complications and design a multilevel attention method to compute the contributions of particular diseases and admissions, thus enhancing personalized interpretability. We present a new hierarchy-enhanced historical prediction proxy task in our self-supervised learning framework to fully utilize EHR data and exploit medical domain knowledge. We conduct a comprehensive set of experiments on widely used publicly available EHR datasets to verify the effectiveness of our model. Our results demonstrate the proposedmore »model's strengths in both predictive tasks and interpretable abilities.« less
  2. Abstract Background

    Natural language processing (NLP) tasks in the health domain often deal with limited amount of labeled data due to high annotation costs and naturally rare observations. To compensate for the lack of training data, health NLP researchers often have to leverage knowledge and resources external to a task at hand. Recently, pretrained large-scale language models such as the Bidirectional Encoder Representations from Transformers (BERT) have been proven to be a powerful way of learning rich linguistic knowledge from massive unlabeled text and transferring that knowledge to downstream tasks. However, previous downstream tasks often used training data at such a large scale that is unlikely to obtain in the health domain. In this work, we aim to study whether BERT can still benefit downstream tasks when training data are relatively small in the context of health NLP.


    We conducted a learning curve analysis to study the behavior of BERT and baseline models as training data size increases. We observed the classification performance of these models on two disease diagnosis data sets, where some diseases are naturally rare and have very limited observations (fewer than 2 out of 10,000). The baselines included commonly used text classification models such as sparse andmore »dense bag-of-words models, long short-term memory networks, and their variants that leveraged external knowledge. To obtain learning curves, we incremented the amount of training examples per disease from small to large, and measured the classification performance in macro-averaged$$F_{1}$$F1score.


    On the task of classifying all diseases, the learning curves of BERT were consistently above all baselines, significantly outperforming them across the spectrum of training data sizes. But under extreme situations where only one or two training documents per disease were available, BERT was outperformed by linear classifiers with carefully engineered bag-of-words features.


    As long as the amount of training documents is not extremely few, fine-tuning a pretrained BERT model is a highly effective approach to health NLP tasks like disease classification. However, in extreme cases where each class has only one or two training documents and no more will be available, simple linear models using bag-of-words features shall be considered.

    « less
  3. With the wide application of electronic health records (EHR) in healthcare facilities, health event prediction with deep learning has gained more and more attention. A common feature of EHR data used for deep-learning-based predictions is historical diagnoses. Existing work mainly regards a diagnosis as an independent disease and does not consider clinical relations among diseases in a visit. Many machine learning approaches assume disease representations are static in different visits of a patient. However, in real practice, multiple diseases that are frequently diagnosed at the same time reflect hidden patterns that are conducive to prognosis. Moreover, the development of a disease is not static since some diseases can emerge or disappear and show various symptoms in different visits of a patient. To effectively utilize this combinational disease information and explore the dynamics of diseases, we propose a novel context-aware learning framework using transition functions on dynamic disease graphs. Specifically, we construct a global disease co-occurrence graph with multiple node properties for disease combinations. We design dynamic subgraphs for each patient's visit to leverage global and local contexts. We further define three diagnosis roles in each visit based on the variation of node properties to model disease transition processes. Experimental resultsmore »on two real-world EHR datasets show that the proposed model outperforms state of the art in predicting health events.« less
  4. Off-label drug use refers to using marketed drugs for indications that are not listed in their FDA labeling information. Such uses are very common and sometimes inevitable in clinical practice. To some extent, off-label drug uses provide a pathway for clinical innovation, however, they could cause serious adverse effects due to lacking scientific research and tests. Since identifying the off-label uses can provide a clue to the stakeholders including healthcare providers, patients, and medication manufacturers to further the investigation on drug efficacy and safety, it raises the demand for a systematic way to detect off-label uses. Given data contributed by health consumers in online health communities (OHCs), we developed an automated approach to detect off-label drug uses based on heterogeneous network mining. We constructed a heterogeneous healthcare network with medical entities (e.g. disease, drug, adverse drug reaction) mined from the text corpus, which involved 50 diseases, 1,297 drugs, and 185 ADRs, and determined 13 meta paths between the drugs and diseases. We developed three metrics to represent the meta-path-based topological features. With the network features, we trained the binary classifiers built on Random Forest algorithm to recognize the known drug-disease associations. The best classification model that used lift to measuremore »path weights obtained F1-score of 0.87, based on which, we identified 1,009 candidates of off-label drug uses and examined their potential by searching evidence from PubMed and FAERS.« less
  5. Off-label drug use is quite common in clinical practice and inevitable to some extent. Such uses might deliver effective treatment and suggest clinical innovation sometimes, however, they have the unknown risk to cause serious outcomes due to lacking scientific support. As gaining information about off-label drug use could present a clue to the stakeholders such as healthcare professionals and medication manufacturers to further the investigation on drug efficacy and safety, it raises the need to develop a systematic way to detect off-label drug uses. Considering the increasing discussions in online health communities (OHCs) among the health consumers, we proposed to harness the large volume of timely information in OHCs to develop an automated method for detecting off-label drug uses from health consumer generated data. From the text corpus, we extracted medical entities (diseases, drugs, and adverse drug reactions) with lexicon-based approaches and measured their interactions with word embedding models, based on which, we constructed a heterogeneous healthcare network. We defined several meta-path-based indicators to describe the drug-disease associations in the heterogeneous network and used them as features to train a binary classifier built on Random Forest algorithm, to recognize the known drug-disease associations. The classification model obtained better results whenmore »incorporating word embedding features and achieved the best performance when using both association rule mining features and word embedding features, with F1-score reaching 0.939, based on which, we identified 2,125 possible off-label drug uses and checked their potential by searching evidence in PubMed and FAERS.« less