skip to main content

This content will become publicly available on February 22, 2023

Title: Context-aware Health Event Prediction via Transition Functions on Dynamic Disease Graphs
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 results more » on two real-world EHR datasets show that the proposed model outperforms state of the art in predicting health events. « less
Authors:
; ;
Award ID(s):
2047843 1948432
Publication Date:
NSF-PAR ID:
10318648
Journal Name:
In the Proceedings of the 36th AAAI Conference on Artificial Intelligence
Sponsoring Org:
National Science Foundation
More Like this
  1. 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 modelmore »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.

    « less
  2. Abstract Background The United States, and especially West Virginia, have a tremendous burden of coronary artery disease (CAD). Undiagnosed familial hypercholesterolemia (FH) is an important factor for CAD in the U.S. Identification of a CAD phenotype is an initial step to find families with FH. Objective We hypothesized that a CAD phenotype detection algorithm that uses discrete data elements from electronic health records (EHRs) can be validated from EHR information housed in a data repository. Methods We developed an algorithm to detect a CAD phenotype which searched through discrete data elements, such as diagnosis, problem lists, medical history, billing, and procedure (International Classificationmore »of Diseases [ICD]-9/10 and Current Procedural Terminology [CPT]) codes. The algorithm was applied to two cohorts of 500 patients, each with varying characteristics. The second (younger) cohort consisted of parents from a school child screening program. We then determined which patients had CAD by systematic, blinded review of EHRs. Following this, we revised the algorithm by refining the acceptable diagnoses and procedures. We ran the second algorithm on the same cohorts and determined the accuracy of the modification. Results CAD phenotype Algorithm I was 89.6% accurate, 94.6% sensitive, and 85.6% specific for group 1. After revising the algorithm (denoted CAD Algorithm II) and applying it to the same groups 1 and 2, sensitivity 98.2%, specificity 87.8%, and accuracy 92.4; accuracy 93% for group 2. Group 1 F1 score was 92.4%. Specific ICD-10 and CPT codes such as “coronary angiography through a vein graft” were more useful than generic terms. Conclusion We have created an algorithm, CAD Algorithm II, that detects CAD on a large scale with high accuracy and sensitivity (recall). It has proven useful among varied patient populations. Use of this algorithm can extend to monitor a registry of patients in an EHR and/or to identify a group such as those with likely FH.« less
  3. Research and experimentation using big data sets, specifically large sets of electronic health records (EHR) and social media data, is demonstrating the potential to understand the spread of diseases and a variety of other issues. Applications of advanced algorithms, machine learning, and artificial intelligence indicate a potential for rapidly advancing improvements in public health. For example, several reports indicate that social media data can be used to predict disease outbreak and spread (Brown, 2015). Since real-world EHR data has complicated security and privacy issues preventing it from being widely used by researchers, there is a real need to synthetically generatemore »EHR data that is realistic and representative. Current EHR generators, such as Syntheaä (Walonoski et al., 2018) only simulate and generate pure medical-related data. However, adding patients’ social media data with their simulated EHR data would make combined data more comprehensive and realistic for healthcare research. This paper presents a patients’ social media data generator that extends an EHR data generator. By adding coherent social media data to EHR data, a variety of issues can be examined for emerging interests, such as where a contagious patient may have been and others with whom they may have been in contact. Social media data, specifically Twitter data, is generated with phrases indicating the onset of symptoms corresponding to the synthetically generated EHR reports of simulated patients. This enables creation of an open data set that is scalable up to a big-data size, and is not subject to the security, privacy concerns, and restrictions of real healthcare data sets. This capability is important to the modeling and simulation community, such as scientists and epidemiologists who are developing algorithms to analyze the spread of diseases. It enables testing a variety of analytics without revealing real-world private patient information.« less
  4. Doshi-Velez, Finale ; Fackler, Jim ; Jung, Ken ; Kale, David ; Ranganath, Rajesh ; Wallace, Byron ; Wiens, Jenna (Ed.)
    Electronic Health Records (EHRs) provide vital contextual information to radiologists and other physicians when making a diagnosis. Unfortunately, because a given patient’s record may contain hundreds of notes and reports, identifying relevant information within these in the short time typically allotted to a case is very difficult. We propose and evaluate models that extract relevant text snippets from patient records to provide a rough case summary intended to aid physicians considering one or more diagnoses. This is hard because direct supervision (i.e., physician annotations of snippets relevant to specific diagnoses in medical records) is prohibitively expensive to collect at scale.more »We propose a distantly supervised strategy in which we use groups of International Classification of Diseases (ICD) codes observed in ‘future’ records as noisy proxies for ‘downstream’ diagnoses. Using this we train a transformer-based neural model to perform extractive summarization conditioned on potential diagnoses. This model defines an attention mechanism that is conditioned on potential diagnoses (queries) provided by the diagnosing physician. We train (via distant supervision) and evaluate variants of this model on EHR data from Brigham and Women’s Hospital in Boston and MIMIC-III (the latter to facilitate reproducibility). Evaluations performed by radiologists demonstrate that these distantly supervised models yield better extractive summaries than do unsupervised approaches. Such models may aid diagnosis by identifying sentences in past patient reports that are clinically relevant to a potential diagnosis. Code is available at https://github.com/dmcinerney/ehr-extraction-models.« less
  5. Real-world networked systems often show dynamic properties with continuously evolving network nodes and topology over time. When learning from dynamic networks, it is beneficial to correlate all temporal networks to fully capture the similarity/relevance between nodes. Recent work for dynamic network representation learning typically trains each single network independently and imposes relevance regularization on the network learning at different time steps. Such a snapshot scheme fails to leverage topology similarity between temporal networks for progressive training. In addition to the static node relationships within each network, nodes could show similar variation patterns (e.g., change of local structures) within the temporalmore »network sequence. Both static node structures and temporal variation patterns can be combined to better characterize node affinities for unified embedding learning. In this paper, we propose Graph Attention Evolving Networks (GAEN) for dynamic network embedding with preserved similarities between nodes derived from their temporal variation patterns. Instead of training graph attention weights for each network independently, we allow model weights to share and evolve across all temporal networks based on their respective topology discrepancies. Experiments and validations, on four real-world dynamic graphs, demonstrate that GAEN outperforms the state-of-the-art in both link prediction and node classification tasks.

    « less