Hierarchical Bayesian nonparametric models for knowledge discovery from electronic medical records

Li, Cheng; Rana, Santu; Phung, Dinh; Venkatesh, Svetha

doi:10.1016/j.knosys.2016.02.005

Cited by 24 publications

(11 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Prior to determining the effective measures towards treatment, the experts should have clear knowledge of symptoms and the related diseases [4]. Learning of the relationships between the various symptoms as attributes and the diseases as conclusions from the ontological representation can be done easily at entity level discovery with the use of triplet representation [5], discovery of diseases through the symptoms in the graphical path [6] and through the mining algorithms [7].…”

Section: Related Workmentioning

confidence: 99%

“…Conditional Independence among the features Summarizing the CPD, for any given Bayesian network, which is a directed acyclic graph, CPD for each vertex s i in the given set of vertices S is defined as Where S 1 , S 2 , … , S n are the states of a node S in a DAG. As denominator in (5) will be 1 remaining as constant for any given inputs, the JPD of the states can be concluded as product of the CPD as shown in (6).…”

Section: Proposed Algorithmic Approachmentioning

confidence: 99%

See 1 more Smart Citation

Integration of Healthcare domain Ontologies using Bayesian Networks

2019

IJITEE

View full text Add to dashboard Cite

Semantic Web (SW) was created with the vision of knowledge sharing. Knowledge from the past and present help predict the future with the use of Machine Learning (ML) algorithms. SW powered with ontologies help in realizing machine interactions supporting automated knowledge extraction. Healthcare as a field of medical domain gives lot of importance for timely accurate decisions with the available features. Representing existing information in terms of ontologies, retrieving the decisions upon establishing interaction between the relevant ontologies within the same domain, knowledge sharing & reusing the existing facts are of great benefit to the medical practitioners and researchers which has lot of open challenges to be resolved in order to realize the same. To address the stated issues, an algorithmic approach – Ontologies Integration algorithm using Bayesian Networks (OIBN) based on Bayesian Belief Networks (BBN) working on Naïve beliefs has been proposed which works on symptoms through the attributes of related ontologies within the same domain exploring the symptom dependencies and their probability of occurrences in combination. Selection of features for integration will follow the steps proposed in Sequential Forward Feature Selection algorithm (SFFS). The observation on the correctness of the presented method over diabetic datasets represented in ontological form with integration of relevant features reveals that the knowledge graphs have been efficiently explored discovering the facts based on the probability theory. The experimental results conclude that the proposed technique is showing enhanced prediction accuracy of 80.95% which is better compared to accuracies of the individual ontologies prior to integration and existing state-of-art technique.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Proposed Algorithmic Approachmentioning

confidence: 99%

Integration of Healthcare domain Ontologies using Bayesian Networks

2019

IJITEE

View full text Add to dashboard Cite

show abstract

“…Reference [19] classified obesity and obesity types in thousands of EMRs by using the Support Vector Machine (SVM) and Naïve Bayes models, and the experimental results indicate that the SVM outperforms other methods. Reference [33] proposed a flexible hierarchical Bayesian nonparametric model to cluster medical data into groups. This work was inspired by the structure of ICD codes that can present semantic relationships between different diseases.…”

Section: B Emr/ehr Data Miningmentioning

confidence: 99%

Discovering Medication Patterns for High-Complexity Drug-Using Diseases Through Electronic Medical Records

et al. 2019

View full text Add to dashboard Cite

An Electronic Medical Record (EMR) is a professional document that contains all data generated during the treatment process. The EMR can utilize various data formats, such as numerical data, text, and images. Mining the information and knowledge hidden in the huge amount of EMR data is an essential requirement for clinical decision support, such as clinical pathway formulation and evidencebased medical research. In this paper, we propose a machine-learning-based framework to mine the hidden medication patterns in EMR text. The framework systematically integrates the Jaccard similarity evaluation, spectral clustering, the modified Latent Dirichlet Allocation and cross-matching among multiple features to find the residuals that describe additional knowledge and clusters hidden in multiple perspectives of highly complex medication patterns. These methods work together, step by step to reveal the underlying medication pattern. We evaluated the method by using real data from EMR text (patients with cirrhotic ascites) from a large hospital in China. The proposed framework outperforms other approaches for medication pattern discovery, especially for this disease with subtle medication treatment variances. The results also revealed little overlap among the discovered patterns; thus, the distinct features of each pattern are well studied through the proposed framework.INDEX TERMS Electronic medical record (EMR), medication pattern, discovery, machine learning, highcomplexity drug-use pattern.

show abstract

“…With the proliferation of Electronic Medical Records (EMRs), such high-quality medical data open a new window for data-driven knowledge discovery towards medical decision support. There are various medical knowledge discovery applications based on EMRs, including medical entity discovery [9], disease topic discovery [12], temporal pattern mining [27], medical event detection [7]. In this paper, we study medical knowledge condition information discovery from EMRs, which plays a crucial role in medical systems and related applications.…”

Section: Related Workmentioning

confidence: 99%

MedTruth: A Semi-supervised Approach to Discovering Knowledge Condition Information from Multi-Source Medical Data

Deng

Shen

et al. 2018

Preprint

View full text Add to dashboard Cite

Knowledge Graph (KG) contains entities and the relations between entities. Due to its representation ability, KG has been successfully applied to support many medical/healthcare tasks. However, in the medical domain, knowledge holds under certain conditions. For example, symptom runny nose highly indicates the existence of disease whooping cough when the patient is a baby rather than the people at other ages. Such conditions for medical knowledge are crucial for decision-making in various medical applications, which is missing in existing medical KGs. In this paper, we aim to discovery medical knowledge conditions from texts to enrich KGs.Electronic Medical Records (EMRs) are systematized collection of clinical data and contain detailed information about patients, thus EMRs can be a good resource to discover medical knowledge conditions. Unfortunately, the amount of available EMRs is limited due to reasons such as regularization. Meanwhile, a large amount of medical question answering (QA) data is available, which can greatly help the studied task. However, the quality of medical QA data is quite diverse, which may degrade the quality of the discovered medical knowledge conditions. In the light of these challenges, we propose a new truth discovery method, MedTruth, for medical knowledge condition discovery, which incorporates prior source quality information into the source reliability estimation procedure, and also utilizes the knowledge triple information for trustworthy information computation. We conduct series of experiments on realworld medical datasets to demonstrate that the proposed method can discover meaningful and accurate conditions for medical knowledge by leveraging both EMR and QA data. Further, the proposed method is tested on synthetic datasets to validate its effectiveness under various scenarios.

show abstract

Hierarchical Bayesian nonparametric models for knowledge discovery from electronic medical records

Cited by 24 publications

References 21 publications

Integration of Healthcare domain Ontologies using Bayesian Networks

Integration of Healthcare domain Ontologies using Bayesian Networks

Discovering Medication Patterns for High-Complexity Drug-Using Diseases Through Electronic Medical Records

MedTruth: A Semi-supervised Approach to Discovering Knowledge Condition Information from Multi-Source Medical Data

Contact Info

Product

Resources

About