Enhancing Model Interpretability and Accuracy for Disease Progression Prediction via Phenotype-BasedPatient Similarity Learning

@article{Wang2019EnhancingMI,
  title={Enhancing Model Interpretability and Accuracy for Disease Progression Prediction via Phenotype-BasedPatient Similarity Learning},
  author={Yue Wang and Tong Wu and Yunlong Wang and Gao Wang},
  journal={Pacific Symposium on Biocomputing. Pacific Symposium on Biocomputing},
  year={2019},
  volume={25},
  pages={
          511-522
        },
  url={https://api.semanticscholar.org/CorpusID:202888555}
}
This paper proposes to learn patient similarity features as phenotypes from the aggregated patient-medical service matrix using non-negative matrix factorization and shows that the phenotype-based similarity features can improve prediction over multiple baselines, including logistic regression, random forest, convolutional neural network, and more.
[PDF] Semantic Reader
9 Citations
Highly Influential Citations
1
Background Citations
2
Methods Citations
1

Figures from this paper

Ask This Paper
BETA
AI-Powered

Our system tries to constrain to information found in this paper. Results quality may vary. Learn more about how we generate these answers.

9 Citations

Leveraging graph-based hierarchical medical entity embedding for healthcare applications

The efficacy of ME2Vec over competitive baselines on diagnosis prediction, readmission prediction, as well as recommending doctors to patients based on their medical conditions are demonstrated.

Unsupervised EHR‐based phenotyping via matrix and tensor decompositions

This paper provides a comprehensive review of low‐rank approximation‐based approaches for computational phenotyping and outlines different approaches for the validation of phenotypes, that is, the assessment of clinical significance.

EAPR: explainable and augmented patient representation learning for disease prediction

EAPR proposes to use two-stage gradient backpropagation to address the problem of unexplainable patient representation learning models due to the complex data enhancement process and the experimental results on real clinical data validate the effectiveness and explainability of the proposed approach.

Structure-aware siamese graph neural networks for encounter-level patient similarity learning

eTREE: Learning Tree-structured Embeddings

This work proposes eTREE, a model that incorporates the (usually ignored) tree structure to enhance the quality of the embeddings and leverages the special uniqueness properties of Nonnegative MF (NMF) to prove identifiability of eTrees.

Using Machine Learning to Classify Schizophrenia Based on Retinal Images

It is suggested that ML techniques can detect patterns in patients with FEP schizophrenia with greater performance using features extracted from OCT images than metrics provided by OCT machines.

Using Machine Learning to Classify Schizophrenia Based on Retinal Images

It is suggested that ML techniques can detect patterns in patients with FEP schizophrenia with greater performance using features extracted from OCT images than metrics provided by OCT machines.

A review of disease risk prediction methods and applications in the omics era.

The general process of omics-based disease risk model construction and the applications in four typical diseases are reviewed and the problems in current studies are highlighted and the potential opportunities and challenges for future clinical practice are explored.

Session Introduction

The size, complexity, and heterogeneity of biologically and medically relevant datasets have been dramatically increasing in recent decades in conjunction with newly developed experimental and…

24 References

Predicting Treatment Initiation from Clinical Time Series Data via Graph-Augmented Time-Sensitive Model

Results show that relational similarity can improve prediction over multiple baselines, for example a 5% incremental over long-short term memory baseline in terms of area under precision-recall curve.

Doctor AI: Predicting Clinical Events via Recurrent Neural Networks

Great generalizability of Doctor AI is demonstrated by adapting the resulting models from one institution to another without losing substantial accuracy, significantly higher than several baselines.

RETAIN: An Interpretable Predictive Model for Healthcare using Reverse Time Attention Mechanism

The REverse Time AttentIoN model (RETAIN) is developed for application to Electronic Health Records (EHR) data and achieves high accuracy while remaining clinically interpretable and is based on a two-level neural attention model that detects influential past visits and significant clinical variables within those visits.

An RNN Architecture with Dynamic Temporal Matching for Personalized Predictions of Parkinson's Disease

A deep model is proposed that directly learns patient similarity from longitudinal and multi-modal patient records with an Recurrent Neural Network (RNN) architecture, which learns the similarity between two longitudinal patient record sequences through dynamically matching temporal patterns in patient sequences.

Rare Disease Detection by Sequence Modeling with Generative Adversarial Networks

A deep learning method for detecting patients with exocrine pancreatic insufficiency (EPI) (a rare disease) and an accurate prediction with 0.56 PR-AUC which outperformed benchmark models in terms of precision and recall is presented.

Learning to Diagnose with LSTM Recurrent Neural Networks

This first study to empirically evaluate the ability of LSTMs to recognize patterns in multivariate time series of clinical measurements considers multilabel classification of diagnoses, and establishes the effectiveness of a simple LSTM network for modeling clinical data.

Using recurrent neural network models for early detection of heart failure onset

Deep learning models adapted to leverage temporal relations appear to improve performance of models for detection of incident heart failure with a short observation window of 12–18 months.

Deep Survival Analysis

Deep survival analysis is introduced, a hierarchical generative approach to survival analysis that scalably handles heterogeneous data types that occur in the EHR and is significantly superior in stratifying patients according to their risk.

Patient Subtyping via Time-Aware LSTM Networks

A patient subtyping model is proposed that leverages the proposed T-LSTM in an auto-encoder to learn a powerful single representation for sequential records of patients, which is then used to cluster patients into clinical subtypes.

EHR Big Data Deep Phenotyping

The big data solution, using flexible markup, provides a route to improved utilization of processing power for organizing patient records in genotype and phenotype research.