CLEP: a hybrid data- and knowledge-driven framework for generating patient representations

Abstract

As machine learning and artificial intelligence increasingly attain a larger number of applications in the biomedical domain, at their core, their utility depends on the data used to train them. Due to the complexity and high dimensionality of biomedical data, there is a need for approaches that combine prior knowledge around known biological interactions with patient data. Here, we present CLEP, a novel approach that generates new patient representations by leveraging both prior knowledge and patient-level data. First, given a patient-level dataset and a knowledge graph containing relations across features that can be mapped to the dataset, CLEP incorporates patients into the knowledge graph as new nodes connected to their most characteristic features. Next, CLEP employs knowledge graph embedding models to generate new patient representations that can ultimately be used for a variety of downstream tasks, ranging from clustering to classification. We demonstrate how using new patient representations generated by CLEP significantly improves performance in classifying between patients and healthy controls for a variety of machine learning models, as compared to the use of the original transcriptomics data. Furthermore, we also show how incorporating patients into a knowledge graph can foster the interpretation and identification of biological features characteristic of a specific disease or patient subgroup. Finally, we released CLEP as an open source Python package together with examples and documentation. Availability CLEP is available to the bioinformatics community as an open source Python package at https://github.com/hybrid-kg/clep under the Apache 2.0 License. Supplementary information Supplementary data are available at Bioinformatics online.

Developing a novel causal inference algorithm for personalized biomedical causal graph learning using meta machine learning

“…The complexity and dimensionality of biomedical data require methodologies that integrate existing biological knowledge with patient-specific data, such as knowledge graphs [ 13 , 14 ]. For example, CLinical Embedding of Patients (CLEP) [ 15 ] incorporates patient-level multi-omics data into a knowledge graph to model the underlying relationships between patients and clinical features for identifying Alzheimer’s patients and their properties. Medical knowledge graphs or deep architectures that utilize patient-level medical data are important for developing accurate and generalizable clinical decision-support models [ 15 – 22 ].…”

“…For example, CLinical Embedding of Patients (CLEP) [ 15 ] incorporates patient-level multi-omics data into a knowledge graph to model the underlying relationships between patients and clinical features for identifying Alzheimer’s patients and their properties. Medical knowledge graphs or deep architectures that utilize patient-level medical data are important for developing accurate and generalizable clinical decision-support models [ 15 – 22 ]. Moreover, the utilization of a personalized biomedical graph enables the identification of patient-specific biological mechanisms [ 7 , 15 , 23 , 24 ], offering further insights into the causal relationships in specific diseases or patient subgroups.…”

Precision Nursing Research Based on Multimodal Knowledge Graph

“…The evolution of multi-modal knowledge graph research has initially revolved around two predominant modes: images and text [23,25]. Literature[26] laid the foundation by de ning multi-modal knowledge graphs, beginning with tasks such as link prediction and ontology matching.…”

Causal reasoning over knowledge graphs leveraging drug-perturbed and disease-specific transcriptomic signatures for drug discovery

“…While homogeneous networks, such as protein-protein interaction networks, can represent relationships between a single entity type, knowledge graphs (KGs) can incorporate a broad range of biological scales, from the genetic and molecular level (e.g., proteins, drugs, and biochemicals), to biological concepts (e.g., phenotypes and diseases). These KGs can then be utilized for several applications in drug discovery, such as providing insights into molecular mechanisms and therapeutic targets [ 1 – 2 ], side effect prediction in the early stages of drug development [ 3 ], target prioritization [ 4 ], and drug repositioning [ 5 ].…”