PubMed Health⌕ Search

SEARCH · PubMed Health

Results for “knowledge graphs”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 37 records · Page 2Linked to original sources
PubMed2025

Integrating explainable artificial intelligence with multiomics systems biology and electronic health record data mining for personalized drug repurposing in Alzheimer's disease.

Alzheimer's disease (AD) is characterized by region- and patient-specific molecular heterogeneity, which hinders therapeutic design. In this study, we introduce PRISM-ML (PRecision-medicine using Interpretable Systems and Multiomics with Machine Learning), an open-source integrated analysis pipeline that combines interpretable machine learning with systems biology and electronic health records data mining to elucidate the molecular diversity of AD and predict promising drug repurposing opportunities. First, we integrated and harmonized transcriptomic (bulk RNA-seq) and genomic (genome-wide association study) data from 2105 brain samples, each with matched data from the same individual (1363 AD patients, 742 controls; 9 tissues), sourced from three independent studies. Random forest classifiers with SHapley Additive exPlanations identified patient-specific biomarkers; unsupervised clustering resolved 36 molecularly distinct subtissues (defined as clusters of samples within a brain tissue that share a specific expression pattern); and gene-gene coexpression networks prioritized 262 high-centrality bottleneck genes as putative regulators of dysregulated pathways. Next, knowledge graph-based drug repurposing predicted six Food and Drug Administration (FDA)-approved drugs that simultaneously target multiple bottleneck genes and multiple AD-relevant pathways. Notably, in a large US de-identified insurance-claims database (n&#x2009;=&#x2009;364&#xa0;733), exposure to promethazine, one of the candidate drugs, was associated with a 57%-62% lower incidence of AD versus an active antihistamine comparator (adjusted hazard ratio 0.38; inverse-probability weighted 0.43; both P&#x2009;<&#x2009;.001), providing real-world support for its repurposing potential. In summary, PRISM-ML, as an explainable multiomics analysis pipeline, is readily transferable to other complex diseases, advancing precision medicine.

Alzheimer Disease↗
PubMed2025

Integrating explainable AI with multiomics systems biology and EHR data mining for personalized drug repurposing in Alzheimer's disease.

Alzheimer's disease (AD) is characterized by region- and patient-specific molecular heterogeneity, which hinders therapeutic design. In this study, we introduce PRISM-ML (PRecision-medicine using Interpretable Systems and Multiomics with Machine Learning), an open-source integrated analysis pipeline that combines interpretable machine learning with systems biology and electronic health record (EHR) data mining to elucidate the molecular diversity of AD and predict promising drug repurposing opportunities. First, we integrated and harmonized transcriptomic (bulk RNA-seq) and genomic (genome-wide association study) data from 2105 brain samples, each with matched data from the same individual (1363 AD patients, 742 controls; nine tissues), sourced from three independent studies. Random forest classifiers with SHapley Additive exPlanations (SHAP) identified patient-specific biomarkers; unsupervised clustering resolved 36 molecularly distinct "subtissues" (clusters of samples); and gene-gene co-expression networks prioritized 262 high-centrality bottleneck genes as putative regulators of dysregulated pathways. Next, knowledge graph-based drug repurposing predicted six FDA-approved drugs that simultaneously target multiple bottleneck genes and multiple AD-relevant pathways. Notably, in a large U.S. de-identified insurance-claims database (n = 364733), exposure to promethazine, one of the candidate drugs, was associated with a 57-62 % lower incidence of AD versus an active antihistamine comparator (adjusted hazard ratio 0.38; inverse-probability weighted 0.43; both p < 0.001), providing real-world support for its repurposing potential. In summary, PRISM-ML, as an explainable multi-omics analysis pipeline, is readily transferable to other complex diseases, advancing precision medicine.

Computational Biology↗
PubMed1998

Ovarian ultrasound image analysis: follicle segmentation.

Ovarian ultrasound is an effective tool in infertility treatment. Repeated measurements of the size and shape of follicles over several days are the primary means of evaluation by physicians. Currently, follicle wall segmentation is achieved by manual tracing which is time consuming and susceptible to inter-operator variation. An automated method for follicle wall segmentation is reported that uses a four-step process based on watershed segmentation and knowledge-based graph search algorithm which utilizes priori information about follicle structure for inner and outer wall detection. The automated technique was tested on 36 ultrasonographic images of women's ovaries. Validation against manually traced borders has shown good correlation of manually defined and computer-determined area measurements (R2 = 0.85 - 0.96). The border positioning errors were small: 0.63+/-0.36 mm for inner border and 0.67+/-0.41 mm for outer border detection. The use of watershed segmentation and graph search methods facilitates fast, accurate inner and outer border detection with minimal user-interaction.

Algorithms↗
PubMed2001

Knowledge representation and retrieval using conceptual graphs and free text document self-organisation techniques.

Hospitals generate and store a large amount of clinical data each year, a significant portion of which is in free text format. Conventional database storage and retrieval algorithms are incapable of effectively processing free text medical data. The rich information and knowledge buried in healthcare records are unavailable for clinical decision-making. We examined a number of techniques for structuring and processing free text documents to effective and efficient for information retrieval and knowledge discovery. One critical success criterion is that the complexity of the techniques must be polynomial both in space and time for them to be able to cope with very large databases. We used conceptual graphs (CG) to capture the structure and semantic information/knowledge contained within the free text medical documents. Ordering and self-organising techniques (lattice techniques and knowledge space) were used to improve organisation of concepts from standard medical nomenclatures and large sets of free text medical documents. Pair-wise union of CG was performed to identify the common generalisation structure and a lattice structure of these CG documents. A combination of all three techniques allowed us to organise a set of 9000 discharge summaries into a generalisation hierarchy that supported efficient and rich information/knowledge retrieval.

Classification↗
PubMed2005

Learning protein secondary structure from sequential and relational data.

We propose a method for sequential supervised learning that exploits explicit knowledge of short- and long-range dependencies. The architecture consists of a recursive and bi-directional neural network that takes as input a sequence along with an associated interaction graph. The interaction graph models (partial) knowledge about long-range dependency relations. We tested the method on the prediction of protein secondary structure, a task in which relations due to beta-strand pairings and other spatial proximities are known to have a significant effect on the prediction accuracy. In this particular task, interactions can be derived from knowledge of protein contact maps at the residue level. Our results show that prediction accuracy can be significantly boosted by the integration of interaction graphs.

Animals↗
PubMed1993

[A method of analysing the reasoning applied in nursing in neuroscience].

It is a preliminary study testing the applicability of a method for knowledge characterization based on graphs theory. 11 (eleven) nurses working in a medical surgical intensive care unit including neurological patients constructed graphs to explain the diagnostic process. The results show that: the knowledge structure is in a developing stage; there is agreement regarding to diagnosis formulation and the data base on which it is founded is diversified and link to a specified form of related causes. The method demonstrated to be feasible and should be reapplied on a larger population.

Adult↗
PubMed1986

[Testing an expert system for hypertension].

An Expert System (ES) has been connected to a database management system for the management and follow-up of hypertensive patients. The patient data base, called Artemis, contains approximately 18,000 medical records. About 90% of the initial informations used by the ES is contained in the medical records of the Artemis data base. The knowledge base consists of 870 rules. A first group of rules allows the description of knowledge structures (hierachies, graphs and mutual exclusions). The second group consists of production rules which describe the dynamic reasoning of the expert. The inference engine uses a combination of forward and backward chaining. The ES produce diagnostic hypotheses (possible causes of hypertension) and therapeutic suggestions before and after requiring additional information (patient supplementary interrogation, biological or radiological investigations). The evaluation of the diagnostic performance of the ES was made on 40 confirmed cases of secondary hypertension (SH) and 40 cases of essential hypertension (EH). The initial initial diagnosis, just after the forward chaining step, was correct in 17 cases of SH and 32 cases of EH. The final diagnosis proposed after several steps of forward and backward chaining was correct in 37 cases (92%) of SH and 36 (90%) of EH. Averages of 5 (EH) and 8 (SH) questions were formulated by the ES to reach the final diagnosis. The integration of the ES to the database is expected to facilitate the validation of the knowledge base and to enhance its overall acceptability. Whether or not such an integration will be useful and accepted as a complementary tool by physicians remains however an open question.

Decision Making, Computer-Assisted↗
PubMed2026

engGNN: a dual-graph neural network for omics-based disease classification and feature selection.

Omics data, such as transcriptomics, proteomics, and metabolomics, provide critical insights into disease mechanisms and clinical outcomes. However, their high dimensionality, small sample sizes, and intricate biological networks pose major challenges for reliable prediction and meaningful interpretation. Graph neural networks offer a promising way to integrate prior knowledge by encoding feature relationships as graphs. Yet, existing methods typically rely solely on either an externally curated feature graph or a data-driven generated graph, which limits their ability to capture complementary information. To address this, we propose the external and generated Graph Neural Network (engGNN), a dual-graph framework that jointly leverages both external biological networks and data-driven generated graphs. Specifically, engGNN constructs a biologically informed undirected feature graph from established network databases and complements it with a directed feature graph derived from tree-ensemble models. This dual-graph design produces more comprehensive representations, thereby improving predictive performance and interpretability. Through extensive simulation studies and real-world applications to three independent gene expression datasets, engGNN consistently demonstrates strong classification performance compared with competitive baselines. Beyond classification, engGNN provides feature- and source-level interpretability, enabling biologically meaningful analyses such as pathway enrichment analysis. Taken together, these results highlight engGNN as a robust, flexible, and interpretable framework for disease classification and biomarker discovery in high-dimensional omics contexts.

Graph Neural Networks↗
PubMed2000

PSG-EXPERT. An expert system for the diagnosis of sleep disorders.

This paper describes PSG-EXPERT, an expert system in the domain of sleep disorders exploring polysomnographic data. The developed software tool is addressed from two points of view: (1)--as an integrated environment for the development of diagnosis-oriented expert systems; (2)--as an auxiliary diagnosis tool in the particular domain of sleep disorders. Developed over a Windows platform, this software tool extends one of the most popular shells--CLIPS (C Language Integrated Production System) with the following features: backward chaining engine; graph-based explanation facilities; knowledge editor including a fuzzy fact editor and a rules editor, with facts-rules integrity checking; belief revision mechanism; built-in case generator and validation module. It therefore provides graphical support for knowledge acquisition, edition, explanation and validation. From an application domain point of view, PSG-Expert is an auxiliary diagnosis system for sleep disorders based on polysomnographic data, that aims at assisting the medical expert in his diagnosis task by providing automatic analysis of polysomnographic data, summarising the results of this analysis in terms of a report of major findings and possible diagnosis consistent with the polysomnographic data. Sleep disorders classification follows the International Classification of Sleep Disorders. Major features of the system include: browsing on patients data records; structured navigation on Sleep Disorders descriptions according to ASDA definitions; internet links to related pages; diagnosis consistent with polysomnographic data; graphical user-interface including graph-based explanatory facilities; uncertainty modelling and belief revision; production of reports; connection to remote databases.

Computer Systems↗
PubMed1997

Toxopert-I: knowledge-based automatic interpretation of serological tests for toxoplasmosis.

Primary infection with Toxoplasma gondii, a parasite found in most regions of the world, is asymptomatic in more than 80% of cases. However, primary infection with Toxoplasma gondii in a pregnant woman might cause fetal infection and severe damage. Most cases do not require treatment. This applies to women without any infection (denoted as seronegative) and women who have acquired the infection before conception (denoted as latent). In contrast, women with postconceptual infection require immediate treatment to prevent or ameliorate fetal infection. We have developed an expert system, called Toxoport-I, designed for routine laboratory work, which automatically interprets serological test results of toxoplasma infection. By using the system the clinician can also examine questionable cases by interactively exploring possible results. We used a popular method of designing expert systems applied to medical interpretation and therapy advice, the rule-based one. In order to meet the requirements of automatic interpretation in toxoplasma serology the following characteristics were introduced: the interpretation of sequences of test results, the possibility of excluding inconsistent test results and the adaptability of the knowledge base. A decision graph that covers the different kinds of infections as well as therapy and recommendations for further tests was designed, implemented and was clinically tested by carrying out a retrospective study including 1000 pregnant women. A comparison of Toxoport-I and the clinician's interpretations yielded sensitivity and specificity rates of over 99% each.

Animals↗
PubMed1993

[A serum cholesterol study in a consulting population and its implications in clinical practice].

OBJECTIVE: To investigate serious cholesterol levels in the population seeking medical attention, utilising HOTH (How Often That High) graphs, and to look at implications these results have for recording of cholesterol incidence. DESIGN: This was a non-longitudinal, retrospective, observational study. SETTING: Carried out at the general practice level, in the Valdefierro Basic Health Area (Zaragoza Province). PATIENTS: 742 subjects aged between 18 and 65 years with no previous history of dyslipemia. MEASUREMENTS AND MAIN RESULTS: After obtaining the figures on cholesterolemia, the HOTH graph was drawn from which, for a given cholesterol value, the percentage of the population with a given value can be seen. The mean cholesterol value was 201 mg/dl. A total of 54% of male subjects and 40% of female subjects were situated above the threshold figure of 200 mg/dl. Furthermore, 15% of male subjects and 10% of female subjects were situated above 250 mg/dl. CONCLUSIONS: The application of recording procedures for cholesterol, based on a globally discriminatory score, reveals a startling number of patients who need tests and treatment. We suggest a strategy based on the different levels according to age, sex, bloodstream cholesterol level, and the existence or otherwise of other cardiovascular risk factors. Lastly, we would mention the usefulness of the HOTH graphs in improving our knowledge of the cholesterol levels in a population.

Adolescent↗
PubMed2006

Comparisons of graph-structure clustering methods for gene expression data.

Although many numerical clustering algorithms have been applied to gene expression data analysis, the essential step is still biological interpretation by manual inspection. The correlation between genetic co-regulation and affiliation to a common biological process is what biologists expect. Here, we introduce some clustering algorithms that are based on graph structure constituted by biological knowledge. After applying a widely used dataset, we compared the result clusters of two of these algorithms in terms of the homogeneity of clusters and coherence of annotation and matching ratio. The results show that the clusters of knowledge-guided analysis are the kernel parts of the clusters of Gene Ontology (GO)-Cluster software, which contains the genes that are most expression correlative and most consistent with biological functions. Moreover, knowledge-guided analysis seems much more applicable than GO-Cluster in a larger dataset.

Algorithms↗
PubMed2005

The emergence of scaling in sequence-based physical models of protein evolution.

It has recently been discovered that many biological systems, when represented as graphs, exhibit a scale-free topology. One such system is the set of structural relationships among protein domains. The scale-free nature of this and other systems has previously been explained using network growth models that, although motivated by biological processes, do not explicitly consider the underlying physics or biology. In this work we explore a sequence-based model for the evolution protein structures and demonstrate that this model is able to recapitulate the scale-free nature observed in graphs of real protein structures. We find that this model also reproduces other statistical feature of the protein domain graph. This represents, to our knowledge, the first such microscopic, physics-based evolutionary model for a scale-free network of biological importance and as such has strong implications for our understanding of the evolution of protein structures and of other biological networks.

Algorithms↗
PubMed1994

Representation of medical concepts of the thyroid gland by physicians in anatomy and pathology.

An experimental study in cognitive psychology is described, concerning the categorization of medical concepts into specific classes, expressed by physicians specialized in anatomic pathology consultations of the thyroid gland. This study belongs to a medical computer science project, called ARISTOTLE, concerning Natural Language Processing of specialized medical reports in anatomic pathology of the thyroid gland. This research has been done for two reasons; first, to specify the characteristics of human expert categorization in an area of medical knowledge and, secondly, to validate the hierarchical organization of a prototype declarative knowledge base. In this experiment, physicians were asked to categorize 121 concepts into 10 proposed classes. These classes and concepts belong to expert knowledge represented in a conceptual graph that was constructed before the experiment. Results show variable semantic distances between concepts of a same class, and dynamic variations of these distances due to contextual representation.

Artificial Intelligence↗
PubMed1995

Toward a medical linguistic knowledge base.

This paper presents the design of a Medical Linguistic Knowledge Base (MLKB). This MLKB is intended to be the multilingual recipient for all the declarative knowledge about languages. It includes words, their syntax and their conceptual representation, typology of concepts of the domain, rules for semantic analysis and conceptual schemata. For that purpose, Sowa's conceptual graphs are considered as an adequate knowledge representation. The MLKB will be an enormous body of information, and the difficulty to feed it and to validate it appears immediately. Therefore, it is necessary to start an international initiative to merge efforts from different groups.

Language↗
PubMed2004

Detection of activity centers in cellular pathways using transcript profiling.

We present a new computational method for identifying regulated pathway components in transcript profiling (TP) experiments by evaluating transcriptional activity in the context of known biological pathways. We construct a graph representing thousands of protein functional relationships by integrating knowledge from public databases and review articles. We use the notion of distance in a graph to define pathway neighborhoods. The pathways perturbed in an experiment are then identified as the subgraph induced by the genes, referred to as activity centers, having significant density of transcriptional activity in their functional neighborhoods. We illustrate the predictive power of this approach by performing and analyzing an experiment of TP53 overexpression in NCI-H125 cells. The detected activity centers are in agreement with the known TP53 activation effects and our independent experimental results. We also apply the method to a serum starvation experiment using HEY cells and investigate the predicted activity of the transcription factor MYC. Finally, we discuss interesting properties of the activity center approach and its possible applications beyond the comparison of two experiments.

Algorithms↗
PubMed1997

THEMPO: a knowledge-based system for therapy planning in pediatric oncology.

This article describes the knowledge-based system THEMPO (Therapy Management in Pediatric Oncology), which supports protocol-directed therapy planning and configuration in pediatric oncology. THEMPO provides a semantic network controlled by graph grammars to cover the different types of knowledge relevant in the domain, and offers a suite of acquisition tools for knowledge base authoring. Medical problem solvers, operating on the oncological network, reason about adequate therapeutic and diagnostic timetables for a patient. Furthermore, a corresponding patient record, also based on semantic networks and graph grammars, has been implemented to represent the course of therapy of an oncological patient.

Antineoplastic Combined Chemotherapy Protocols↗