SEARCH · PubMed Health

Results for “network analysis”

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 19 recordsLinked to original sources

PubMed2025

MiNEApy: enhancing enrichment network analysis in metabolic networks.

MOTIVATION: Modeling genome-scale metabolic networks (GEMs) helps understand metabolic fluxes in cells at a specific state under defined environmental conditions or perturbations. Elementary flux modes (EFMs) are powerful tools for simplifying complex metabolic networks into smaller, more manageable pathways. However, the enumeration of all EFMs, especially within GEMs, poses significant challenges due to computational complexity. Additionally, traditional EFM approaches often fail to capture essential aspects of metabolism, such as co-factor balancing and by-product generation. The previously developed Minimum Network Enrichment Analysis (MiNEA) method addresses these limitations by enumerating alternative minimal networks for given biomass building blocks and metabolic tasks. MiNEA facilitates a deeper understanding of metabolic task flexibility and context-specific metabolic routes by integrating condition-specific transcriptomics, proteomics, and metabolomics data. This approach offers significant improvements in the analysis of metabolic pathways, providing more comprehensive insights into cellular metabolism. RESULTS: Here, I present MiNEApy, a Python package reimplementation of MiNEA, which computes minimal networks and performs enrichment analysis. I demonstrate the application of MiNEApy on both a small-scale and a genome-scale model of the bacterium Escherichia coli, showcasing its ability to conduct minimal network enrichment analysis using minimal networks and context-specific data. AVAILABILITY AND IMPLEMENTATION: MiNEApy can be accessed at: https://github.com/vpandey-om/mineapy.

Metabolic Networks and Pathways↗

PubMed1992

A demonstration that breast cancer recurrence can be predicted by neural network analysis.

Neural Network Analysis, a form of artificial intelligence, was successfully used to predict the clinical outcome of node-positive breast cancer patients. A Neural Network was trained to predict clinical outcome using prognostic information from 1008 patients. During training, the network received as input information tumor hormone receptor status, DNA index and S-phase determination by flow cytometry, tumor size, number of axillary lymph nodes involved with tumor, and age of the patient, as well as length of clinical followup, relapse status, and time of relapse. The ability of the trained Network to determine relapse probability was then validated in a separate set of 960 patients. The Neural Network was as powerful as Cox Regression Modeling in identifying breast cancer patients at high and low risk for relapse.

Axilla↗

PubMed1991

Nuclear grading of breast carcinoma by image analysis. Classification by multivariate and neural network analysis.

The use of nuclear grade as a prognostic indicator for breast carcinoma has been limited by interobserver variability. Advances in image analysis and automated cell classification offer one approach to this problem. The authors used the CAS-100 (Cell Analysis System. Elmhurst, IL) system to measure and analyze nuclear morphometric and texture features of cytologic preparations from 35 breast carcinomas (well, moderate, and poorly differentiated) as well as benign lesions. Morphometric and Markovian texture feature data from breast cancer nuclei of various grades comprised a training set, which was then used to establish classification criteria by multivariate (Bayesian) analysis and to train a neural network system. Both systems were tested for the ability to classify the nuclear grade of individual nuclei. There was good agreement between computer classification and the grade assigned by human observer to individual nuclei using either Bayesian or neural network analysis. Thirty-one unknown cases, which were assigned an overall grade by an observer, were then analyzed by computer, and an overall grade assigned based on the grade of nucleus most frequently present. Using this method, both classification systems were able to assign a "correct" grade to low-grade lesions (approximately 70% correct) more often than to high-grade tumors (approximately 20%). Difficulty in computer assignment of high-grade tumors was explained by nuclear heterogeneity in these tumors (i.e., although the percentage of high-grade nuclei was increased compared with that of low-grade tumors, high-grade nuclei frequently did not predominate). The authors present this study to demonstrate the feasibility of using image analysis as an objective means of nuclear grading. Further studies will be needed to establish criteria for assigning overall nuclear grade based on computer analysis of imaging data.

Artificial Intelligence↗

PubMed1976

The application of network analysis to the study of branching patterns of large dendritic fields.

Network analysis of dendritic fields not only defines the topology and connectivity of segments of an arborescence, but offers a means of discovering how networks grow. An important theory has recently been formulated29 suggesting that dendritic branching patterns may be established by synaptogenic interaction of dendritic growth cones with growing axons. This thesis may be verified through network analysis since the theory predicts that growth at pendant vertices will predominate in dendritic networks, that dendritic growth will be directed into areas of maximal synaptogenic activity and that arc lengths will be inversely related, and the order of branching at vertices directly related, to the magnitude of the synaptogenic activity operating about growing dendritic terminals. The possibility of a preponderance of terminal growth may be detected by comparing the topologies in an observed dendritic network with those of a series of hypothetical growth models. This paper provides the frequency table for models grown by monochotomous, dichotomous and trichotomous branching on random pendant vertices and random arcs for large networks in which 'set theory' contingencies are included. The paper also describes a method of calculating branching probabilities from the measurement of segment lengths, which is a means of testing the last mentioned prediction of the synaptogenic theory of denddritic growth. The method of network analysis is then discussed in relation to probable dendritic growth patterns, the constancy of segment lengths and the interaction of extrinsic and intrinsic factors in determining branching probabilities.

Dendrites↗

PubMed1975

Application of network analysis to the study of the branching patterns of dendritic fields.

The technique of network analysis has been used to define the connectivity and growth of networks generated by monochotomous, dichotomous, and trichotomous branching. The number of distinct topologic branching patterns exhibited by networks with a given number of pendant arcs is defined mathematically; when all types are represented, a complete pendant arc series is formed. The frequency of occurrence of topologic types in these series is unique for a given hypothesis of growth. The growth of the small dendritic arrays such as the basal dendritic fields of neocortical pyramids may be studied by comparing the actual frequency of topologic types with those computed according to given hypotheses. For larger dendritic networks such as those of Purkinje cells in the cerebellum it is only practicable to use the topologic types formed by the peripheral parts of the tree as a basis for comparison. Individual dendritic segments can be ordered sequentially to define their hierarchical arrangement; the frequency of orders in a given network always forms an inverse geometric series. The ratio between orders is called the "bifurcation ratio," and the relationship in a given large series between adjacent orders becomes stabilized to a fixed or "established" bifurcation ratio at the periphery of the tree only. This "established ratio" characterizes the pattern of growth of the network. In the proximal part of the tree the ratio between adjacent orders is unstable and accounts for the variability of the overall bifurcation ratio exhibited by different networks with the same fundamental growth pattern and for the deviation of the overall from the established bifurcation ratio. For a given size of network the overall bifurcation ratio may be similar regardless of the mode of growth. It is concluded that the precise definition of branching structures afforded by network analysis makes this technique well suited for the study of the connectivity, growth, and morphology of dendritic trees.

Computers↗

PubMed2025

Integrated proteomic network analysis reveals PTPRC as a central hub protein orchestrating co-expression modules and metabolic dysregulation in renal carcinoma: PTPRC protein molecular action.

The occurrence of renal carcinoma is closely related to a variety of molecular mechanisms and metabolic disorders. PTPRC (protein tyrosine phosphatase receptor C), as an important regulatory protein, was studied to reveal the role of PTPRC in renal carcinoma through comprehensive proteomic network analysis, especially its core position in the coordination of co-expression modules and metabolic disorders. This study was the first to download and process multiple publicly available renal cancer transcriptome data to conduct differential gene expression analysis across datasets. Functional enrichment and disease ontology analysis were performed on the transcriptome of renal cancer, and weighted gene co-expression network (WGCNA) was constructed. The results showed that comprehensive principal component analysis revealed significant differences in the transcriptome of renal cancer, and functional annotation revealed specific pathways associated with renal cancer. WGCNA analysis identified tumor-associated co-expression modules, while multi-omics analysis further identified core regulatory networks including PTPRC. As a central hub protein, PTPRC plays an important coordinating role in the co-expression module and metabolic dysregulation of renal carcinoma. This discovery provides a new perspective for understanding the molecular mechanism of kidney cancer.

Humans↗

PubMed1992

Network analysis of intermediary metabolism using linear optimization. I. Development of mathematical formalism.

Analysis of metabolic networks using linear optimization theory allows one to quantify and understand the limitations imposed on the cell by its metabolic stoichiometry, and to understand how the flux through each pathway influences the overall behavior of metabolism. A stoichiometric matrix accounting for the major pathways involved in energy and mass transformations in the cell was used in our analysis. The auxiliary parameters of linear optimization, the so-called shadow prices, identify the intermediates and cofactors that cause the growth to be limited on each nutrient. This formalism was used to examine how well the cell balances its needs for carbon, nitrogen, and energy during growth on different substrates. The relative values of glucose and glutamine as nutrients were compared by varying the ratio of rates of glucose to glutamine uptakes, and calculating the maximum growth rate. The optimum value of this ratio is between 2-7, similar to experimentally observed ratios. The theoretical maximum growth rate was calculated for growth on each amino acid, and the amino acids catabolized directly to glutamate were found to be the optimal nutrients. The importance of each reaction in the network can be examined both by selectively limiting the flux through the reaction, and by the value of the reduced cost for that reaction. Some reactions, such as malic enzyme and glutamate dehydrogenase, may be inhibited or deleted with little or no adverse effect on the calculated cell growth rate.

Adenosine Triphosphate↗

PubMed2025

Exploring prognostic genes in the immune microenvironment of acute myeloid leukemia via weighted gene co-expression network analysis.

BACKGROUND: Acute myeloid leukemia (AML) is a heterogeneous blood cancer that arises from transformed myeloid precursor cells in a compromised bone marrow microenvironment. This environment is essential for AML initiation, progression, and relapse. Alongside oncogenic changes in hematopoietic cells, immunological dysregulation also contributes to leukemogenesis. The present study is aimed to identify prognostic genes in stromal and immune cells associated with AML using the weighted gene co-expression network analysis (WGCNA). METHODS: Gene expression profiles were retrieved from The Cancer Genome Atlas database, and immune and stromal cell scores were calculated using the ESTIMATE (Estimation of STromal and Immune cells in MAlignant Tumor tissues using Expression data) method. These scores helped identify differentially expressed genes (DEGs), which were then used to create gene clusters through WGCNA. To explore the functions of genes linked to AML subtypes, Gene Ontology and Kyoto Encyclopedia of Genes and Genomes enrichment analyses were performed. A protein-protein interaction network was developed to identify hub genes. The top 18 hub genes were identified using the cytoHubba plug-in in Cytoscape software, and survival analysis was conducted with the Gene Expression Profiling Interactive Analysis 2 online tool. RESULTS: A total of 1097 DEGs were identified, with 601 being upregulated and 496 downregulated. WGCNA analysis indicated that the gray module, comprising 165 genes, had the strongest association with AML subtypes (Cor > 0.3; P < .05). Gene Ontology enrichment analysis demonstrated that the 18 identified hub genes were predominantly associated with neutrophil activation, immune response, secretory granule membrane, and pattern recognition receptor activity. Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis revealed that the DEGs were mainly involved in pathways related to phagosome, lysosome, tuberculosis, leishmaniasis, and neutrophil extracellular trap formation. Kaplan-Meier survival analysis of the top 18 hub genes indicated that ITGAM, IL10, and CD163 were significantly correlated with survival outcomes in AML. CONCLUSION: Key stromal and immune-related genes influencing AML patient outcomes were identified, highlighting their potential as therapeutic targets. These discoveries provide deeper insights into the molecular mechanisms driving AML pathogenesis and subtype differentiation.

Leukemia, Myeloid, Acute↗

PubMed2024

Comparison of the antibiotic resistance mechanisms in a gram-positive and a gram-negative bacterium by gene networks analysis.

Nowadays, the emergence of some microbial species resistant to antibiotics, both gram-positive and gram-negative bacteria, is due to changes in molecular activities, biological processes and their cellular structure in order to survive. The aim of the gene network analysis for the drug-resistant Enterococcus faecium as gram-positive and Salmonella Typhimurium as gram-negative bacteria was to gain insights into the important interactions between hub genes involved in key molecular pathways associated with cellular adaptations and the comparison of survival mechanisms of these two bacteria exposed to ciprofloxacin. To identify the gene clusters and hub genes, the gene networks in drug-resistant E. faecium and S. Typhimurium were analyzed using Cytoscape. Subsequently, the putative regulatory elements were found by examining the promoter regions of the hub genes and their gene ontology (GO) was determined. In addition, the interaction between milRNAs and up-regulated genes was predicted. RcsC and D920_01853 have been identified as the most important of the hub genes in S. Typhimurium and E. faecium, respectively. The enrichment analysis of hub genes revealed the importance of efflux pumps, and different enzymatic and binding activities in both bacteria. However, E. faecium specifically increases phospholipid biosynthesis and isopentenyl diphosphate biosynthesis, whereas S. Typhimurium focuses on phosphorelay signal transduction, transcriptional regulation, and protein autophosphorylation. The similarities in the GO findings of the promoters suggest common pathways for survival and basic physiological functions of both bacteria, including peptidoglycan production, glucose transport and cellular homeostasis. The genes with the most interactions with milRNAs include dpiB, rcsC and kdpD in S. Typhimurium and EFAU004_01228, EFAU004_02016 and EFAU004_00870 in E. faecium, respectively. The results showed that gram-positive and gram-negative bacteria have different mechanisms to survive under antibiotic stress. By deciphering their intricate adaptations, we can develop more effective therapeutic approaches and combat the challenges posed by multidrug-resistant bacteria.

Anti-Bacterial Agents↗

PubMed2024

Decoding the genetic landscape of allergic rhinitis: a comprehensive network analysis revealing key genes and potential therapeutic targets.

BACKGROUND: Allergic Rhinitis (AR), an inflammatory affliction impacting the upper respiratory tract, has been registering a substantial surge in incidence across the globe. METHODS: We embarked on examination of differentially expressed genes (DEGs) and the Weighted Gene Co-Expression Network Analysis (WGCNA). With this armory of genes identified, we engaged the tools of Gene Ontology (GO) and the Kyoto Encyclopedia of Genes and Genomes (KEGG). Our study continued with the establishment of a protein-protein interaction (PPI) network and the application of LASSO regression. Finally, we leveraged a docking model to elucidate potential drug-gene interactions involving these key genes. RESULTS: Through WGCNA and different express genes screening, PPI network was performed, identifying top 20 key genes, including CD44, CD69, CD274. LASSO regression identified three independent factors, STARD5, CST1, and CHAC1, that were significantly associated with AR. A predictive model was developed with an AUC value over 0.75. Also, 105 potential therapeutic agents were discovered, including Fluorouracil, Cyclophosphamide, Doxorubicin, and Hydrocortisone, offering promising therapeutic strategies for AR. CONCLUSION: By fuzing DEGs with key genes derived from WGCNA, this study has illuminated a comprehensive network of gene interactions involved in the pathogenesis of AR, paving the way for future biomarker and therapeutic target discovery in AR.

Humans↗

PubMed1975

Network analysis of dendritic fields of pyramidal cells in neocortex and Purkinje cells in the cerebellum of the rat.

The connectivity within the dendritic array of Purkinje cells in the cerebellum and pyramidal cells of the neocortex of the rat, stained by the Golgi-Cox method, has been quantified by the method of network analysis. Connectivity was characterized either by applying the system of Strahler ordering, which assigns a relative order of magnitude to each branch of the arborescence or by the identification of unique topological branching patterns within the tree. The former method has been used to define the entire dendritic array of the Purkinje cell and the apical system of neocortical pyramids. It has been shown that the relation between the numbers of branches of successive Strahler order in Purkinje cells form an inverse geometric series in which the highest order is unity and the ratio between successive orders approximates to 3. On the other hand, the apical dendrites of neocortical pyramids exhibit two bifurcation ratios, i.e. a ratio of 3 between low orders and a ratio of 4 between higher orders. A computer simulation technique was used to generate networks of a size comparable with the Purkinje cell networks and grown according to two hypotheses namely, a 'terminal growth model' in which additional segments were added randomly to the terminal branches only and a 'segmental growth model' in which additional segments were added randomly to any branch within the array including terminal branches. Subsequent ordering of the simulated trees revealed that the relation between the numbers of successive orders for networks generated according to the 'segmental model' tended towards an inverse geometric series with a ratio of 4 and that generated according to the 'terminal model' tended towards a ratio of 3. This result showed that the dendritic tree of Purkinje cells grow in a manner indistinguishable from a system adding branches to random terminal segments and that neocortical apical dendrites add their collateral branches to random segments of the apical shaft but that the collateral branches themselves grow by random terminal branching. The possibility that such conclusions may be influenced by loss of branches incurred by either a failure of impregnation, by sectioning, or by environmental influences was investigated by means of a computer technique...

Animals↗

PubMed2026

Nested co-expression network analysis identifies compact gene clusters in a black box.

MOTIVATION: Digital analysis of biological systems requires methods capable of identifying both broad and nested gene modules reflecting complex biological processes. Existing transcriptomic methods often miss compact gene sets corresponding to subprocesses in specialized cell types, limiting insights into functional heterogeneity. RESULTS: We present Nested-WGCNA, a two-stage unsupervised network analysis algorithm designed to identify coarse-grained and fine-grained gene modules. Applied to bulk RNA-Seq data, Nested-WGCNA reveals stable modules reproducible across datasets. When validated against scRNA-Seq data, these modules correspond to both major and minor immune cell subtypes. Application to immunotherapy response datasets uncovers predictive and prognostic biomarkers, highlighting its utility in treatment stratification and biomarker discovery. AVAILABILITY: The NestedWGCNA source code and analysis pipeline are available on GitHub (https://github.com/ilyada/NestedWGCNA) and archived on Zenodo (https://doi.org/10.5281/zenodo.18959244).

Algorithms↗

PubMed1992

Network analysis of Korean health insurance policy-making process.

This study examines how the decision-making process evolved in Korea during the initial phases of introduction and implementation of National Health Insurance. This study analyses the official documents and interviews views made with government officials and related personnel. We used the method of network analysis and multidimensional scaling in order to demonstrate how the major participants in the decision-making process developed and changed under the contemporary political situations. In the pre-implementation stage around 1976, major concerns were concentrated around the issues of financial support for social insurance, the fee schedule and who ought to be covered first. The total number of participants of the health or health-related organization was 61, which included the President, the Minister of Health and Social Affairs, representatives of special interest groups, etc. In the actual implementation period of 1982, different issues were brought up by the major participants. The number of participants in this period declined to 44 with the deletion of 19 and with the addition of two newly formed health insurance organizations. By 1988, as the implementation reached its final decision period, disagreements were centered on progressive premium rating and the administration of National Health Insurance. The number of participants increased to 60 after the addition of 16 participants. The analysis of this paper may provide some insight for other countries which wish to establish National Health Insurance; as reference to the policy-making process, it may provide some suggestions for when to initiate and how to formulate National Health Insurance policies.

Humans↗

PubMed1991

Use of neural network analysis to classify electroencephalographic patterns against depth of midazolam sedation in intensive care unit patients.

The electroencephalographic (EEG) analog signal is complex and cannot easily be described by univariate variables. Clear visual changes in the EEG power spectrum can be present with little or no change in univariate variable values. A method that could produce a single value based on the total data available in the EEG power spectrum would be very useful in monitoring EEG changes. Neural network analysis is a technique that can take multiple inputs and produce a single output value using complicated processing patterns that require training to establish. We examined the usefulness of a series of neural network models to classify 63 EEG patterns against sedation level in 26 mechanically ventilated patients requiring midazolam for long-term sedation. During a stable period of sedation, a 4- to 60-minute period of EEG data was obtained concurrently with a sedation level from 1 (follows commands) to 7 (no or gag response to suctioning of the endotracheal tube). The EEG power spectrum was divided into equal frequency bands, and the log absolute powers in each of these bands were used as inputs for a series of neural network models. The output target was the sedation level associated with each set of EEG data. Networks were trained on a subset of EEG power/sedation score data pairs, and the ability to classify the remaining data pairs was tested. Using a t-test comparison with a random set of sedation levels, we found that trained neural network models classified EEG patterns against sedation level successfully (p less than 0.001).(ABSTRACT TRUNCATED AT 250 WORDS)

Adolescent↗

PubMed2026

Crosstalk mediators implicated in the Stevens-Johnson Syndrome through gene regulatory network analysis.

Stevens-Johnson syndrome (SJS) is a rare and severe mucocutaneous disorder often triggered by medications or infections. Our previous research identified that four key genes, Ikzf1, Ptger3, Mavs, and Tlr3 are involved in SJS susceptibility and the conjunctival epithelial innate immune response, demonstrating their role in regulating interferon-stimulated genes. However, the interplay among these regulatory factors remains unclear. This study aimed to elucidate the crosstalk mechanisms between the pathways regulated by these four genes in conjunctival epithelial cells. We constructed a comprehensive gene regulatory network using transcriptomic data from murine conjunctival epithelial cells under 16 distinct conditions, including polyI:C stimulation across wild-type, knockout, and transgenic backgrounds for the key genes. A targeted network analysis systematically identified numerous candidate genes mediating the crosstalk between the regulatory pathways initiated by Ikzf1, Ptger3, Mavs, and Tlr3. The identified candidates suggest the involvement of diverse signaling pathways previously unlinked to SJS pathology. Our findings suggest that the pathogenesis of SJS may arise not from the dysfunction of isolated genes but from the disruption of a balance maintained by intricate pathway crosstalk.

Animals↗

PubMed2026

Functional and Nutritional Potential of Chickpea Protein Hydrolysates: A Systematic Review and Plant-protein Network Analysis.

Chickpea is a protein-rich legume increasingly explored as a substrate for functional plant-based ingredients. Chickpea protein hydrolysates (CPHs) and chickpea-derived peptides (CPs), obtained through enzymatic hydrolysis or simulated gastrointestinal digestion, may provide technological and biological properties while supporting the valorization of chickpea fractions and by-products. This review integrates a network analysis of title-abstract terms from 5,728 unique Scopus and PubMed records on plant protein hydrolysates with a systematic review of 72 studies focused on CPH production, peptide characterization, bioactivity, and translational gaps. The evidence indicates that CPHs and CPs show promising antioxidant, antihypertensive, antidiabetic, anti-inflammatory, lipid-lowering, immunomodulatory, antimicrobial, and anticancer-related activities, mainly supported by biochemical assays, cell models, and animal studies. However, heterogeneous hydrolysis protocols, incomplete peptide characterization, inconsistent bioactivity methods, limited scale-up evidence, and the absence of human intervention trials restrict translation. Future studies should prioritize standardized protocols, mechanistic validation, bioavailability, sensory and regulatory assessment, food-matrix validation, and clinical trials.

Cicer↗

PubMed2026

Genetic dissection of cardiac iron regulation using transcriptome network analysis and systems genetics in BXD mice.

Cardiac iron homeostasis is essential for myocardial energy metabolism and contractile function, yet the genetic and molecular mechanisms governing iron levels within the heart remain poorly understood. We used a systems genetics approach to dissect the transcriptional regulation of cardiac iron homeostasis. Myocardial iron level varies substantially across BXD strains (40-112 μg/g) and is under heritable genetic control (H2 = 0.38). Elevated cardiac iron is associated with reduced ventricular mass, increased ventricular ectopy, and prolonged atrioventricular conduction in the BXD population. Weighted gene co-expression network analysis of the BXD heart transcriptome identified a co-expression module that was significantly and negatively correlated with cardiac iron levels in both young and old BXD mice and enriched for pathways related to metabolic regulation, cyclic AMP (cAMP) signaling, circadian entrainment, and cardiovascular physiology. The module showed substantial overlap with a curated cardiac iron gene set, and cross-species enrichment analysis confirmed its conservation in human cardiomyopathy differentially expressed genes (enrichment ratio = 1.49; false discovery rate [FDR] = 0.0342). Quantitative trait locus (QTL) mapping of the first principal component of the overlapping module iron genes (n = 38), corroborated by individual gene mapping, identified trans-eQTL hotspots on multiple chromosomes, implicating Fcho2, Gcc2, and Rmdn1 as candidate upstream regulators operating through sequential steps of intracellular iron trafficking. Together, these findings establish a systems-level map of cardiac iron gene regulation, identify candidate genetic regulators, and provide a molecular framework linking disruption of iron-related transcriptional networks to structural and electrical cardiac dysfunction with implications for iron-related heart diseases.

BXD mouse population↗

PubMed1992

Neuronal network analysis of serum electrophoresis.

AIMS: To advise a system of neuronal networks which can classify the densitometric patterns of serum electrophoresis. METHODS: Digitised data containing 83 normal and 132 pathological serum protein electrophoresis patterns were presented to four neuronal networks containing 1900 neurons. Network 1 evaluates the integrated values of the albumin, alpha 1, alpha 2, beta and gamma fractions together with total protein (Biuret method). Networks 2, 3, and 4 analyse the shape of the albumin, beta and gamma fractions. To increase the sensitivity for the detection of monoclonal gammopathies a Fourier transformation was applied to the beta and gamma fractions. RESULTS: After a learning period of 20 minutes (back-propagation learning algorithm) the system was tested with a set of electrophoresis patterns comprising 446 routinely collected samples. It differentiated between physiological and pathological curves with a sensitivity of 97.5% and a specificity of 98.8%, with 86% correct diagnoses. All monoclonal gammopathies were recognised by the Fourier detector. CONCLUSIONS: Neuronal networks could be useful for certain medical uses. Unlike rule based systems, neuronal networks do not have to be programmed but have the capacity to "learn" quickly.

Blood Protein Electrophoresis↗