SEARCH · PubMed Health

Results for “generative artificial intelligence”

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 199 records · Page 11Linked to original sources

PubMed2003

Artificial neural networks for species identification by taxonomists.

This paper is a study of the value of applying artificial neural networks (ANNs), specifically a multilayer perceptron (MLP), to identification of higher plants using morphological characters collected by conventional means. A practical methodology is thus demonstrated to enable botanical or zoological taxonomists to use ANNs as advisory tools for identification purposes. A comparison is made between the ability of the neural network and that of traditional methods for plant identification by means of a case study in the flowering plant genus Lithops N.E. Brown (Aizoaceae). In particular, a comparison is made with taxonomic keys generated by means of the DELTA system. The ANN is found to perform better than the DELTA key generator, for conditions where the available data is limited, and species relatively difficult to distinguish.

Aizoaceae↗

PubMed2005

The next generation of literature analysis: integration of genomic analysis into text mining.

Text-mining systems are indispensable tools to reduce the increasing flux of information in scientific literature to topics pertinent to a particular interest in focus. Most of the scientific literature is published as unstructured free text, complicating the development of data processing tools, which rely on structured information. To overcome the problems of free text analysis, structured, hand-curated information derived from literature is integrated in text-mining systems to improve precision and recall. In this paper several text-mining approaches are reviewed and the next step in development of text-mining systems, which is based on a concept of multiple lines of evidence, is described: results from literature analysis are combined with evidence from experiments and genome analysis to improve the accuracy of results and to generate additional knowledge beyond what is known solely from literature.

Abstracting and Indexing↗

PubMed2004

Doctor Dunsel.

Repressingly, a robot scientist can generate functional genomic hypothesis and carry out experiments, increasing the working scientist's feelings of redundancy.

Artificial Intelligence↗

PubMed1991

Using comparative clinical information to understand practice patterns and affect organizational change.

The University Hospital Consortium is collecting clinical, administrative and financial data from its members to develop a Clinical Information Network. The value of this collective data lies in how comparative information about peer hospitals and physicians in the same specialty can be used to influence practice. The raw data from each hospital is analyzed, classified, normalized and stored in a data repository which is easily accessible. This data becomes information when it is presented in a variety of ways, and is supported by a knowledge-base of health care rules. The "drilling down" technique to progressive levels of detail serves the needs of all levels in the organization--executives, managers, and analysts. The system combines the power of a mainframe for the data repository with the ease of use of a PC-based workstation. With an open-ended approach, the users can ask a variety of questions of the data, as well as perform statistical analysis, create graphical presentations and generate explanations of the analysis techniques.

Artificial Intelligence↗

PubMed2026

Immunopeptidomics-guided cancer vaccine design: Advances, challenges, and emerging opportunities.

Selecting clinically relevant tumor antigens remains a major challenge in the development of therapeutic cancer vaccines. Although computational approaches have considerably improved neoantigen prediction, many candidate epitopes identified in silico are not ultimately presented on the tumor cell surface. The emergence of immunopeptidomics has provided direct access to naturally processed HLA-associated peptides and has offered new opportunities for antigen discovery. Increasing evidence has shown that information derived from the immunopeptidome becomes considerably more informative when interpreted alongside genomic, transcriptomic, and proteomic data. This integrative view has broadened the spectrum of targetable antigens and has also revealed important limitations related to peptide abundance, HLA diversity, tumor heterogeneity, and the imperfect relationship between antigen presentation and immunogenicity. These issues have renewed interest in multi-antigen vaccine strategies designed to better reflect the complexity of tumor antigen landscapes. Advances in bioinformatics and artificial intelligence are facilitating the interpretation of increasingly complex datasets and are beginning to support more systematic approaches to antigen prioritization. In this review, we discuss how immunopeptidomics is contributing to next-generation cancer vaccine development, summarize the major translational challenges, and highlight emerging concepts that may improve the clinical applicability of immunopeptidomics-guided immunotherapy.

Cancer immunotherapy↗

PubMed1985

Electronic databases.

Electronic databases corresponding to most of the world's currently published literature and many other types of information are publicly available through online systems. Scientific databases that give references for publications are numerous and widely used; scientific numeric databases that are open to the public are far fewer and less used. Online retrieval systems are becoming easier to use as a result of the introduction of artificial intelligence techniques and user-friendly front ends and gateways. Issues related to electronic databases include public-private sector competition, transborder data flow, copyright, downloading, and the changing roles in database generation and processing.

Bibliographies as Topic↗

PubMed2026

Adenoviral Vectors in Gene Therapy: A Detailed Overview.

Adenoviral vectors (AdVs) represent one of the most extensively researched platforms in the realm of gene therapy, providing advantages such as high transduction efficiency, large transgene capacity, and broad tropism. This review provides a detailed and structured overview of AdVs, highlighting their biology, gene delivery mechanisms, clinical applications, and challenges limiting their broader therapeutic applicability. The study also explores recent progress in vector engineering, such as rare serotypes, capsid modifications, third-generation vectors, as well as strategies for immune modulation and toxicity reduction. AdVs are used in therapies for genetic disorders, oncology, and vaccinology, alongside innovations such as CRISPR-Cas9, nanotechnology, and artificial intelligence design. Nevertheless, persistent hurdles, including vector immunogenicity, hepatotoxicity, scalability, and the lack of durable expression, prevent widespread clinical use. This review consolidates current knowledge and presents a future perspective on how AdVs may evolve as powerful, adaptable, and precise tools in modern gene therapy. By contextualizing strengths and unresolved challenges, this work aims to give researchers and clinicians a balanced foundation for evaluating their future roles in translational medicine.

Humans↗

PubMedDate not supplied

Friends reunited? Evolutionary robotics and representational explanation.

Robotics as practiced within the artificial life community is no longer the bitter enemy of representational explanation in the way that it sometimes seemed to be in the heady, revolutionary days of the 1990s. This rapprochement is, however, fragile, because the field of evolutionary robotics continues to pose two important challenges to the idea that real-time intelligent action must or should be explained by appeal to inner representations. The first of these challenges, the threat from nontrivial causal spread, occurs when extra-neural factors account for the kind of adaptive richness and flexibility normally associated with representation-based control. The second, the threat from continuous reciprocal causation, occurs when the causal contributions made by the systemic components collectively responsible for behavior generation are massively context-sensitive and variable over time. I argue that while the threat from nontrivial causal spread can be resisted, the threat from continuous reciprocal causation provides a stern test for our representational intuitions.

Artificial Intelligence↗

PubMed1996

Similarity and complementarity of molecular shapes: applicability of a topological analysis approach.

Developments based on a topological analysis approach of electron density maps are presented and applied to two different fields: the interpretation of electron density maps of proteins and the description of shape complementarity between a cyclodextrin host and a guest molecule. A global representation of the electron density distribution, through the location, identification and linkage of its critical points (points where the gradient of the density vanishes, i.e., peaks and passes), is generated using the program ORCRIT. On one hand, the interpretation of protein electron density maps is based on similarity evaluations between graphs of critical points and known structures. So far, the method has been applied to 3 A resolution maps for the recognition of secondary structure motifs using a procedure relevant to expert systems in artificial intelligence. Satisfying matches between critical point graphs and their corresponding protein structure depict the ability of the topological analysis to catch the essential secondary structural features in electron density maps. On the other hand, mapping the accessible volume of a host molecule is achieved by representing the peaks as ellipsoids with axes related to local curvature of the electron density function. Related energies of the interacting species can also be estimated. A qualitative comparison is made between the results generated by the topological analysis and energy values obtained by conventional molecular mechanics calculations. A positive comparison and a close complementarity between cyclodextrin and ligands shows that the topological analysis method gives a good representation of the electron density function.

Computer Simulation↗

PubMed1992

Hemodynamic and oxygen transport patterns for outcome prediction, therapeutic goals, and clinical algorithms to improve outcome. Feasibility of artificial intelligence to customize algorithms.

A generalized decision tree or clinical algorithm for treatment of high-risk elective surgical patients was developed from a physiologic model based on empirical data. First, a large data bank was used to do the following: (1) describe temporal hemodynamic and oxygen transport patterns that interrelate cardiac, pulmonary, and tissue perfusion functions in survivors and nonsurvivors; (2) define optimal therapeutic goals based on the supranormal oxygen transport values of high-risk postoperative survivors; (3) compare the relative effectiveness of alternative therapies in a wide variety of clinical and physiologic conditions; and (4) to develop criteria for titration of therapy to the endpoints of the supranormal optimal goals using cardiac index (CI), oxygen delivery (DO2), and oxygen consumption (VO2) as proxy outcome measures. Second, a general purpose algorithm was generated from these data and tested in preoperatively randomized clinical trials of high-risk surgical patients. Improved outcome was demonstrated with this generalized algorithm. The concept that the supranormal values represent compensations that have survival value has been corroborated by several other groups. We now propose a unique approach to refine the generalized algorithm to develop customized algorithms and individualized decision analysis for each patient's unique problems. The present article describes a preliminary evaluation of the feasibility of artificial intelligence techniques to accomplish individualized algorithms that may further improve patient care and outcome.

Algorithms↗

PubMed1991

A machine learning approach to computer-aided molecular design.

Preliminary results of a machine learning application concerning computer-aided molecular design applied to drug discovery are presented. The artificial intelligence techniques of machine learning use a sample of active and inactive compounds, which is viewed as a set of positive and negative examples, to allow the induction of a molecular model characterizing the interaction between the compounds and a target molecule. The algorithm is based on a twofold phase. In the first one--the specialization step--the program identifies a number of active/inactive pairs of compounds which appear to be the most useful in order to make the learning process as effective as possible and generates a dictionary of molecular fragments, deemed to be responsible for the activity of the compounds. In the second phase--the generalization step--the fragments thus generated are combined and generalized in order to select the most plausible hypothesis with respect to the sample of compounds. A knowledge base concerning physical and chemical properties is utilized during the inductive process.

Amino Acid Sequence↗

PubMed2005

Improved classification of mass spectrometry database search results using newer machine learning approaches.

Manual analysis of mass spectrometry data is a current bottleneck in high throughput proteomics. In particular, the need to manually validate the results of mass spectrometry database searching algorithms can be prohibitively time-consuming. Development of software tools that attempt to quantify the confidence in the assignment of a protein or peptide identity to a mass spectrum is an area of active interest. We sought to extend work in this area by investigating the potential of recent machine learning algorithms to improve the accuracy of these approaches and as a flexible framework for accommodating new data features. Specifically we demonstrated the ability of boosting and random forest approaches to improve the discrimination of true hits from false positive identifications in the results of mass spectrometry database search engines compared with thresholding and other machine learning approaches. We accommodated additional attributes obtainable from database search results, including a factor addressing proton mobility. Performance was evaluated using publically available electrospray data and a new collection of MALDI data generated from purified human reference proteins.

Amino Acid Sequence↗

PubMed1998

A computational system for modelling flexible protein-protein and protein-DNA docking.

A computational system is described that predicts the structure of protein/protein and protein/DNA complexes starting from unbound coordinate sets. The approach is (i) a global search with rigid-body docking for complexes with shape complementarity and favourable electrostatics; (ii) use of distance constraints from experimental (or predicted) knowledge of critical residues; (iii) use of pair potential to screen docked complexes and (iv) refinement and further screening by protein-side chain optimisation and interfacial energy minimisation. The system has been applied to model ten protein/protein and eight protein-repressor/DNA (steps i to iii only) complexes. In general a few complexes, one of which is close to the true structure, can be generated.

Algorithms↗

PubMed2025

Challenges and Opportunities in Analyzing Cancer-Associated Microbiomes.

The study of cancer-associated microbiomes has gained significant attention in recent years, spurred by advances in high-throughput sequencing and metagenomic analysis. Microbiome research holds promise for identifying noninvasive biomarkers and possibly new paradigms for cancer treatment. In this review, we explore the key computational challenges and opportunities in analyzing cancer-associated microbiomes (in tumor/normal tissues and other body sites, e.g., gut, oral, and skin), focusing on sequencing-driven strategies and associated considerations for taxonomic and functional characterization. The discussion covers the strengths and limitations of current analysis tools for identifying contamination, determining compositional bias, and resolving species and strains, as well as the statistical, metabolic, and network inferences that are essential to uncover host-microbiome interactions. Several key considerations are required to guide the choice of databases used for metagenomic analysis in such studies. Recent advances in spatial and single-cell technologies have provided insights into cancer-associated microbiomes, and Artificial Intelligence-driven protein function prediction might enable rapid advances in this field. Finally, we provide a perspective on how the field can evolve to manage the ever-growing size of datasets and generate robust and testable hypotheses. This article is part of a special series: Driving Cancer Discoveries with Computational Research, Data Science, and Machine Learning/AI .

Humans↗

PubMed2025

Real-world deployment of a fine-tuned pathology foundation model for lung cancer biomarker detection.

Artificial intelligence models using digital histopathology slides stained with hematoxylin and eosin offer promising, tissue-preserving diagnostic tools for patients with cancer. Despite their advantages, their clinical utility in real-world settings remains unproven. Assessing EGFR mutations in lung adenocarcinoma demands rapid, accurate and cost-effective tests that preserve tissue for genomic sequencing. PCR-based assays provide rapid results but with reduced accuracy compared with next-generation sequencing and require additional tissue. Computational biomarkers leveraging modern foundation models can address these limitations. Here we assembled a large international clinical dataset of digital lung adenocarcinoma slides (N = 8,461) to develop a computational EGFR biomarker. Our model fine-tunes an open-source foundation model, improving task-specific performance with out-of-center generalization and clinical-grade accuracy on primary and metastatic specimens (mean area under the curve: internal 0.847, external 0.870). To evaluate real-world clinical translation, we conducted a prospective silent trial of the biomarker on primary samples, achieving an area under the curve of 0.890. The artificial-intelligence-assisted workflow reduced the number of rapid molecular tests needed by up to 43% while maintaining the current clinical standard performance. Our retrospective and prospective analyses demonstrate the real-world clinical utility of a computational pathology biomarker.

Humans↗

PubMed2004

Why neural networks should not be used for HIV-1 protease cleavage site prediction.

UNLABELLED: Several papers have been published where nonlinear machine learning algorithms, e.g. artificial neural networks, support vector machines and decision trees, have been used to model the specificity of the HIV-1 protease and extract specificity rules. We show that the dataset used in these studies is linearly separable and that it is a misuse of nonlinear classifiers to apply them to this problem. The best solution on this dataset is achieved using a linear classifier like the simple perceptron or the linear support vector machine, and it is straightforward to extract rules from these linear models. We identify key residues in peptides that are efficiently cleaved by the HIV-1 protease and list the most prominent rules, relating them to experimental results for the HIV-1 protease. MOTIVATION: Understanding HIV-1 protease specificity is important when designing HIV inhibitors and several different machine learning algorithms have been applied to the problem. However, little progress has been made in understanding the specificity because nonlinear and overly complex models have been used. RESULTS: We show that the problem is much easier than what has previously been reported and that linear classifiers like the simple perceptron or linear support vector machines are at least as good predictors as nonlinear algorithms. We also show how sets of specificity rules can be generated from the resulting linear classifiers. AVAILABILITY: The datasets used are available at http://www.hh.se/staff/bioinf/

Algorithms↗

PubMed2006

OR/AND neurons and the development of interpretable logic models.

In this paper, we are concerned with the concept of fuzzy logic networks and logic-based data analysis realized within this framework. The networks under discussion are homogeneous architectures comprising of OR/AND neurons originally introduced by Hirota and Pedrycz. Being treated here as generic processing units, OR/AND neurons are neurofuzzy constructs that exhibit well-defined logic characteristics and are endowed with a high level of parametric flexibility and come with significant interpretation abilities. The composite logic nature of the logic neurons becomes instrumental in covering a broad spectrum of logic dependencies whose character spread in-between between those being captured by plain and and or logic descriptors (connectives). From the functional standpoint, the developed network realizes a logic approximation of multidimensional mappings between unit hypercubes, that is transformations from [0, 1]n to [0, 1]m. The way in which the structure of the network has been formed is highly modular and becomes reflective of a general concept of decomposition of logic expressions and Boolean functions (as being commonly encountered in two-valued logic). In essence, given a collection of input variables, selected is their subset and transformed into new composite variable, which in turn is used in the consecutive module of the network. These intermediate synthetic variables are the result of the successive problem (mapping) decomposition. The development of the network is realized through genetic optimization. This helps address important issues of structural optimization (where we are concerned with a selection of a subset of variables and their allocation within the network) and reaching a global minimum when carrying out an extensive parametric optimization (adjustments of the connections of the neurons). The paper offers a comprehensive and user-interactive design procedure including a simple pruning mechanism whose intention is to enhance the interpretability of the network while reducing its size. The experimental studies comprise of three parts. First, we demonstrate the performance of the network on Boolean data (that leads to some useful comparative observations considering a wealth of optimization tools available in two-valued logic and digital systems). Second, we discuss synthetic multivalued data that helps focus on the approximation abilities of the network. Finally, show the generation of logic expressions describing selected data sets coming from the machine learning repository.

Algorithms↗

PubMed2005

A method for generating precise temporal patterns of retinal spiking using prosthetic stimulation.

The goal of retinal prosthetic devices is to generate meaningful visual information in patients that have lost outer retinal function. To accomplish this, these devices should generate patterns of ganglion cell activity that closely resemble the spatial and temporal components of those patterns that are normally elicited by light. Here, we developed a stimulus paradigm that generates precise temporal patterns of activity in retinal ganglion cells, including those patterns normally generated by light. Electrical stimulus pulses (> or =1-ms duration) elicited activity in neurons distal to the ganglion cells; this resulted in ganglion cell spiking that could last as long as 100 ms. However, short pulses, <0.15 ms, elicited only a single spike within 0.7 ms of the leading edge of the pulse. Trains of these short pulses elicited one spike per pulse at frequencies < or =250 Hz. Patterns of short electrical pulses (derived from normal light elicited spike patterns) were delivered to ganglion cells and generated spike patterns that replicated the normal light patterns. Finally, we found that one spike per pulse was elicited over almost a 2.5:1 range of stimulus amplitudes. Thus a common stimulus amplitude could accommodate a 2.5:1 range of activation thresholds, e.g., caused by differences arising from cell biophysical properties or from variations in electrode-to-cell distance arising when a multielectrode array is placed on the retina. This stimulus paradigm can generate the temporal resolution required for a prosthetic device.

Action Potentials↗