SEARCH · PubMed Health

Results for “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 73 records · Page 4Linked to original sources

PubMed2026

A safety-centric perspective on innovation and risk in the use of artificial intelligence in genomics.

Adopting a safety-centric approach, this article explores how generative artificial intelligence (AI), and more specifically, foundation models for biological sequences, can exacerbate data quality issues, technical biases, and dual-use potential, particularly in critical applications such as clinical genetics, precision medicine, and pathogen engineering. This work centres on how misuse risks emerge throughout the innovation pipeline and how these intersect with the growing accessibility of generative genomic models. Particular attention is given to dual-use governance and infrastructure hardening in sequence analysis workflows. The work aims to provide scientists, regulators, and policymakers with a toolkit to discuss beneficial innovation in genomic AI while maintaining robust safeguards against harm and misuse.

Genomics↗

PubMed2026

Multimodal artificial intelligence and machine learning in oncology: from data integration to precision cancer care.

Cancer remains a major global health burden, with approximately 20 million new cases and 9.7 million cancer-related deaths reported globally in 2022. While advances in radiological imaging, molecular profiling, and clinical data have enhanced the interpretation of disease progression, the availability of multiple such modalities still does not meet the needs of a large patient population. This narrative review focuses on the role of multimodal artificial intelligence and machine learning in bridging the gap in interpreting heterogeneous modalities to improve risk prediction, prognostic assessment, and treatment decision-making in precision oncology. Multimodal frameworks such as Pathomic Fusion illustrate how complementary histopathological and genomic information can be integrated for cancer diagnosis and prognostic modeling. Multimodal models have demonstrated potential in virtual biopsy, cancer screening, prognostic prediction, radiotherapy planning, intraoperative guidance, and clinical-trial design using digital twins and synthetic control arms. The major limitations of incorporating multimodal artificial intelligence and machine learning in oncology include data heterogeneity, demographic or institutional biases, and reproducibility challenges that hinder translation. Accordingly, appropriate data-governance strategies, fairness audits, and privacy-preserving approaches such as federated learning should be considered where appropriate. Future progress will depend on the development of standardized benchmarking datasets, robust external validation, seamless integration with electronic health records and picture archiving and communication systems, and the implementation of explainable, secure, and clinically validated multimodal artificial intelligence frameworks that support precision oncology in routine clinical practice.

deep learning↗

PubMed2026

Artificial intelligence in genitourinary oncology: publication trends and systematic review.

OBJECTIVE: To conduct an analysis of publication trends and a systematic review of randomized controlled trials (RCTs) to characterize the current state of artificial intelligence (AI) use in genitourinary (GU) oncology, as AI has emerged as a transformative tool in healthcare with potential applications in diagnostics, treatment planning, and prognostication. METHODS: We searched the Medical Literature Analysis and Retrieval System Online (MEDLINE), Excerpta Medica dataBASE (EMBASE; Ovid), and Cumulative Index to Nursing and Allied Health Literature (CINAHL) Ultimate for studies related to AI and GU oncology, excluding non-English papers, non-human studies, review articles, and articles using AI solely for manuscript writing. Publication trends were analysed from 2013 to 2023 and categorized by study design and cancer type. RCTs were evaluated through systematic review using Covidence (Veritas Health Innovation Ltd, Melbourne, Victoria, Australia) for screening and data extraction. Two reviewers independently assessed all studies, with risk of bias (RoB) evaluated using the Cochrane RoB 2.0 tool. RESULTS: Of 2409 articles identified, 1220 met inclusion criteria. These included 962 retrospective articles, 175 prospective studies, 79 studies with combined retrospective/prospective methods, and four RCTs. Studies most commonly addressed prostate (n = 923), renal (n = 274), and urothelial (n = 194) cancers. Publications grew from 14 in 2013 to 362 in 2023, with substantial acceleration in 2019. Four RCTs were identified - one in urothelial cancer and three in prostate cancer. Two RCTs evaluated AI-based diagnostics, demonstrating improved performance over conventional methods; the remaining two RCTs evaluated AI in prognostication and treatment planning, showing improved gains in imaging interpretation and operational efficiency. RoB varied across studies, primarily related to randomisation and deviations from intended interventions. CONCLUSIONS: Artificial intelligence research in GU oncology has grown, although high-level evidence from RCTs remains limited. Existing trials underscore AI's promise in diagnostics, prognostication, and treatment planning, and the rapidly evolving nature of this field warrants continued prospective investigation.

Humans↗

PubMed2026

Emulated trial of artificial intelligence use and subsequent depressive outcomes in a survey of US adults.

BACKGROUND: Generative artificial intelligence (AI) use has been suggested to have adverse mental health consequences but a causal relationship has not been examined. OBJECTIVE: To simulate a randomised controlled trial of AI use in a work, school or personal context by applying target trial emulation to multiple waves of data from a nationally representative survey. METHODS: We conducted a target trial emulation using non-probability survey data from three waves of a nationally representative survey conducted between 18 June 2024 and 8 January 2025. Participants aged ≥18 years reported generative AI use frequency at baseline. High-frequency use was defined as multiple times per week or more. The primary outcome was depressive symptom severity measured using the Patient Health Questionnaire 9-item (PHQ-9) at follow-up. Generalised causal forests assessed heterogeneity of treatment effects. FINDINGS: Among 19 099 participants assessed at baseline, 2862 (15.0%) reported AI use at least multiple times per week. A subset of 3109 (16.3%) returned for follow-up. In the primary weighted analysis, high-frequency use was not significantly associated with change in PHQ-9 score at follow-up (mean difference -0.18, 95% CI -0.94 to 0.59; p=0.65). Multiple sensitivity analyses using alternate outcome definitions also did not identify significant causal effects. Generalised causal forests yielded no significant evidence of heterogeneity of effect (p=0.81). CONCLUSIONS: In an emulated randomised trial among US adults, generative AI use was not associated with subsequent depressive symptoms. This result does not support the premise that AI use causes greater depressive symptoms, although adverse outcomes among vulnerable individuals cannot be excluded. CLINICAL IMPLICATIONS: AI use is unlikely to cause increased depressive symptoms among most US adults. Continued monitoring should clarify potential risks among vulnerable populations.

Humans↗

PubMedDate not supplied

Artificial intelligence techniques for cancer treatment planning.

An artificial intelligence system, NEWCHEM, for the development of new oncology therapies is described. This system takes into account the most recent advances in molecular and cellular biology and in cell-drug interaction, and aims to guide experimentation in the design of new optimal protocols. Further work is being carried out, aimed to embody in the system all the basic knowledge of biology, physiopathology and pharmacology, to reason qualitatively from first principles so as to be able to suggest cancer therapies.

Animals↗

PubMed2001

Artificial intelligence: a new approach for prescription and monitoring of hemodialysis therapy.

The effect of dialysis on patients is conventionally predicted using a formal mathematical model. This approach requires many assumptions of the processes involved, and validation of these may be difficult. The validity of dialysis urea modeling using a formal mathematical model has been challenged. Artificial intelligence using neural networks (NNs) has been used to solve complex problems without needing a mathematical model or an understanding of the mechanisms involved. In this study, we applied an NN model to study and predict concentrations of urea during a hemodialysis session. We measured blood concentrations of urea, patient weight, and total urea removal by direct dialysate quantification (DDQ) at 30-minute intervals during the session (in 15 chronic hemodialysis patients). The NN model was trained to recognize the evolution of measured urea concentrations and was subsequently able to predict hemodialysis session time needed to reach a target solute removal index (SRI) in patients not previously studied by the NN model (in another 15 chronic hemodialysis patients). Comparing results of the NN model with the DDQ model, the prediction error was 10.9%, with a not significant difference between predicted total urea nitrogen (UN) removal and measured UN removal by DDQ. NN model predictions of time showed a not significant difference with actual intervals needed to reach the same SRI level at the same patient conditions, except for the prediction of SRI at the first 30-minute interval, which showed a significant difference (P = 0.001). This indicates the sensitivity of the NN model to what is called patient clearance time; the prediction error was 8.3%. From our results, we conclude that artificial intelligence applications in urea kinetics can give an idea of intradialysis profiling according to individual clinical needs. In theory, this approach can be extended easily to other solutes, making the NN model a step forward to achieving artificial-intelligent dialysis control.

Adult↗

PubMed1996

Artificial intelligence in pediatrics: important clinical signs in newborn syndromes.

New methods are warranted in the field of syndromology. This study is an exploration into whether an artificial intelligence method (ID3) could provide a new angle for approaching syndromes. Diagnosing syndromes in the newborn is difficult. The accepted approach is to look for individual clinical signs that add up to a syndrome diagnosis. Of all possible clinical signs, one would want to extract the signs with the strongest predictive power. I used the ID3 algorithm to extract predictive clinical signs from a catalogue of syndromes (Birth Defects Encyclopedia Online; BDEO). Using information from BDEO, files of randomly generated "patients" were created. The signs consistently high in the identification tree were long philtrum, short palpebral fissures, low-set ears, and hepatosplenomegaly. The program used featured a crude "expert system" based on the ID3 algorithm. When using one-half of the data set as a training set and the other half as a testbed, a correct classification rate of 92.1-98.1% was attained. When the ID3 expert system was tested against cases from a clinical database (Pictures of Standard Syndromes and Undiagnosed Malformations), the correct classification rate was less than 20%. This may not necessarily reflect faults with the ID3 approach, but possibly biases in the clinical database. In syndromology no "criterion standards" exist that can confirm a diagnosis. The statistical method of cluster analysis does not require prior knowledge of diagnoses and will make a tree of syndromes based upon clinical signs. A cluster analysis was performed as a validity check to provide a tree for comparison with the ID3 tree. There was a reasonable degree of agreement between the two. Applying artificial intelligence methods to this field highlights problems with basic assumptions and philosophical aspects of syndrome diagnosis.

Algorithms↗

PubMed2026

Automated CEAP Classification of Venous Duplex Reports Using Multimodal Artificial Intelligence.

OBJECTIVE: To develop and internally validate a prototype multimodal artificial intelligence system for automated CEAP (Clinical, Etiological, Anatomical and Pathophysiological) classification of venous duplex ultrasound (VDUS) reports, integrating natural language processing of free-text components with computer vision analysis of hand-drawn anatomical diagrams. METHODS: Single centre retrospective observational study using routinely collected clinical data. One thousand consecutive venous duplex ultrasound reports from Cambridge University Hospitals NHS Foundation Trust, UK (July 2024 - May 2025) were labelled according to the CEAP classification, excluding the Etiological component, which could not be reliably determined from duplex reports alone. Transfer learning was applied using ClinicalBERT for text and MobileNetV3 for diagrammatic data. Clinical classes were predicted from request line text. Text- and image-based pathophysiological models were developed for four anatomical territories (Great Saphenous Vein, Small Saphenous Vein, Deep system, Perforators), combined using late fusion with probability averaging. RESULTS: The clinical CEAP model achieved accuracy of 0.91, macro-F1 of 0.82, and macro-AUC of 0.98. Pathophysiological prediction varied, with text models broadly outperforming image models. Fusion yielded heterogeneous benefits, improving SSV performance but reducing Deep system accuracy. The performance of the final pathophysiological CEAP fusion models varied across anatomical territories: accuracy ranged from 0.70-0.92 and macro-AUC from 0.80-0.92. CONCLUSION: This study demonstrates the feasibility of automated CEAP classification from VDUS reports. Despite class imbalance affecting minority class predictions, the strong discriminatory performance validates this multimodal ML model for extracting clinically meaningful information from real-world data. This approach offers potential, pending external validation, to streamline vascular services through automated triage and guideline-compliant decision making.

Artificial intelligence↗

PubMed2003

Estimation of urban runoff and water quality using remote sensing and artificial intelligence.

Water quality and quantity of runoff are strongly dependent on the landuse and landcover (LULC) criteria. In this study, we developed a more improved parameter estimation procedure for the environmental model using remote sensing (RS) and artificial intelligence (AI) techniques. Landsat TM multi-band (7bands) and Korea Multi-Purpose Satellite (KOMPSAT) panchromatic data were selected for input data processing. We employed two kinds of artificial intelligence techniques, RBF-NN (radial-basis-function neural network) and ANN (artificial neural network), to classify LULC of the study area. A bootstrap resampling method, a statistical technique, was employed to generate the confidence intervals and distribution of the unit load. SWMM was used to simulate the urban runoff and water quality and applied to the study watershed. The condition of urban flow and non-point contaminations was simulated with rainfall-runoff and measured water quality data. The estimated total runoff, peak time, and pollutant generation varied considerably according to the classification accuracy and percentile unit load applied. The proposed procedure would efficiently be applied to water quality and runoff simulation in a rapidly changing urban area.

Artificial Intelligence↗

PubMed2026

How Following Medical Artificial Intelligence Advice Can Mitigate Malpractice Liability: Cross-National Insights from a Randomized Trial.

Artificial intelligence (AI) increasingly influences clinical decision-making, yet its recommendations may diverge from standard care. Although malpractice concerns are thought to discourage physicians from following AI advice, experimental evidence from the United States suggests the opposite: lay jurors are more likely to hold physicians liable when they reject AI recommendations. Whether this pattern extends to systems in which court-appointed experts, not lay jurors, determine liability remains unknown. Methods: To examine how physicians and laypeople in expert-based and lay-juror legal systems evaluate physicians' acceptance or rejection of AI recommendations, particularly when those recommendations deviate from standard care, we designed a randomized vignette study: a 2 × 2 factorial design varying the AI recommendation (standard vs. nonstandard care) and a fictional physician's decision (accept vs. reject). The study was conducted online in 2023 among nationally representative samples of U.S. and German adults and from 2023 to 2024 among German physicians. In total, 387 German physicians, 2291 U.S. adults, and 2283 German adults participated; those not completing the survey or failing attention checks were excluded per preregistered criteria. Participants were randomly assigned to 1 of 4 vignettes, varying the AI recommendation (standard vs. nonstandard care) and physician's decision (accept vs. reject). The reasonableness of the fictional physician's decision was measured, rated by participants on a Likert scale. Results: Analysis, following preregistered exclusion criteria, included 248 German physicians, 1202 U.S. adults, and 1358 German adults. Physicians accepting standard-care AI recommendations were rated more reasonable than those rejecting them (U.S. laypeople: t = 5.36; 95% CI, 0.45-0.97; P < 0.001; German physicians: t = 2.47; 95% CI, 0.14-1.30; P = 0.02; German laypeople: t = 4.14; 95% CI, 0.27-0.76; P < 0.001). Ratings of physicians accepting versus rejecting AI nonstandard-care recommendations were statistically equivalent. Equivalence was tested at an α-value of 0.05 using a two 1-sided tests procedure, reported with 90% CIs per standard convention (U.S. laypeople: t = -4.90; 90% CI, -0.1 to 0.36; P < 0.001; German physicians: t = -1.76; 90% CI, -0.12 to 0.67; P = 0.04; German laypeople: t = 5.35; 90% CI, -0.35 to 0.06; P < 0.001). Conclusion: Across the United States and Germany, samples representative of lay jurors and court-appointed experts viewed accepting standard-care AI advice as more reasonable, whereas accepting or rejecting nonstandard-care AI advice was judged similarly. Contrary to predictions, malpractice liability regimes do not necessarily pose a barrier to AI use in precision medicine.

Artificial Intelligence↗

PubMed2003

Abstraction and reformulation in artificial intelligence.

This paper contributes in two ways to the aims of this special issue on abstraction. The first is to show that there are compelling reasons motivating the use of abstraction in the purely computational realm of artificial intelligence. The second is to contribute to the overall discussion of the nature of abstraction by providing examples of the abstraction processes currently used in artificial intelligence. Although each type of abstraction is specific to a somewhat narrow context, it is hoped that collectively they illustrate the richness and variety of abstraction in its fullest sense.

Artificial Intelligence↗

PubMed1996

Artificial intelligence analysis of paraspinal power spectra.

OBJECTIVE: As an aid to discrimination of sufferers with back pain an artificial intelligence neural network was constructed to differentiate paraspinal power spectra. DESIGN: Clinical investigation using surface electromyography. METHOD: The surface electromyogram power spectra from 60 subjects, 33 non-back-pain sufferers and 27 chronic back pain sufferers were used to construct a back propagation neural network that was then tested. Subjects were placed on a test frame in 30 degrees of lumbar forward flexion. An isometric load of two-thirds maximum voluntary contraction was held constant for 30 s whilst surface electromyograms were recorded at the level of the L(4-5). Paraspinal power spectra were calculated and loaded into the input layer of a three-layer back propagation network. The neural network classified the spectra into normal or back pain type. RESULTS: The back propagation neural was shown to have satisfactory convergence with a specificity of 79% and a sensitivity of 80%. CONCLUSIONS: Artificial intelligence neural networks appear to be a useful method of differentiating paraspinal power spectra in back-pain sufferers.

Journal Article↗

PubMed2004

The role of networks and artificial intelligence in nanotechnology design and analysis.

Techniques with their origins in artificial intelligence have had a great impact on many areas of biomedicine. Expert-based systems have been used to develop computer-assisted decision aids. Neural networks have been used extensively in disease classification and more recently in many bioinformatics applications including genomics and drug design. Network theory in general has proved useful in modeling all aspects of biomedicine from healthcare organizational structure to biochemical pathways. These methods show promise in applications involving nanotechnology both in the design phase and in interpretation of system functioning.

Artificial Intelligence↗

PubMed2001

Application of artificial intelligence in audiology.

In this paper, machine learning methods based on artificial intelligence theory are applied to the computer-aided decision making of some otoneurological diseases, for example Ménière's disease. Three methods explored are decision trees, genetic algorithms and neural networks. By using such a machine learning method, the decision-making program is trained with a representative training set of cases and tested with another set. The machine learning methods are useful also for our otoneurological expert system, One, which is based on a pattern recognition approach. The methods are able to differentiate most of the cases tested between the six diseases included, provided that a sufficiently large training set is available.

Algorithms↗

PubMed1991

Artificial intelligence methods for theory representation and hypothesis formation.

This article describes artificial intelligence methods for representing theories in molecular biology, and for improving the predictive power of these theories using experimental data. A program called GENSIM provides a framework for representing theories that includes descriptions of classes of biological objects (genes, enzymes, etc.), and processes that specify potential interactions among these objects (such as enzymatic reactions). GENSIM can employ a theory specified within this framework to predict the outcomes of biological experiments. A program called HYPGENE comes into play when the observed outcome of an experiment does not match the outcome predicted by GENSIM. HYPGENE works backward from the error in GENSIMs prediction to postulate changes to both the theory embodied by GENSIM, and the presumed initial conditions of the experiment. I view HYPGENEs hypothesis generation task as a design problem, and I have adapted AI methods developed for design and planning to this task. These techniques were developed in conjunction with an in-depth study of the discovery of the gene regulation mechanism of attenuation in the E. coli tryptophan operon. Both GENSIM and HYPGENE have been tested on sample problems from the history of attenuation, and produced many of the same solutions as biologists did.

Artificial Intelligence↗

PubMed1994

Artificial intelligence within the chemical laboratory.

Various techniques within the area of artificial intelligence such as expert systems and neural networks may play a role during the problem-solving processes within the clinical biochemical laboratory. Neural network analysis provides a non-algorithmic approach to information processing, which results in the ability of the computer to form associations and to recognize patterns or classes among data. It belongs to the machine learning techniques which also include probabilistic techniques such as discriminant function analysis and logistic regression and information theoretical techniques. These techniques may be used to extract knowledge from example patients to optimize decision limits and identify clinically important laboratory quantities. An expert system may be defined as a computer program that can give advice in a well-defined area of expertise and is able to explain its reasoning. Declarative knowledge consists of statements about logical or empirical relationships between things. Expert systems typically separate declarative knowledge residing in a knowledge base from the inference engine: an algorithm that dynamically directs and controls the system when it searches its knowledge base. A tool is an expert system without a knowledge base. The developer of an expert system uses a tool by entering knowledge into the system. Many, if not the majority of problems encountered at the laboratory level are procedural. A problem is procedural if it is possible to write up a step-by-step description of the expert's work or if it can be represented by a decision tree. To solve problems of this type only small expert system tools and/or conventional programming are required.(ABSTRACT TRUNCATED AT 250 WORDS)

Artificial Intelligence↗

PubMed2006

A review on integration of artificial intelligence into water quality modelling.

With the development of computing technology, numerical models are often employed to simulate flow and water quality processes in coastal environments. However, the emphasis has conventionally been placed on algorithmic procedures to solve specific problems. These numerical models, being insufficiently user-friendly, lack knowledge transfers in model interpretation. This results in significant constraints on model uses and large gaps between model developers and practitioners. It is a difficult task for novice application users to select an appropriate numerical model. It is desirable to incorporate the existing heuristic knowledge about model manipulation and to furnish intelligent manipulation of calibration parameters. The advancement in artificial intelligence (AI) during the past decade rendered it possible to integrate the technologies into numerical modelling systems in order to bridge the gaps. The objective of this paper is to review the current state-of-the-art of the integration of AI into water quality modelling. Algorithms and methods studied include knowledge-based system, genetic algorithm, artificial neural network, and fuzzy inference system. These techniques can contribute to the integrated model in different aspects and may not be mutually exclusive to one another. Some future directions for further development and their potentials are explored and presented.

Algorithms↗

PubMed1994

Artificial intelligence versus logistic regression statistical modelling to predict cardiac complications after noncardiac surgery.

The traditional approach to developing models predictive of cardiac events has been to perform logistic regression (LR) analysis on a variety of potential predictors. An alternative to use an artificial intelligence system called a neural network (NN) which simulates biological intelligence. To evaluate the potential applicability of the latter method, we compared the ability of LR and NN techniques to predict cardiac events after noncardiac surgery. A total of 200 patients (training group) underwent cardiac risk assessment before major noncardiac surgery using 17 clinical parameters and 7 quantitative indices based on dipyridamole-thallium imaging. There were 21 post-operative myocardial infarctions and/or cardiac deaths. Data from the training group were used to develop two predictive models: one based on backward stepwise LR multivariate statistical analysis and the other one using a neural network. Both models were then validated on a second group of 160 consecutive patients also referred for preoperative risk stratification (validation group). The NN consisted of 14 input, 29 hidden, and 1 output neurons and used a back-propagation algorithm (learning rate 0.2, training tolerance 0.5, sigmoid transfer function). The sensitivity, specificity, positive and negative predictive accuracies for the prediction of postoperative events in the validation group of 160 patients were, respectively, 67% (6/9), 82% (124/151), 18% (6/33), and 98% (124/127) for LR, and 67% (6/9), 96% (145/151), 50% (6/12), and 98% (145/148) for the NN, with a difference in specificity which attained statistical significance (p < 0.01). Artificial intelligence may provide a useful alternative to conventional LR statistical analysis for the purpose of preoperative cardiac risk assessment.

Artificial Intelligence↗