SEARCH · Search PubMed

Results for “AI-assisted literature review”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

SourceResearch connectionFrom yearThrough year

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

1,207 records · Page 3Linked to original sources

PubMed2026

Clinical applications of digital twin technology in In Vitro Fertilisation.

BACKGROUND: Digital twin technology, originating from aerospace and manufacturing industries, has emerged as a transformative tool in healthcare. In vitro fertilisation (IVF) faces persistent challenges including suboptimal embryo selection, unpredictable treatment outcomes, and limited personalisation of protocols. Despite advances in assisted reproductive technology, existing literature exhibits fragmentation: artificial intelligence applications in embryo selection, ovarian stimulation, and endometrial assessment have been developed independently without systematic integration into comprehensive treatment frameworks. Digital twin technology offers unprecedented opportunities to create virtual replicas of biological systems, enabling real-time monitoring, predictive modelling, and personalised treatment strategies. AIM: This narrative review aims to critically examine the current applications of digital twin technology in IVF, evaluate its potential benefits and limitations, synthesize existing evidence into an integrative conceptual model, and identify future directions for implementation in reproductive medicine. METHOD: A comprehensive narrative review was conducted using PubMed, Scopus, Web of Science, and IEEE Xplore databases. A narrative review approach was selected over systematic review to accommodate the heterogeneity of evidence types in this emerging field, including theoretical frameworks, simulation studies, and proof-of-concept implementations that would be excluded from systematic reviews. Search terms included "digital twin," "IVF," "in vitro fertilisation," "assisted reproductive technology," "embryo selection," and "predictive modelling." Studies published between 2015 and 2025 were included, focusing on original research articles, systematic reviews, and proof-of-concept studies describing digital twin applications in reproductive medicine. RESULTS: Digital twin technology in IVF demonstrates significant potential across multiple domains including embryo development simulation, ovarian response prediction, endometrial receptivity modelling, and personalised stimulation protocols. Current applications integrate artificial intelligence, machine learning algorithms, time-lapse imaging, and omics data to create comprehensive virtual models. Early evidence suggests improvements in embryo selection accuracy, ovarian response prediction, and treatment protocol optimization, though large-scale randomized controlled trials remain limited. Implementation challenges include data integration complexity, computational requirements, regulatory considerations, and validation requirements. CONCLUSION: Digital twin technology represents a paradigm shift in IVF practice, offering personalised, predictive, and precision medicine approaches. This review synthesizes existing evidence to propose an integrative conceptual model for digital twin implementation across the IVF treatment spectrum, identifies critical knowledge gaps, and establishes research priorities to advance clinical translation. Despite current limitations, continued advancement promises improved success rates and patient outcomes.

Humans↗

PubMed2026

Fundamentals of pacemakers ECG interpretation - part 2.

BACKGROUND: Modern pacemakers incorporate arrhythmia-response algorithms, ventricular pacing minimization protocols, and safety mechanisms that generate ECG patterns indistinguishable from pathological AV block, sensing malfunction, or device-mediated tachycardia. Failure to recognize these algorithm-driven signatures leads to unnecessary interventions, misdiagnosis, and inappropriate device reprogramming. This manuscript is the second in a two-part series on pacemaker ECG interpretation. METHODS: We conducted a narrative review of peer-reviewed literature and device-specific documentation on algorithm-driven ECG behavior, synthesizing evidence across arrhythmia recognition, upper rate physiology, ventricular pacing minimization, mode switching, safety mechanisms, and hysteresis algorithms. RESULTS: Pacemaker-mediated tachycardia produces regular paced wide-complex tachycardia locked at the upper tracking rate, initiated by any event with retrograde VA conduction. Ventricular tachycardia is identified by QRS morphology diverging from the known paced pattern, absent pacing spikes, and AV dissociation. Upper rate Wenckebach behavior mimics Mobitz type I AV block; 2:1 upper rate response mimics second-degree AV block. Ventricular pacing minimization algorithms produce isolated nonconducted P waves and prolonged AV intervals that simulate pathological conduction disease. Mode switching causes abrupt rate drops misidentified as output failure. Ventricular safety pacing generates a conspicuously short, fixed AV interval. Three discrete pacing artifacts reflect AV-sequential cardiac resynchronization therapy (CRT), ventricular safety pacing in CRT, or His-bundle pacing with backup RV output. Rate and AV hysteresis produce pauses and wandering AV intervals mimicking oversensing or Wenckebach periodicity. CONCLUSIONS: Recognizing algorithm-driven ECG patterns requires knowledge of device timing intervals and refractory periods, which lets clinicians distinguish programmed behavior from true malfunction or cardiac arrhythmia.

Humans↗

PubMed2026

Evaluation of a cornea-specialized large language model for diagnostic and management accuracy in complex corneal cases.

PURPOSE: To evaluate whether a cornea-specialized large language model (LLM) enhanced with retrieval-augmented generation (RAG) improves clinicians' diagnostic and management accuracy in complex corneal cases compared to a general-purpose GPT-4o model and unaided clinician performance. METHODS: This prospective, randomized, masked evaluation study involved three cornea trainees who each independently reviewed 39 real-world corneal cases under three experimental conditions: unaided, GPT-4o-assisted, and assisted by a cornea-specialized GPT-4o model. The cornea-specialized model was constructed by embedding over 200 publicly available Wikipedia articles into GPT-4o's RAG framework. Participants provided open-ended diagnoses and selected the next-step management options (multiple choice). They were allowed up to three GPT-4o queries per case, and the AI-assisted arms were randomized to minimize bias. Accuracy for both tasks was compared against expert reference standards using McNemar's test. RESULTS: Diagnostic accuracy was 48.7%, 20.5%, and 38.5% unaided, improving to 69.2%, 46.2%, and 59.0% with general GPT-4o (p<0.04). The cornea-specialized GPT-4o further improved accuracy to 71.8%, 48.7%, and 74.4%, with improvements over unaided performance for all clinicians (p<0.01). For next-step decisions, unaided accuracy was 76.9%, 87.2%, and 59.0%. With the specialized model, Ophthalmologist 3 improved to 71.8% (p<0.05), Ophthalmologist 1 remained high at 82.1%, and Ophthalmologist 2 declined to 64.1% (p<0.05). CONCLUSIONS: A cornea-specialized LLM enhanced with RAG improved diagnostic accuracy in complex corneal cases, particularly among clinicians with lower baseline performance. Effects on management accuracy were inconsistent. Future studies should explore the use of open-ended management tasks and examine whether smaller, curated retrieval corpora yield better model performance.

Humans↗

PubMed2026

Comparative evaluation of molecular technologies for the identification of prevalent non-tuberculous mycobacteria in pulmonary infections: a systematic review and meta-analysis.

BACKGROUND: The increasing prevalence of non-tuberculous mycobacteria pulmonary disease (NTM PD) is a burden to public health. Successful management of NTM PD critically depends on accurate species identification and reliable drug susceptibility testing to guide appropriate antibiotic therapy. Emerging molecular technologies offer rapid diagnostic solutions compared to conventional methods, but their performance varies. This study aims to provide a comprehensive evaluation of current molecular techniques for NTM identification and to present a global antibiotic resistance profile. METHODS: A systematic literature search was conducted in PubMed and Web of Science for studies published between 2005 and 2024. Studies applying molecular methods for NTM identification and resistance detection in humans were included. Data on study characteristics, diagnostic methods, sample types, sample sizes, identification sensitivity, and drug susceptibility results were extracted. Meta-analysis was performed using R with the meta4diag package. The quality of included studies was assessed using the QUADAS-2 tool. RESULTS: The analysis included 49 studies on NTM identification and 33 studies on antibiotic resistance. For species identification, all evaluated molecular technologies (MALDI-TOF MS, PCR-based methods, Sequencing, DNA chip, and DNA strip) demonstrated high pooled sensitivities (>0.92). Subgroup analysis revealed that sample type significantly affected performance for MALDI-TOF MS. Preliminary analysis of antibiotic resistance rates revealed varying patterns. For slowly growing mycobacteria, a significantly high Ethambutol resistance rate was observed in M. avium (69.20%). Among rapidly growing mycobacteria, resistance to Imipenem was notable (54.22%), and Clarithromycin resistance varied significantly within the Mycobacterium abscessus complex. CONCLUSION: Emerging molecular technologies have revolutionized the methodology for NTM identification with excellent performance. However, their performance can be influenced by sample type, particularly for MALDI-TOF MS. The alarming and heterogeneous antibiotic resistance patterns also highlight the critical need for rapid and accurate species identification and drug susceptibility testing to inform effective therapeutic strategies. Key messagesMolecular technologies demonstrate high accuracy for NTM identification.Antibiotic resistance is a serious concern with variations among NTM species and subspecies.Rapid and accurate species identification and drug susceptibility testing are crucial for guiding effective clinical management of NTM PD.

Humans↗

PubMed2026

The mighty microproteins: from versatile cellular regulators to precision medicine therapeutics.

Microproteins, are tiny proteins encoded by small open reading frame (sORF), translation of these non-canonical open reading frames (ncORFs) has been implicated in diverse biological processes and diseases. This review summarizes recent developments in the discovery, biogenesis, and functional characterization of microproteins, and their involvement in various disease, with special focus on their roles in cancer, cardiovascular, metabolic, neurodegenerative and immune-related disorders. We emphasize the regulation of key cellular pathways by microproteins, including mitochondrial homeostasis, apoptosis, metabolic reprogramming, and immune signaling, all of which affect disease initiation and progression. Emerging evidence also supports their potential as disease biomarkers and therapeutic candidates for precision medicine. Finally, the review critically discusses the current challenges including discrepancies in microprotein annotation, the limitations of ribosome profiling and proteogenomic approaches, the gap between computationally predicted and experimentally validated microproteins, and the need for rigorous orthogonal validation by means of CRISPR-based genome editing, ribosome release assays, mutational analysis, high-resolution mass spectrometry, and functional studies. Finally, we review recent development of AI-assisted ORF prediction, single-cell translatomics, spatial proteomics, and integrated multi-omics as emerging technologies reshaping. Microprotein discovery and functional annotation. Finally, we discuss the translational potential of microproteins and highlight the remaining challenges to clinical application, including peptide stability, pharmacokinetics, tissue-specific delivery, immunogenicity, and the need for rigorous preclinical and clinical validation. Together, this review provides an updated and critical overview of the rapidly evolving microprotein field and highlights future research priorities for translating these molecules into clinically useful biomarkers and precision therapeutics.

Microproteins↗

PubMed2026

AI-enabled viral genomics: from virus discovery to host prediction and emerging variant forecasting.

The rapid expansion of metagenomic sequencing has generated vast repositories of viral sequence data that far outpace our capacity to interpret them using conventional approaches. Highly divergent sequences, sparse functional annotation, and taxonomically uneven sampling present fundamental challenges for reference-dependent methods, which lose sensitivity precisely for novel and understudied viruses with high public health relevance. Artificial intelligence (AI) provides a new avenue to address these challenges by enabling predictive inference from viral genomes and proteins while reducing dependence on sequence similarity. In this Review, we discuss representative advances in AI for virus discovery, taxonomic classification and functional annotation, prediction of host range and zoonotic potential, and efforts toward forecasting emerging variants. These advances are transforming viral genomics from a largely descriptive discipline into one with increasing predictive capability. We also critically assess the major challenges that constrain current approaches, including the availability of high-quality and representative datasets, rigorous model evaluation, biological interpretability and responsible governance for increasingly capable AI models.

Artificial Intelligence↗

PubMed2026

Beyond predictive performance: A systematic review and critical methodological appraisal of AI/ML and conventional modelling strategies in breast, colorectal, and pancreatic Cancer.

BACKGROUND: Predictive modelling for cancer risk, treatment-related complications, and survival is central to precision oncology. Conventional logistic regression (LR) and Cox proportional hazards (CoxPH) regression remain widely used but are limited when modelling nonlinear interactions, high-dimensional imaging features, and multimodal clinical-metabolic predictors. Artificial intelligence (AI) and machine learning (ML) methods offer expanded capability through automated feature extraction, ensemble learning, and flexible survival modelling, but the evidence on when AI/ML adds value over conventional models across cancer sites and predictive tasks remains fragmented. OBJECTIVE: To systematically evaluate the methodological performance, validation strategies, and translational limitations of AI/ML models compared with conventional statistical models in published predictive-modelling studies for breast, colorectal, or pancreatic cancer. METHODS: PubMed, Scopus, and Web of Science were searched for studies published between January 2019 and March 2025. Two reviewers independently conducted title-and-abstract screening, full-text eligibility assessment, and PROBAST risk-of-bias assessment. Sixty-five studies (n = 907,567 participants) were narratively synthesised by cancer site, predictive task, model family, comparator, validation strategy, predictor modality, and calibration or explainability reporting. RESULTS: The 65 studies comprised breast cancer (n = 35), colorectal cancer (n = 21), and pancreatic cancer (n = 9). AI/ML superiority over LR and CoxPH was task- and data-dependent. CNN- and U-Net-based models predominated in imaging and body-composition tasks, tree-based ensembles consistently outperformed LR for tabular perioperative complication prediction, and CoxPH remained competitive, and in the largest pancreatic risk study, superior to XGBoost (C-index 0.802 vs 0.723) in well-structured datasets. PROBAST analysis-domain risk was moderate in 54 of 65 studies (83%), driven by limited external validation, sparse calibration reporting (11/65), and few decision-curve analyses (7/65). CONCLUSION: AI/ML adds the most methodological value in imaging-derived feature extraction and nonlinear perioperative prediction, while conventional regression remains preferable in large, structured datasets with linear predictors. Clinical translation requires standardised body-composition definitions, external validation, calibration assessment, decision-curve analysis, and explainability, in line with TRIPOD+AI and CLAIM standards.

Humans↗

PubMed2026

AI-driven snapshot hyperspectral imaging for on-line sorting systems in food industry: From real-time sensing to intelligent decision-making.

High-throughput food sorting requires rapid, non-destructive detection of external defects, foreign materials, and internal quality attributes in heterogeneous food matrices. Conventional scanning hyperspectral imaging may suffer from motion-induced spatial-spectral mismatches, whereas snapshot hyperspectral imaging (S-HSI) captures spectral images within a single integration time. However, its advantage is limited by trade-offs in resolution, signal-to-noise ratio (SNR), reconstruction uncertainty, and calibration stability, which are further amplified by variable tissue structure, surface reflection, moisture, and fat distribution in foods. This review critically examines artificial intelligence (AI)-driven S-HSI for on-line food sorting within a sensing-representation-decision-execution framework. Compact architectures are compared according to their physical constraints, food-sorting suitability, and ability to support mapping between spectral responses and physicochemical quality attributes. AI strategies are reviewed for spectral reconstruction, image restoration, spatial-spectral representation, band selection, uncertainty-aware decision-making, and edge implementation. AI can partially compensate for snapshot-specific limitations, but current evidence remains largely limited to laboratory or prototype studies. Future work should link system performance to food safety and quality outcomes by reporting throughput, decision latency, calibration drift, missed-detection risk, false-rejection cost, and closed-loop sorting success.

Hyperspectral Imaging↗

PubMed2026

A systematic approach to standardizing the visual appearance of endometriotic lesions for artificial intelligence recognition.

INTRODUCTION: Numerous studies have shown that the diagnostic performance and reproducibility of visual recognition of endometriosis during laparoscopy are poor. The use of artificial intelligence (AI) seems relevant for exhaustive lesion recognition. Standardization of the visual classification of lesions, in the form of an ontology, is an essential prerequisite to enable medical experts to annotate surgical data consistently and subsequently allow engineers to train and build an artificial intelligence tool for endometriosis recognition. MATERIAL AND METHODS: A systematic search was conducted in the MEDLINE (via PubMed), EMBASE, and the Cochrane Library databases up to May 2022, aiming to identify studies describing the laparoscopic visual appearance of superficial endometriosis, endometriomas, and deep infiltrating endometriosis. The accumulated data in the literature concerning the visual appearance of the different forms of endometriosis were used to create an ontology that could be used for artificial intelligence applications. RESULTS: Out of 932 articles screened, 35 studies were selected based on the inclusion criteria of human subjects with histologically confirmed endometriosis lesions visualized via laparoscopy. The selected studies were reviewed to develop a visual ontology of endometriosis lesions observed via laparoscopy. The lesions were categorized into 4 classes and further subdivided into 11 subclasses: superficial (black, red, white, or subtle), adhesions (dense or filmy), deep (obliteration, retraction, or deformation), and ovarian (endometrioma or chocolate fluid). The positive predictive value (PPV) varied across lesion types: black lesions (PPV 47%-97%), red lesions (PPV 33%-100%), white lesions (PPV 20%-81%), and ovarian endometriosis (PPV 42%-98%). Nonspecific lesions such as adhesions (PPV 16%-50%) and subtle superficial lesions (PPV 0%-67%) presented lower PPVs. Deep endometriosis lesions, often buried within organs, required indirect signs (obliteration, retraction, deformation) for identification. CONCLUSIONS: The visual ontology proposed in this systematic search could facilitate the detection and classification of endometriosis lesions using artificial intelligence. This study highlights the challenges of reaching a consensus on lesion recognition and classification in AI projects due to the diverse visual presentations of endometriosis.

Humans↗

PubMed2026

Artificial Intelligence Cannot Replace Peer Reviewers but May Help Editors Triage: A Comparative Analysis of a Large Language Model and Human Reviewer Recommendations at the American Journal of Sports Medicine.

BACKGROUND: The peer review system faces increasing strain from rising manuscript volumes, reviewer fatigue, and well-documented interreviewer disagreement. Large language models (LLMs) have shown potential to support the peer review process, but their ability to replicate editorial decisions at high-impact medical journals and their utility as manuscript screening tools remain unknown. PURPOSE: To compare the agreement between an LLM and the final editorial decision on manuscripts submitted to the American Journal of Sports Medicine and to evaluate the potential of LLMs as a manuscript screening tool. STUDY DESIGN: Cross-sectional agreement study. METHODS: Fifty-four manuscripts randomly selected from submissions to the American Journal of Sports Medicine (September 2024-October 2024) were reviewed by a locally deployed LLM (Ministral 3 14B; Mistral AI) using a standardized prompt. The artificial intelligence (AI) produced a categorical recommendation (reject, cascade, revision, or accept) and a numerical score (0-100) for each manuscript. Agreement with the final editorial decision was assessed by Cohen kappa (4-category model) for pooled human reviewers (n = 139 reviews) and the AI (n = 54). Screening performance was evaluated by positive predictive value (PPV), sensitivity, and specificity. RESULTS: Pooled human reviewers demonstrated fair agreement with the final decision (κ = 0.181 [P < .001]; 42.4% agreement), while the AI demonstrated slight, nonsignificant agreement (κ = 0.126 [P = .099]; 37.0% agreement). The AI recommended revision for 61.1% of manuscripts, of which 72.7% were ultimately rejected or cascaded, demonstrating systematic "revision bias." When the AI recommended rejection, 54.5% of those manuscripts were ultimately rejected and 27.3% were cascaded; when the AI recommended cascade, 50% were rejected and 50% were cascaded. However, when the AI recommended rejection or cascade (n = 21), 90.5% received a final decision of rejection or cascade (PPV, 90.5%; specificity, 81.8%). Manuscripts with an AI score <70 were rejected or cascaded 88.0% of the time (PPV, 88.0%). CONCLUSION: AI cannot replicate the nuanced judgment of human peer reviewers at a high-impact sports medicine journal. When AI recommended rejection or cascade, 90.5% of manuscripts received that final decision (descriptive PPV, 90.5%; 95% CI, 71.1%-97.3%), suggesting potential utility as an exploratory first-pass screening tool warranting further validation in larger cohorts. However, AI could not reliably distinguish manuscripts destined for outright rejection from those that would be cascaded to a sister journal-an important limitation for editorial triage applications.

Sports Medicine↗

PubMed2026

Nurse-led titration models of care for heart failure reduced ejection fraction: a systematic narrative review of characteristics, patient outcomes, and healthcare resource utilization.

AIMS: Nurse-led titration (NLT) models of care assist with delivery of guideline directed medical therapy for patients with heart failure with reduced ejection fraction (HFrEF). Effectiveness of NLT is established but there is limited information of characteristics of models, patient outcomes and healthcare resource utilization. To build upon the existing evidence by providing a systematic narrative review of the literature of NLT of medications for patients with HFrEF. This review syntheses characteristics of NLT models of care, patient outcomes and healthcare resource utilization. METHODS AND RESULTS: A systematic narrative literature review with systematic search strategy, identification of results, thematic analysis and narrative synthesis. A search was conducted from 2012 to 2025 in Medline, Cinahl complete, Embase and Cochrane. Sixteen studies of NLT models of care were identified from 1944 screened records. Characteristics of models of care were participation of nurses, multidisciplinary teams, follow-up and common features of service delivery. Patient outcomes of mortality were favourable for those that received NLT. There is some evidence of changes in healthcare resource utilization; studies in which the NLT groups received more HF nurse visits and greater HF medication use also reported reduced rehospitalizations. CONCLUSION: Findings reinforce the published benefits of NLT. Additional studies examining adverse events and quality-of-life outcomes are needed to strengthen the evidence base. Several studies suggest a shift in resource use with NLT, highlighting the need for an economic evaluation to inform a cost-effective model of care.

Humans↗

PubMed2026

Applications of quantum AI in brain disorder diagnosis: A systematic review.

BACKGROUND AND OBJECTIVE: Brain disorder diagnosis and prediction remain challenging because neuroimaging, electrophysiological, behavioral, and multimodal data are high-dimensional, noisy, heterogeneous, and limited by small clinical cohorts. This systematic review synthesised applications of quantum artificial intelligence (QAI) for brain disorder diagnosis, prediction, detection, and monitoring. METHODS: Following PRISMA guidelines, studies published from 2016 to 13 January 2026 were retrieved from Scopus, Web of Science, and IEEE Xplore. After screening, 36 studies met the eligibility criteria and were qualitatively analysed according to disorder category, data modality, QAI method, implementation setting, validation strategy, and performance. RESULTS: At the broader disease-group level, neurodegenerative disorders were the most frequently investigated, followed by mental health and psychiatric disorders. At the individual level, Parkinson's disease and schizophrenia were the leading applications, followed by depression, anxiety, Alzheimer's disease, and stress-related tasks. MRI-based modalities were the most frequently used data source, followed by multimodal data and EEG. Methodologically, primary QAI approaches were dominated by quantum neural and QDL architectures, followed by quantum-inspired optimization or feature-selection methods and quantum-kernel/conventional QML classifiers. Qiskit/IBM Quantum and PennyLane were the most frequently reported quantum software frameworks. However, most studies relied on simulators, classical quantum-inspired implementations, or unclear implementation settings, with limited real-hardware evaluation. CONCLUSIONS: QAI shows emerging potential for brain disorder analysis, particularly through hybrid quantum-classical learning, quantum neural architectures, quantum-kernel methods, and quantum-inspired optimization. Nevertheless, current evidence remains preliminary and requires larger datasets, subject-level and external validation, fair classical benchmarking, noise-resilient circuits, real quantum hardware evaluation, explainability, and clinical validation.

Humans↗

PubMed2026

Privacy, security, and reliability risks of artificial intelligence in healthcare: a systematic review of empirical evidence.

BACKGROUND: Artificial intelligence (AI) is increasingly integrated into healthcare information systems, supporting clinical decision-making, imaging analysis, and predictive modeling. While these applications offer operational and clinical benefits, they also introduce emerging risks to patient privacy, data security, and system reliability. OBJECTIVE: To systematically review empirical evidence on privacy breaches, security vulnerabilities, and misuse associated with AI applications in healthcare settings. METHODS: PubMed, Embase, Web of Science, Scopus, IEEE Xplore, and ACM Digital Library were searched for empirical studies published between January 2015 and November 2025 that evaluated AI use or misuse in clinical diagnosis, treatment, or decision-making. Two reviewers independently screened studies and extracted data using a standardized form. Findings were synthesized narratively due to heterogeneity in study designs, AI methods, and reported outcomes. RESULTS: Of 7,285 records identified through database searches and 205 through citation screening, 22 empirical studies met the inclusion criteria, spanning multiple clinical domains and data modalities, predominantly medical imaging applications. Five recurring threat categories were identified: patient re-identification, membership inference, unauthorized access and adversarial exploitation, input manipulation, and misuse or overinterpretation of AI outputs. Across studies, AI models were shown to encode latent biometric signals across diverse data types, limiting the effectiveness of traditional anonymization and synthetic data approaches. Adversarial attacks and input manipulation were also shown to compromise diagnostic performance and system integrity. CONCLUSION: This systematic review provides empirical evidence suggesting that contemporary AI systems in healthcare introduce privacy and security risks that may challenge traditional assumptions about data protection. These findings underscore the need for privacy- and security-by-design approaches and governance frameworks that address risks across the AI lifecycle.

Humans↗

PubMed2026

The Role of Artificial Intelligence for Intimate Partner Violence Prevention: A Systematic Review.

INTRODUCTION: Intimate partner violence (IPV), encompassing physical, sexual, emotional and economic abuse, remains a pervasive global health concern. Traditional prevention efforts face obstacles such as underreporting, delayed detection and limited personalised support. Emerging artificial intelligence (AI) approaches offer new opportunities to enhance IPV prevention. AIM: This systematic review maps and synthesises evidence on AI-driven tools in IPV prevention based on studies published between 2004 and 2024. METHODS: Following PRISMA 2020 guidelines and PROSPERO registration, we searched PubMed, Embase, CINAHL, PsycINFO, IEEE Xplore and Web of Science. Eligible studies explicitly evaluated AI technologies targeting IPV prediction, screening, intervention or support delivery. Study quality was appraised using the Mixed Methods Appraisal Tool (MMAT). RESULTS: Of 1304 records initially identified, 41 studies met eligibility criteria. AI applications ranged from machine learning (ML) for risk prediction and natural language processing (NLP) for IPV detection in clinical and social media data, to image analysis for forensic evaluation and chatbot-based support. Predictive modelling demonstrated strong discriminative performance, while NLP-based screening detected IPV with notable sensitivity. Chatbots showed feasibility and user acceptability, but evidence of their direct impact on reducing IPV incidence was limited, with one randomised controlled trial showing a modest reduction. Key challenges identified included algorithmic bias, data privacy risks and barriers to integration across health and social care systems. DISCUSSION: AI-informed interventions show promise for improving IPV detection, risk assessment, and scalable support, but questions remain about long-term effectiveness, ethical fairness, transparency and equitable implementation. Future interdisciplinary research should address these concerns to responsibly deploy AI in IPV prevention. RELEVANCE TO CLINICAL PRACTICE: The findings highlight the importance of trauma-informed, culturally responsive care and provider training in AI applications. Nurse-led innovation and policy advocacy will be crucial for safe, equitable integration of AI in IPV prevention.

Artificial Intelligence↗

PubMed2026

The Role of Artificial Intelligence Combined With Digital Cholangioscopy for Indeterminant and Malignant Biliary Strictures: A Systematic Review and Meta-analysis.

BACKGROUND: Current endoscopic retrograde cholangiopancreatography (ERCP) and cholangioscopic-based diagnostic sampling for indeterminant biliary strictures remain suboptimal. Artificial intelligence (AI)-based algorithms by means of computer vision in machine learning have been applied to cholangioscopy in an effort to improve diagnostic yield. The aim of this study was to perform a systematic review and meta-analysis to evaluate the diagnostic performance of AI-based diagnostic performance of AI-associated cholangioscopic diagnosis of indeterminant or malignant biliary strictures. METHODS: Individualized searches were developed in accordance with PRISMA and MOOSE guidelines, and meta-analysis according to Cochrane Diagnostic Test Accuracy working group methodology. A bivariate model was used to compute pooled sensitivity and specificity, likelihood ratio, diagnostic odds ratio, and summary receiver operating characteristics curve (SROC). RESULTS: Five studies (n=675 lesions; a total of 2,685,674 cholangioscopic images) were included. All but one study analyzed a deep learning AI-based system using a convoluted neural network (CNN) with an average image processing speed of 30 to 60 frames per second. The pooled sensitivity and specificity were 95% (95% CI: 85-98) and 88% (95% CI: 76-94), with a diagnostic accuracy (SROC) of 97% (95% CI: 95-98). Sensitivity analysis of CNN studies (4 studies, 538 patients) demonstrated a pooled sensitivity, specificity, and accuracy (SROC) of 95% (95% CI: 82-99), 88% (95% CI: 72-95), and 97% (95% CI: 95-98), respectively. CONCLUSIONS: Artificial intelligence-based machine learning of cholangioscopy images appears to be a promising modality for the diagnosis of indeterminant and malignant biliary strictures.

Humans↗

PubMed2026

Performance of AI-Based Screening Tools for Obstructive Sleep Apnea Across Apnea-Hypopnea Index Thresholds: Systematic Review and Meta-Analysis.

BACKGROUND: Obstructive sleep apnea (OSA) is highly prevalent but remains substantially underdiagnosed. Polysomnography (PSG) is the reference standard, but its cost and limited availability constrain large-scale case identification. AI-based screening tools may support risk stratification and referral prioritization, but their diagnostic accuracy across apnea-hypopnea index (AHI) thresholds remains uncertain. OBJECTIVE: This review aimed to systematically evaluate the diagnostic accuracy of AI-based OSA screening tools at AHI thresholds of ≥5, ≥15, and ≥30 events/hour, with emphasis on models using non-PSG-derived inputs. METHODS: PubMed, Embase, Scopus, and Web of Science were searched for studies published from January 1, 2016, to May 3, 2026. Eligible studies included adults evaluated for suspected OSA or recruited from population-based cohorts, assessed AI-based models intended or interpretable for OSA screening, risk prediction, or screening-oriented severity classification, used PSG as the reference standard, and reported sufficient data to construct or reconstruct 2×2 contingency tables. Diagnostic accuracy was synthesized separately by AHI threshold and input source using bivariate random-effects models, with 95% CIs and prediction intervals (PIs). Risk of bias and certainty of evidence were assessed using QUADAS-2 (Quality Assessment of Diagnostic Accuracy Studies 2) and GRADE (Grading of Recommendations Assessment, Development, and Evaluation), respectively. RESULTS: A total of 60 studies were included, of which 47 contributed data to the meta-analysis. At AHI thresholds of ≥5, ≥15, and ≥30 events/hour, pooled sensitivities were 0.94 (95% CI 0.92-0.96; 95% PI 0.71-0.99), 0.87 (95% CI 0.84-0.89; 95% PI 0.66-0.96), and 0.83 (95% CI 0.79-0.87; 95% PI 0.61-0.94), respectively; the corresponding specificities were 0.77 (95% CI 0.69-0.84; 95% PI 0.30-0.96), 0.81 (95% CI 0.75-0.85; 95% PI 0.39-0.96), and 0.91 (95% CI 0.87-0.94; 95% PI 0.55-0.99), respectively. The corresponding areas under the summary receiver operating characteristic curves were 0.943, 0.907, and 0.920. For non-PSG-derived tools, sensitivities were 0.92, 0.85, and 0.81, and specificities were 0.70, 0.74, and 0.85 at the 3 thresholds, respectively. For PSG-derived models, sensitivities were 0.96, 0.90, and 0.85, and specificities were 0.82, 0.88, and 0.96, respectively. Exploratory subgroup analyses suggested performance variation across selected study and model characteristics, including region, algorithmic framework, data source, and validation method. CONCLUSIONS: AI-based tools showed generally favorable screening performance for OSA across clinically relevant AHI thresholds, although wide PIs suggest variable performance across future comparable populations and settings. By synthesizing diagnostic accuracy across 3 AHI thresholds and distinguishing non-PSG-derived from PSG-derived models, this review extends previous broad or modality-specific reviews and offers a clinically interpretable, pathway-specific basis for linking model performance to intended use. The findings may clarify potential roles for non-PSG-derived tools in front-end screening and referral prioritization and for PSG-derived models in reduced-channel assessment and sleep-laboratory workflow support. Given substantial heterogeneity, limited external validation, and low or very low certainty of evidence, prospective validation is needed before routine implementation.

Humans↗

PubMed2026

Decoding the spatiotemporal patterns of food spoilage microbial communities: Integrating multi-omics and artificial intelligence to enable precision preservation.

In the global food supply chain, food wastage caused by spoilage has resulted in significant economic losses, food shortages, and environmental pressure. This process is fundamentally driven by the spatiotemporal dynamics of microbial communities. However, traditional research methods struggle to elucidate the complex mechanisms of spatial heterogeneity, interspecies interactions, and functional succession. This limits the development of effective preservation strategies. This review systematically reviews the cutting-edge progress of integrating multi-omics technologies and artificial intelligence (AI) to study food spoilage microbial communities, breaking through this bottleneck. We propose an intelligent theoretical framework that could potentially analyze microbial metabolic activities and predict dynamic shelf life if implemented. The conceptual framework integrates multidimensional data, including spatial metabolomics, temporal metatranscriptomics, single-cell transcriptomics, and longitudinal metagenomics. It can also be combined with AI models, such as graph neural networks. The article elaborates on the principles and applications of spatio-temporal monitoring technologies, such as nano secondary ion mass spectrometry, hyperspectral imaging, and the Internet of Things sensing. Through illustrative cases of typical perishable foods, it also explores how such a multi-omics - AI system might be applied to spoilage warning and precise intervention. Additionally, the article addresses the current challenges in data coverage, model generalization, and federated learning implementation. Then the research further explores emerging areas such as engineered probiotics, edge AI, and microfluidic sensing. These areas are targeted at transforming food preservation from an empirical control approach to a data-driven, precise regulatory framework. This transformation provides theoretical support and technical approaches for developing a smart, sustainable food preservation system.

Multiomics↗

PubMed2026

AI Health message intervention: The role of message customization and message source in breast cancer screening among women of color.

OBJECTIVES: To examine the effectiveness of breast cancer screening messages with varying levels of customization (generic, targeted, and tailored) and to compare AI-generated versus human-generated messages. METHODS: A between-subjects experimental design with a control condition was employed. Message content followed a standardized structure and varied by level of customization: generic, targeted (demographic-based), and tailored (perceived susceptibility- and barrier-based). Messages were developed by either the authors or GenAI (ChatGPT-4o). A total of 391 participants recruited via Prolific were randomly assigned to five groups (generic, targeted-human, targeted-AI, tailored-human, and tailored-AI). Self-efficacy, behavioral intentions, attitudes, and message believability were measured using different scales. RESULTS: Customized (tailoring and targeting) health messages performed comparably to generic messages in shaping positive health outcomes. GenAI-generated messages also produced outcomes comparable to those of human-generated messages under standardized conditions. Significant negative indirect effects through message believability for the human-tailored condition was found relative to the generic condition. CONCLUSIONS: GenAI may be a useful tool for developing and customizing scalable health messages. Its effectiveness depends not only on customization but also on maintaining message quality, including readability, clarity, coherence, naturalness, and credibility. PRACTICAL IMPLICATIONS: GenAI may support health practitioners in developing customized and scalable breast cancer messages. However, professional review remains necessary to ensure that the message is culturally appropriate, responsive to patient concerns, and suitable for use alongside patient-provider communication.

Humans↗