Search PubMedSearch

SEARCH · Search PubMed

Results for “Uncertainty modelling”

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.

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

841 records · Page 3Linked to original sources

Methylation profiling in CNS tumor diagnostics: a single-centre real-world experience from Central Europe.

Genome-wide DNA methylation profiling has transformed neuro-oncology by providing an objective, machine learning-based taxonomy that mitigates interobserver variability and refines the histo-molecular criteria of the current WHO classification. We evaluate the real-world diagnostic performance and clinical utility of this modality in a prospective, consecutively accrued three-year cohort of 291 central nervous system (CNS) tumors across a mixed adult-pediatric population. Successful profiling was completed in 95.9% of cases. Using the Epignostix classifier, a high-confidence diagnostic match (calibrated score [CS]&#x2009;&#x2265;&#x2009;0.84) was achieved in 70.3% of analyzable samples, while 26.5% returned lower-confidence scores (&#x2265;&#x2009;0.3 to <&#x2009;0.84) and only 3.2% remained completely unclassifiable (CS&#x2009;<&#x2009;0.3). When integrated into a comprehensive diagnostic framework, methylation profiling provided clinically useful results in 81.1% of cases, establishing diagnoses in 70 cases submitted for molecular subclassification and resolving diagnostic uncertainty or prompting major revisions in 149 histologically challenging tumors. Within truly ambiguous lesions, integration of methylome data dictated tumor grade modifications in 38.8% of cases (upgrading in 29.4% and downgrading in 9.4%), shifting patient risk stratification. Crucially, over half (52.7%) of the lower-confidence cases yielded meaningful clinical integration when supported by histomorphology and ancillary genetic or immunohistochemical markers, demonstrating that rigid score cutoffs should not dictate assay failure. Discrepant or misleading classifications occurred in 1.9%. Updating bioinformatic pipelines from version 11b4 to 12.8 rescued multiple ambiguous entries, increasing overall clinical utility to 84.1%. These findings demonstrate that integrating computational epigenomics with classical neuropathology enhances diagnostic precision, while highlighting the ongoing need for careful clinical-pathological correlation.

Central nervous system tumors

Influenza A virus in Swiss pig herds with respiratory disease: Seasonality and age dependence.

Influenza A virus (IAV) is an important respiratory pathogen in pigs and poses a zoonotic risk to humans in close contact. While IAV epidemiology has been extensively studied in large-scale production systems, data from Switzerland - characterized by small herds and limited live pig imports - remain scarce. This exploratory nationwide cross-sectional study aimed to assess the association between herd-level IAV detection and reported respiratory disease in pig herds, and to explore associations with husbandry-, animal-, and human health-related factors. Between November 2023 and April 2025, 25 Swiss pig herds with caretaker-suspected respiratory symptoms were investigated. In each herd, five nasal swabs were collected and analyzed by quantitative PCR. Herd managers completed an interview, and clinical examinations were performed. Overall, 56&#x2009;% (95&#x2009;% CI: 37,1 - 73,3) of herds tested positive for IAV, comparable to reports from other European countries. The estimated intra-herd detection rate was 49,6&#x2009;% (95&#x2009;% CI: 31,2 - 68,0). Respiratory disease outbreaks associated with IAV detection showed indications of seasonal variation, with no positive herds identified during summer. Across age groups, pigs aged 11-14 weeks had a higher likelihood of IAV detection, with 15,79-fold increased odds (95&#x2009;% CI: 1,50 - 860,4), although with considerable uncertainty. The interpretation is limited by the small sample size, heterogeneous data, and reliance on single time-point qPCR detection. The results suggest that IAV detection in clinically apparent respiratory outbreaks may follow seasonal patterns in Swiss pig herds. Weaners and newly introduced fattening pigs may play a role in such respiratory outbreaks and could represent relevant targets for IAV surveillance in Switzerland. Continued monitoring and the implemen tation of appropriate control measures remain important given the virus's zoonotic potential and impact on pig health.

Animals

Cognitive behavioural therapy-based interventions on stress outcomes in pregnant women: A systematic review and meta-analysis.

BACKGROUND: Stress symptoms were the most common psychological problem in pregnancy. Cognitive behavioural therapy-based interventions are effective for antenatal depression and anxiety symptoms; but there are fewer studies for stress symptoms. OBJECTIVE: The review aims to (1) examine the effectiveness of cognitive behavioural therapy-based interventions in reducing stress outcomes (pregnancy-specific stress symptoms, generic symptoms, and objective stress) in pregnant women, and (2) identify significant moderators affecting the effectiveness of the intervention. DESIGN: Systematic review, meta-analysis, and meta-regression analysis of randomised controlled trials. METHODS: We conducted a three-step search (12 databases, 4 clinical registries, and citation searches) in English and Chinese up to July 24, 2025, by two independent reviewers. Meta-analysis, subgroup, and meta-regression analyses were performed using the R software. Quality assessment and certainty of the evidence were assessed with the Cochrane risk-of-bias tool version 2 and Grading of Recommendations, Assessment, Development, and Evaluation criteria. Publication bias was assessed using funnel plots and Egger's test. RESULTS: We included 20 randomised controlled trials involving a total of 6966 pregnant women from nine countries. Random-effects meta-analyses found that interventions significantly alleviated pregnancy-specific stress symptoms (Hedges' g&#xa0;=&#xa0;-0.84, 95% Confidence Interval, CI -1.42, -0.26, p&#xa0;<&#xa0;.01, I2&#xa0;=&#xa0;92.3%), reduced generic stress symptoms (g&#xa0;=&#xa0;-0.64, 95% CI -1.09, -0.20, p&#xa0;<&#xa0;.01, I2&#xa0;=&#xa0;87.4%) with median and large effect sizes at post-intervention. No effect was found in lowering cortisol levels (g&#xa0;=&#xa0;-0.99, 95% CI -2.58, -0.60, p&#xa0;=&#xa0;.12, I2&#xa0;=&#xa0;84%) at post-intervention. Subgroup and meta-regression analyses indicated that region, age of participants, use of intention-to-treat, missing data management analyses, frequency, modalities, and approaches of interventions, use of different comparators, and attrition rate were significant factors affecting the effectiveness of interventions. Subgroup analyses suggested that the intensity of intervention should be more than once per week using a blended mode among Asian populations. Multivariate meta-regression analyses indicated that both younger age (&#x3b2;&#xa0;=&#xa0;0.13, p&#xa0;=&#xa0;.02) and a lower attrition rate (&#x3b2;&#xa0;=&#xa0;0.03, p&#xa0;=&#xa0;.03) significantly improved the effectiveness of interventions. The overall certainty of the evidence was rated as either very low or low. CONCLUSIONS: Cognitive behavioural therapy-based interventions can supplement antenatal care to alleviate pregnancy-specific stress symptoms and generic stress symptoms, particularly in young Asian women. However, the evidence has some uncertainties. These findings should be interpreted with caution due to substantial heterogeneity. Well-designed trials on a large-scale with long-term follow-ups were needed. REGISTRATION: PROSPERO registration ID: CRD420251115913.

Humans

Comparative Efficacy of Different AI Systems for Polyp Detection by Size During Colonoscopy: Systematic Review and Network Meta-Analysis.

BACKGROUND: Colorectal cancer remains a leading cause of death despite being largely preventable through polypectomy. AI systems designed to enhance polyp detection during colonoscopy have shown promise, but the extent to which they improve detection of different-sized polyps remains unclear. OBJECTIVE: This study compared the size-stratified efficacy of AI-assisted colonoscopy vs standard colonoscopy using the Hartung-Knapp-Sidik-Jonkman (HKSJ) method, and generated exploratory rankings while acknowledging all cross-platform comparisons are indirect. METHODS: This systematic review and network meta-analysis (NMA) searched PubMed, Embase, Cochrane CENTRAL, and Web of Science from inception to July 25, 2026, supplemented by citation searching. We included randomized controlled trials (RCTs) comparing AI-assisted vs standard colonoscopy in adults (&#x2265;18 years of age), reporting mean polyp detection counts stratified by size (&#x2264;5 mm, 6-9 mm, and &#x2265;10 mm). Two reviewers screened studies, extracted data, and assessed risk of bias using the Cochrane Risk of Bias 2.0. We conducted frequentist NMA using the HKSJ method with restricted maximum likelihood estimation, calculated 95% prediction intervals (PIs), and assessed heterogeneity using I2 and &#x3c4;2. Certainty of evidence was rated using the GRADE (Grading of Recommendations Assessment, Development, and Evaluation) framework. RESULTS: A total of 13 RCTs (4156 participants) compared 8 AI systems to standard colonoscopy, forming a network without direct AI comparisons. For diminutive polyps (&#x2264;5 mm), AI showed a modest advantage (standardized mean difference [SMD] 0.21, 95% CI 0.07 to 0.35, 95% PI -1.12 to 1.54), but substantial heterogeneity (I2=86.6%) and wide PI crossing the null indicated high uncertainty. EndoScreener showed the most consistent evidence (SMD 0.36, 95% CI 0.18-0.54). For small and large polyps, effects were minimal (SMD 0.02, 95% CI -0.02 to 0.06, 95% PI -0.03 to 0.07; SMD 0.01, 95% CI 0.00-0.02, 95% PI -0.01 to 0.03). GRADE certainty was very low for diminutive polyps and low for small and large polyps. Sensitivity analysis excluding Tianjin YuJin did not materially change findings. CONCLUSIONS: AI may modestly enhance diminutive polyp detection, but effects on small and large polyps are minimal, with no platform superiority. Given very low to low certainty, findings are hypothesis-generating. This exploratory NMA provides size-stratified comparisons that can inform future head-to-head trial design. Unlike prior reviews aggregating all polyp sizes, we show the overall AI benefit is driven by diminutive polyp detection, providing a framework for targeted deployment-prioritizing AI for diminutive polyp screening, with limited value for larger lesions. Head-to-head trials are urgently needed. TRIAL REGISTRATION: PROSPERO International Prospective Register of Systematic Reviews CRD420251266932; https://www.crd.york.ac.uk/PROSPERO/view/CRD420251266932.

Colonoscopy

Restrictive vs Liberal Transfusion Strategy in Traumatic Brain Injury: A Secondary Analysis of the TRAIN Trial.

IMPORTANCE: Anemia is a prevalent condition among patients with traumatic brain injury (TBI); however, the optimal hemoglobin (Hb) threshold to initiate red blood cell transfusion (RBCT) is not well defined. OBJECTIVE: To assess which of 2 different Hb thresholds for guiding RBCT in patients with anemia and TBI is associated with a more favorable neurological outcome. DESIGN, SETTING, AND PARTICIPANTS: This was a preplanned secondary analysis of the Transfusion Strategies in Acute Brain Injured Patients multicentric randomized clinical trial, conducted in 72 intensive care units across 22 countries between September 1, 2017, and December 31, 2022. Follow-up was completed June 30, 2023. Only patients with TBI were included in the present analysis, conducted from February to May 2025. INTERVENTIONS: Liberal (transfusion at Hb <9 g/dL [to convert to g/L, multiply by 10.0]) vs restrictive (transfusion at Hb <7 g/dL) RBCT strategy over a maximum of 28 days. MAIN OUTCOME AND MEASURES: The primary outcome was the occurrence of unfavorable neurological outcome, defined as a Glasgow Outcome Scale Extended score of 1 to 5 (overall range, 1-8, with higher scores indicating more favorable outcome) at 180 days. In addition, 14 prespecified serious adverse events, including infection and cerebral ischemia, were assessed. Data were analyzed using both the intention-to-treat and per-protocol principles. RESULTS: Of 486 patients who presented with TBI (mean [SD] age, 46.8 [17.6] years; 347 [71.4%] male), 475 were included in the primary outcome analysis: 236 were randomized to the liberal transfusion strategy group and 239 to the restrictive transfusion strategy group. Both groups had similar baseline characteristics. In total, 534 RBCTs were administered in the liberal transfusion strategy group, compared with 246 RBCTs in the restrictive group. At 180 days after randomization, 138 patients (58.5%) in the liberal group had unfavorable neurological outcome compared with 161 patients (67.4%) in the restrictive group (relative risk [RR], 0.86 [95% CI, 0.75-1.00]; P&#x2009;=&#x2009;.047; fragility index&#x2009;=&#x2009;1). There were no significant differences in the occurrence of secondary outcomes (eg, 28-day mortality: 42 of 240 [17.5%] vs 51 of 244 [20.9%]; RR, 0.84 [95% CI, 0.58-1.21]; P&#x2009;=&#x2009;.34) or serious adverse events (eg, RR, 1.13 [95% CI, 0.88-1.43]; P&#x2009;=&#x2009;.34 for infection and RR, 0.87 [95% CI, 0.40-1.90]; P&#x2009;=&#x2009;.72 for cerebral ischemia). After adjustment for several confounders, being randomized to the liberal group was associated with a lower observed probability of unfavorable neurological outcome (odds ratio, 0.60 [95% CI, 0.38-0.94]; P&#x2009;=&#x2009;.03). CONCLUSIONS AND RELEVANCE: In this secondary analysis of a multicenter randomized clinical trial, a liberal RBCT strategy was associated with a lower risk than a restrictive RBCT strategy of unfavorable neurological outcome at 180 days among patients with TBI. These findings should be interpreted with caution in light of the inherent uncertainty of the estimate. TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT02968654.

Humans

APOL1 kidney disease: a critical narrative review of molecular mechanisms, clinical heterogeneity, and the emerging therapeutic landscape.

BACKGROUND: The G1 and G2 variants of the APOL1 gene represent significant genetic risk factors for APOL1 kidney disease and contribute substantially to the excess burden of renal disease observed in individuals of African ancestry. Importantly, both variants exhibit incomplete penetrance, with only approximately 15-20% of high-risk genotype carriers ultimately developing overt nephropathy. OBJECTIVE: To provide a critically appraised, clinically oriented narrative synthesis of APOL1 kidney disease that (i) assigns an explicit certainty rating to each major mechanistic and clinical claim, (ii) identifies where published estimates diverge, where associations remain contested, and where conclusions have been overstated in the secondary literature, and (iii) aligns terminology, testing guidance and therapeutic expectations with the conclusions of the 2025 KDIGO Controversies Conference and with clinical trial data available to August 2026. METHODS: This literature narrative review was performed using a literature search of PubMed and Scopus focusing on APOL1-related nephropathy. Mainly studies published from 2010 to 2026 were considered; however, some selected historical papers from 2005 to 2010 were used for better understanding of the underlying mechanisms and history. Used search terms were "APOL1," "APOL1 risk variants," "chronic kidney disease," AMPLITUDE trial, MZE829, HORIZON trial, "focal segmental glomerulosclerosis," "HIV-associated nephropathy," "podocyte injury," "inaxaplin," "VX-147," KDIGO 2025, and "antisense oligonucleotides." Trial status and topline results for agents in development were additionally verified against ClinicalTrials.gov registrations and sponsor disclosures. The literature search was last updated on 10 August 2026. The inclusion criteria of the study were peer-reviewed original articles, genome-wide association studies, randomised controlled trials, translational studies, mechanistic investigations, and high-quality review articles published in the English language. Exclusion criteria included conference abstracts without peer review, duplicate papers, non-English publications with unreliable translation, and case reports with no relevance to the underlying mechanisms. More attention was paid to studies focusing on molecular pathogenesis of APOL1 nephropathy, second-hit pathophysiology, genotypes/phenotypes, and new therapies (e.g. inhibitors such as Inaxaplin). The review method and design have been prepared according to SANRA (Scale for the Assessment of Narrative Review Articles) criteria. Among eligible articles, priority was given to studies with larger sample sizes, more recent publication dates, higher-impact peer-reviewed journals, and direct clinical or mechanistic relevance to APOL1-associated nephropathy; where multiple studies addressed the same question, the most methodologically rigorous and most recent source was preferentially cited. To move beyond description, each principal claim carried forward into this review was assigned a qualitative certainty rating (high, moderate, low or very low) on the basis of study design, consistency across independent cohorts, directness of the evidence to human disease, and precision of the estimate. These ratings, together with the study design that would be required to resolve each remaining uncertainty, are presented in Table&#xa0;5. This grading represents a structured judgement by the authors and is not a formal GRADE assessment. RESULTS: Pathogenic actions of APOL1 risk alleles depend on toxic gain-of-function activities that result from the disruption of ion channels. Mitochondrial dysfunction, endoplasmic reticulum stress, and inflammasome activation play roles as secondary downstream modulators of podocyte damage. The existence of incomplete penetrance and lack of symptoms in people with high-risk alleles highlights the need for secondary triggers, including environmental, infectious, and inflammatory factors, for disease onset and progression. High-risk APOL1 genotypes increase the likelihood of rapidly progressing kidney diseases like FSGS, which amplify susceptibility in HIVAN when accompanied by secondary causes like HIV infection. Management is mainly through renin-angiotensin antagonists, but recent treatments include antisense oligonucleotides, immunomodulators, and small molecule inhibitors like inaxaplin. Although promising, inaxaplin (VX-147) showed a ~47% reduction in urine protein/creatinine ratio (UPCR) in Phase 2a trial; however, these findings are based on a relatively small sample size, an open-label study design, and short-term follow-up, and therefore require confirmation in ongoing Phase 3 studies. As this is a narrative review rather than a primary study, no new patient-level data are reported. Across the studies synthesised, high-risk APOL1 genotypes were consistently associated with podocyte injury and with a faster decline in kidney function than low-risk genotypes; however, the magnitude of this association varied substantially with how cohorts were ascertained. The association is robust and reproducible for focal segmental glomerulosclerosis, HIV-associated nephropathy, and hypertension-attributed kidney failure, and remains inconsistent for diabetic kidney disease. Therapeutic development has accelerated, but the supporting clinical evidence remains early phase. Inaxaplin (VX-147) reduced the urine protein-to-creatinine ratio by approximately 47.6% at week 13 in a 16-participant, single-group, open-label Phase 2a study, and is now being evaluated in the randomised, double-blind, placebo-controlled Phase 2/3 AMPLITUDE trial (NCT05312879), whose pre-specified week 48 interim analysis is anticipated in early 2027. MZE829, an orally administered APOL1 inhibitor, produced a mean 35.6% reduction in the urine albumin-to-creatinine ratio at 12&#xa0;weeks in the Phase 2 HORIZON study; because HORIZON was a small, open-label, single-arm basket study (15 participants enrolled, 12 evaluable) whose primary endpoints were safety and tolerability, this reduction is neither placebo adjusted nor the result of a formal test of efficacy. To date, no APOL1-targeted agent has demonstrated benefit on a hard kidney endpoint. CONCLUSION: APOL1 is the clearest current example of a genetically defined, mechanism-targetable kidney disease, but its evidence base is uneven. The genetic association is firmly established; whereas much of the mechanistic literature derives from overexpression systems, several downstream pathways remain contested, and every APOL1-targeted therapy is so far supported only by short-term, surrogate-endpoint data. The principal unresolved issues are the determinants of incomplete penetrance, the absence of a validated progression biomarker and of any model reproducing the common slowly progressive phenotype, and the long-term efficacy and safety of APOL1-directed therapy. Genotype-guided risk stratification is therefore best regarded as clinically reasonable but not yet proven, and routine population-level screening is not currently supported.

AMPLITUDE trial

Adolescent health across Asia Pacific, 2000-23: a systematic analysis for the Global Burden of Disease Study 2023.

BACKGROUND: The Asia Pacific region is home to more than half of the world's 1&#xb7;93 billion adolescents (aged 10-24 years). Addressing adolescent health in this region is of global importance, but to date a systematic analysis of key contributors to disease in adolescents has not been done, which is a barrier to responsive action. This systematic analysis of the Global Burden of Diseases, Injuries, and Risk Factors Study (GBD) 2023 aims to provide a comprehensive assessment of adolescent health across the Asia Pacific region, at both the subregional and national levels, encompassing burden of disease, mortality, and prevalence of adolescent risk factors. METHODS: As part of GBD 2023, we obtained estimates for cause-specific mortality, disability-adjusted life-years (DALYs), and risk factor prevalence by sex for adolescents aged 10-24 years and 5-year age groups (10-14 years, 15-19 years, and 20-24 years) across 44 countries and territories (hereafter referred to collectively as Asia Pacific), grouped by seven UN subregions, from 2000 to 2023. We extracted GBD 2023 population counts and estimates of number and rate (per 100&#x2008;000 population) for mortality and disease burden (DALYs). Risk prevalence estimates were obtained directly from the Institute for Health Metrics and Evaluation, and binge drinking estimates were sourced from WHO. Estimates are reported with 95% uncertainty intervals (UIs) where possible. UIs were estimated by running 250 draws of the posterior distribution, ordering the draws, and selecting the 2&#xb7;5th and 97&#xb7;5th percentiles for each metric. FINDINGS: In 2023, in adolescents across Asia Pacific, there were 637&#x2008;496 deaths and a total disease burden of 115&#xb7;8 million DALYs, representing 34&#xb7;1% of global adolescent deaths and 40&#xb7;6% of the global adolescent burden of disease. Non-communicable diseases (NCDs; particularly mental disorders) were the leading causes of disease burden and mortality (64&#xb7;8% of DALYs and 43&#xb7;9% of deaths). Unintentional and transport injuries were also leading causes of death (14&#xb7;7% of deaths due to transport injury and 13&#xb7;3% of deaths due to unintentional injury) and leading causes of disease burden particularly among males in south-eastern Asia. In Melanesia, Micronesia, and some parts of south-eastern Asia (Cambodia, Indonesia, Laos, the Philippines, and Timor-Leste), respiratory infections and tuberculosis remained important contributors. Southern Asia had the largest reduction (1&#xb7;5% per year) in all-cause DALYs over the study period, and Australia and New Zealand (0&#xb7;2% per year) had the smallest, with females in Australia and New Zealand showing a slight increase contrary to regional trends. Eastern Asia had the largest reduction (2&#xb7;8% per year) in all-cause mortality rate and Melanesia (0&#xb7;8% per year) the smallest. Risk factors generally had between-subregion and within-subregion variation; however, some regional trends stood out, with overweight and obesity increasing in all countries across the region, and binge drinking increasing in more countries than not. In 2023, prevalence of smoking in males exceeded that in females in every country, from 40% difference in Timor-Leste to less than 1% difference in Australia. Anaemia prevalence is decreasing in all countries, but female prevalence was higher and reducing at a slower rate than in males. Bullying prevalence was slightly higher in Polynesia, Micronesia, and Melanesia combined, Australia and New Zealand, and eastern Asia compared with southern and south-eastern Asian subregions. INTERPRETATION: Several patterns were consistent across the region: the dominance of mental disorders and NCDs, the universal rise in overweight and obesity (particularly high in Oceanic countries but increasing rapidly in south and south-eastern Asia), and persistent sex-specific challenges across subregions: unintentional injuries and smoking in males, and anaemia in females. Actions to tackle shared risk factors (while accounting for context-specific local health profiles, workforce deficits, cultural factors, and health system capacity) should not be forgone due to local variation. Future research could focus on subnational variation, intersecting inequalities, and multi-sectoral interventions targeting shared risk factors. Priority actions should include regional investment in adolescent mental health services and obesity prevention, targeted injury reduction strategies for high-risk populations, and sex-specific approaches to smoking cessation and anaemia reduction, delivered through local health systems with the capacity and cultural responsiveness to meet local needs. FUNDING: Gates Foundation and Australian Government.

Humans

Statistical test to compare the linkage model and the admixture model based on central limit results.

In the Admixture Model, the probability that an individual carries a certain allele at a specific marker depends on the allele frequencies in K ancestral populations and the proportion of the individual's genome originating from these populations. The markers are assumed to be independent. The Linkage Model is a Hidden Markov Model that extends the Admixture Model by incorporating linkage between neighboring loci. We prove consistency and asymptotic normality of maximum likelihood estimators for the ancestry of individuals in the Linkage Model, complementing earlier results by (Pfaff et al., 2004; Pfaffelhuber and Rohde, 2022; Heinzel, 2025) for the Admixture Model. These results are used to prove that a statistical test that allows for model selection between the Admixture Model and the Linkage Model is an asymptotic level-&#x3b1;-test. Finally, we demonstrate the practical relevance of our results by applying the test to real-world data from The 1000 Genomes Project Consortium (2015).

Genetic Linkage

Modelling the effects of biological intervention in a dynamical gene network.

Cellular response to environmental and internal signals can be modeled by dynamical gene regulatory networks (GRN). In the literature, three main classes of gene network models can be distinguished: (1) non-quantitative (or data-based) models which do not describe the probability distribution of gene expressions; (2) quantitative models which fully describe the probability distribution of all genes co-expression; and (3) mechanistic models which allow for a causal interpretation of gene interactions. We propose two rigorous frameworks to model gene alteration in a dynamical GRN, depending on whether the network model is quantitative or mechanistic. We explain how these models can be used for design of experiment, or, if additional alteration data are available, for validation purposes or to improve the parameter estimation of the original model. We apply these methods to the Gaussian graphical model, which is quantitative but non-mechanistic, and to mechanistic models of Bayesian networks and penalized linear regression.

Gene Regulatory Networks

The landscape of pruning for large language models: A systematic review and unified taxonomy.

Confronting the inherent tension between the exceptional capabilities and the immense computational costs of Large Language Models (LLMs), pruning has become a crucial technique for achieving efficient deployment. However, a systematic analytical framework dedicated specifically to LLM pruning remains absent. In this paper, we aim to bridge this gap. We first elucidate the theoretical foundations that underpin the effectiveness of pruning, namely overparameterization and redundancy, and then propose a multidimensional taxonomy that organizes existing approaches along the axes of granularity, timing, and criteria. Building upon this unified perspective, we further analyze performance recovery mechanisms and the broader evaluation ecosystem, while also exploring forward-looking challenges such as interpretability, automation, and hardware-algorithm co-design. Through this comprehensive synthesis, we seek to provide an integrated and coherent analytical lens for advancing both research and practice in LLM pruning.

Large Language Models

Penalized Cumulative Probability Model for a Continuous Outcome Subject to Detection Limits.

Mixed-type outcome data occur when the outcome variable's distribution is a mixture of both continuous and discrete ordinal variables. Such mixed-type outcomes are common in biomedical, psychological, and the health sciences, particularly for variables having either a detection or quantitation limit. When interest lies in identifying a combination of genomic features associated with a mixed-type outcome, any method used would require a variable selection strategy for high-dimensional data. Unfortunately, few variable selection methods exist for modeling a mixed-type outcome when the covariate space is high dimensional. This study develops a high-dimensional penalized cumulative probability model (CPM), to allow for the identification of genomic features associated with mixed-type outcome of interest. We demonstrated how such model may be estimated using the iterative penalization procedure-the generalized monotone incremental forward stagewise (GMIFS) algorithm. The Model-X knockoffs procedure was combined with the estimation algorithm to control the false discovery rates (FDR) when performing variable selection. Through extensive simulation studies, our penalized CPM was shown to outperform alternative methods in terms of controlled variable selection performance by achieving high statistical power with the FDR being controlled at the target level. We demonstrate the utility of our method by applying it to predict estimated glomeruli filtration rate (eGFR) in kidney transplant recipients at 24&#x2009;months post-transplant using baseline gene expression data as predictors. Our CPM model identified five genes associated with this mixed-type outcome which have important links to renal disease, which may provide prognostic guidance for kidney transplantation recipients.

Models, Statistical

Construction of precision clinical-proteomics risk model based on machine learning for predicting heart failure in type II diabetes mellitus.

BACKGROUND AND AIMS: Heart failure (HF) is a severe complication in type 2 diabetes mellitus (T2DM), but current risk stratification scores have limited predictive accuracy. We aimed to develop novel prediction tools integrating clinical variables with proteomics to improve risk stratification of hospitalization for HF in T2DM. METHODS AND RESULTS: In this study, we included 2111 UK Biobank participants with T2DM but no prior HF, and profiled 2920 proteins to predict 10-year incident HF hospitalization. Participants were randomly divided into training (70%), tuning (10%), and validation (20%) sets.Three prediction models were developed: a Clinical model based on demographic characteristics, comorbidities, medication use, and laboratory indices; a Protein model based on 40 proteins selected by the Light Gradient Boosting Machine (LGBM); and the Clinical OMics and Protein ASSessment for Heart Failure (COMPASS-HF) model, which integrated both clinical variables and the LGBM-selected proteins. Models were evaluated for area under the curve (AUC), sensitivity, and specificity. During follow-up, 168 participants (7.96%) developed incident HF. The COMPASS-HF model showed better discrimination than the Clinical model, with an AUC of 0.897 (95% CI: 0.850-0.945) versus 0.790 (95% CI: 0.723-0.856). It also demonstrated higher sensitivity (0.882; 95% CI: 0.725-0.967) and consistent performance in subgroups. COMPASS-HF effectively stratified risk of hospitalization for HF, with cumulative incidence rates of 31.9% in the high-risk group and 1.2% in the low-risk group. CONCLUSIONS: By combining clinical and proteomic variables, we developed a high-performance HF prediction model for T2DM, enabling precise risk stratification and informing early intervention strategies.

Humans

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&#xa0;=&#xa0;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&#xa0;=&#xa0;35), colorectal cancer (n&#xa0;=&#xa0;21), and pancreatic cancer (n&#xa0;=&#xa0;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

Integrative modeling of the genome structure and dynamics in fission yeast.

Genome organization in the nucleus is highly structured and dynamic. Recent advances in genomic technology have enabled the measurement of genome-wide architecture and locus-specific motion, yielding contact maps and live-cell trajectories. However, these outcomes are derived from different modalities and are not directly comparable, with their quantitative integration being a key challenge. Here we establish a genome-wide live-cell imaging platform in fission yeast Schizosaccharomyces pombe, tracking 131 chromosomal loci, along with the spindle pole body (SPB) and nucleolus, to construct a quantitative map of locus dynamics. By integrating these dynamics with contact data through polymer modeling of Hi-C data, we build a physics-based "digital twin" of the S. pombe genome consistent with the spatiotemporal dynamics of interphase chromatin. We validate it against genome-wide mobility patterns and known architectural features, including centromere and telomere clustering. The model also identifies distinct dynamical regimes: centromere- and telomere-proximal loci relax within [Formula: see text]150 s, whereas the remaining loci relax within [Formula: see text]70 s. We measure semiperiodic dynamics of SPB motion, including a characteristic peak near 225 s and [Formula: see text] fluctuations. We use the model with SPB-directed forcing to show how these low-frequency components propagate through the genome to drive genome-wide chromatin displacements. Together, this predictive physics-based modeling framework integrates genome structure and dynamics to reveal how nuclear mechanical driving forces shape chromosome motion, linking mechanically driven chromatin responses to genome maintenance and regulation.

Schizosaccharomyces

Revealing the Shared Genetic Architecture of Metabolic Dysfunction-Associated Steatotic Liver Disease-Related Traits Through Genomic Structural Equation Modeling.

Although individual traits related to metabolic dysfunction-associated steatotic liver disease (MASLD) have been investigated through large-scale genome-wide association studies (GWASs), the shared genetic susceptibility across these traits remains unclear. We therefore conducted a multivariate GWAS of key MASLD-related traits to elucidate their common genetic architecture. We applied genomic structural equation modeling to model a latent genetic factor (MASLD-F) underlying genetically correlated MASLD-related traits, leveraging their GWAS-derived genetic correlations. We then performed functional annotations, including fine-mapping, transcriptome-wide association study, and cell- and tissue-type-specific enrichment analyses, and conducted Mendelian randomization analyses to identify modifiable risk factors. Our multivariate MASLD-F GWAS identified 50 independent variants across 48 genomic loci. Transcriptomic imputation identified several MASLD-F-associated genes, including ARNTL, NPC1, BTBD10, VDAC2, TSKU, SFMBT1, and ABHD17C. We observed significant enrichment of MASLD-F-related genetic signals predominantly in brain tissues, pancreatic islets, and the adrenal gland. Additionally, six modifiable risk factors and four modifiable protective factors for MASLD-F were identified. These findings reveal a complex shared genetic architecture underlying MASLD components, thereby expanding our understanding of disease pathogenesis and providing novel insights for precision medicine and public health interventions.

Humans

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