Search PubMedSearch

SEARCH · Search PubMed

Results for “algorithm for CV safety”

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.

2,642 recordsLinked to original sources

Embedding cardiovascular risk assessment into routine BTK inhibitor management in chronic lymphocytic leukemia.

INTRODUCTION: Cardiovascular (CV) toxicities remain a major challenge during Bruton tyrosine kinase inhibitor (BTKi) therapy for chronic lymphocytic leukemia (CLL). Selecting the optimal BTKi based solely on a history of overt CV disease may underestimate underlying cardiovascular vulnerability. AREAS COVERED: We performed a targeted, non-systematic review of PubMed and MEDLINE to examine the association between baseline CV comorbidities and BTKi-related CV toxicities in CLL. Current evidence indicates that preferential use of BTKis with more favorable CV safety profiles, coupled with appropriate cardio-oncology surveillance, reduces the risk of CV adverse events in patients with pre-existing CV disease. In patients without established CV disease, the Systematic Coronary Risk Evaluation 2 (SCORE2) and SCORE2-Older Persons (SCORE2-OP) may help identify clinically meaningful latent CV risk, enabling early optimization of modifiable risk factors in line with the proactive cardiovascular management strategy endorsed by the 2026 European Hematology Association (EHA) CLL guidelines. EXPERT OPINION: A structured, risk-adapted approach integrating standardized CV risk assessment, early management of modifiable risk factors, individualized BTKi selection, and multidisciplinary cardio-oncology collaboration may improve the safety and tolerability of BTKi therapy in CLL. Pending prospective validation, SCORE2 and SCORE2-OP should complement, rather than replace, dedicated cardio-oncology evaluation.

Humans

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

Magnesium administration for vasospasm prevention in acute aneurysmal SAH: a multicenter randomized controlled trial.

Aneurysmal subarachnoid hemorrhage (aSAH) is associated with significant morbidity and mortality, with cerebral vasospasm (CV) and delayed cerebral ischemia (DCI) being the primary contributors to poor outcomes. Magnesium sulfate (MgSO₄) has demonstrated neuroprotective and vasodilatory properties in preclinical models. This study aimed to evaluate the effect of targeted serum magnesium (Mg) maintenance on CV and exploratory clinical outcomes following aSAH. We conducted a prospective, multicenter, single-blind RCT across four neurocritical care units in Korea between 2019 and 2024. A total of 121 aSAH patients were randomized to receive either IV MgSO₄or placebo within six hours of admission. Mg was infused to maintain serum concentrations between 2.0 and 3.0 mg/dL for 14 days. The primary outcome was incidence of CV assessed by transcranial doppler. Secondary outcomes included DCI, ICU and hospital length of stay, modified rankin scale (mRS) at 30 days. There was no significant difference in overall CV incidence; however, the Mg group demonstrated significantly lower mean flow velocity and Lindegaard ratio on days 4-9, indicating reduced vasospasm severity. In exploratory multivariable analyses, a median serum Mg concentration > 2.5 mg/dL during the first 14 hospital days was independently associated with lower risks of CV and DCI. No significant differences were found in mRS scores, ICU and hospital stay, or serious adverse events between groups. Early targeted Mg administration improved TCD-derived hemodynamic markers during the peak vasospasm window; however, it did not significantly reduce CV incidence, DCI, ICU or hospital stay, or 30-day functional outcome.

Humans

Prevalence and cardiometabolic impact of mild autonomous cortisol secretion in primary aldosteronism: a systematic review and meta-analysis.

Many primary aldosteronism (PA) patients harbor concomitant mild autonomous cortisol secretion (MACS), termed "Connshing syndrome." The prevalence and cardiometabolic impact of this co-secretion have not been quantitatively synthesized. To determine the pooled prevalence of MACS in PA and evaluate its association with cardiovascular (CV) events, type 2 diabetes mellitus (T2DM), and obesity. PubMed, Embase, Cochrane, Web of Science, and Scopus through January 2026. Following PRISMA 2020 and MOOSE guidelines (PROSPERO: CRD420261334965), studies reporting MACS prevalence [post-1 mg dexamethasone suppression test (DST) cortisol ≥1.8 μg/dl] in PA were included. Prevalence was pooled using random-effects models. Odds ratios (ORs) were calculated for cardiometabolic outcomes. Certainty was evaluated using GRADE. Fourteen studies (2356 patients) were included. Pooled MACS prevalence was 26.2% [95% confidence interval (CI): 24.1-28.4; I2 = 24.2%; prediction interval: 21.6-31.4%], consistent across European (27.0%) and East Asian (25.2%) cohorts ( P = 0.62). MACS + PA patients had higher odds of CV events (OR 1.60; 95% CI: 1.07-2.38; P = 0.021) and T2DM (OR 1.37; 95% CI: 1.00-1.86; P = 0.048), but not obesity (OR 1.10; P = 0.411). Sensitivity analyses confirmed robustness. GRADE certainty was moderate for prevalence and low for CV events. Approximately one in four PA patients has MACS, associated with a 60% higher CV event risk. These findings support routine DST screening in PA and have implications for perioperative management and cardiometabolic risk stratification.

Humans

An automated geometric modeling framework in GATE for the design and optimization of high-sensitivity converging-beam SPECT collimators.

Objective.The trade-off between detection sensitivity and spatial resolution is a fundamental challenge in designing organ-dedicated Single-photon emission computed tomography (SPECT) collimators. While converging-hole geometries offer a solution, their optimization is often hindered by the lack of flexible computational tools capable of modeling large-scale, non-parallel hole arrays. This study aims to develop an automated geometric modeling framework to facilitate the design and evaluation of complex converging- and diverging-hole collimators within standard Monte Carlo environments.Approach.We developed a specialized modeling framework by implementing custom C++ classes and a vector-based alignment algorithm within GATE. This platform enables automated, orientation-consistent construction of large-scale converging arrays not natively supported by standard implementations. A high-sensitivity pure cone-beam collimator (CBC) was designed using this framework. The evaluation used hot-rod, disc, and Jaszczak phantoms for physical characterization, while XCAT and dedicated brain models were employed for clinical tasks, including cardiac, brain perfusion, and DaTscan SPECT simulations.Main results.The CBC achieved a nearly fourfold sensitivity increase compared to a conventional low-energy high-resolution parallel-hole collimator at a 20 cm radius of rotation, while maintaining comparable spatial resolution. Despite a 52.3% field of view reduction, the CBC yielded a 2.2-fold noise reduction (CV: 11.7% vs 25.9%) and mitigated partial volume effects via geometric magnification. XCAT and brain phantom simulations confirmed enhanced anatomical definition and contrast recovery in cardiac, perfusion, and DaTscan tasks.Significance.This work provides an efficient computational tool for rapid design space exploration of advanced collimator geometries. The results demonstrate that the proposed CBC design offers a significant sensitivity advantage, making it highly suitable for high-performance, small-volume clinical applications such as brain and cardiac molecular imaging.

Tomography, Emission-Computed, Single-Photon

Blood pressure management after endovascular thrombectomy in acute ischemic stroke: association with symptomatic intracranial hemorrhage and functional outcome at 3 months.

BACKGROUND: No clear consensus exists on ideal systolic blood pressure (SBP) targets after endovascular thrombectomy (EVT) following an acute ischemic stroke (AIS). This study investigated the association between SBP parameters within the first 24 h after EVT and 3-month functional outcomes and the risk of symptomatic intracranial hemorrhage (sICH). METHODS: We retrospectively collected and prospectively followed clinical, and radiological data for patients undergoing EVT for AIS from 2016 to 2024, including 2-hourly BP measurements during the first 24 h and SBP variability assessed by standard deviation (SD) and coefficient of variation (CV). Outcomes included 3-month functional status and sICH, and their associations with post-EVT BP metrics were analyzed. RESULTS: A total of 268 post EVT patients were included with a median age of 61 years (IQR, 51-69). Mean SBP was 129.67 ± 17.17 mm Hg, with SBP variability (SD 12.6 ± 5.4 mm Hg; CV 9.6 ± 3.8 %), while good functional outcome and sICH occurred in 39.7 % and 4.9 % of patients, respectively. Multivariate regression showed that higher admission NIHSS (>15) [0.90 (95 %CI, [0.86, 0.95], p = 0.000)], recanalization status [1.88 (95 %CI, [1.43, 2.48], p = 0.00)], and SBP-CV ≥ 10 [0.44 (95 %CI, [0.2, 0.94], p = 0.036)] was independently associated with poor 3-month functional outcome, while higher admission NIHSS (>15) [0.87 (95 %CI, [0.77,0.98], p = 0.02)] and diabetes [0.12 (95 %CI, [0.03, 0.54], p = 0.006)] predicted increased risk of sICH. CONCLUSIONS: The study showed that reduced BP variability during the first 24 h post-EVT was associated with better 3-month functional outcomes. A clear association between SBP and sICH risk was not demonstrated.

Humans

Robotic Needle Insertion for CT-guided Percutaneous Biopsy of Thoracoabdominal Lesions: A Prospective Multicenter Randomized Trial.

Purpose To compare safety and feasibility between a novel CT-guided robotic system and the conventional freehand technique for puncture biopsy of thoracoabdominal lesions. Materials and Methods In this prospective multicenter randomized trial, individuals with suspected lesions were enrolled between July 2023 and April 2024 across three university teaching hospitals and randomized to the robot-assisted group (n = 82) or the freehand group (n = 83). Procedure outcomes included the technical success rate, targeting error, number of CT scans and needle adjustments, puncture time, and complications. Descriptive and inferential statistics were calculated. Results A total of 165 participants (mean age, 60 years &#xb1; 10 [SD]; 83 male) were included. Compared with the freehand group, the robot-assisted group demonstrated a higher technical success rate (97.56% [80 of 82] vs 62.65% [52 of 83], P < .001), lower targeting error (mean Euclidean deviation: 1.7 mm &#xb1; 1.1 vs 4.5 mm &#xb1; 3.9, P < .001), and fewer CT scans (mean, 4.3 &#xb1; 1.9 vs 5.2 &#xb1; 2.3; P = .002) and needle adjustments (mean, 0.7 &#xb1; 0.7 vs 1.6 &#xb1; 1.6; P = .003). Despite differences in geometric precision, both groups achieved 100% (82 of 82 and 83 of 83) diagnostic yield. The median puncture time was comparable between groups (5.5 minutes &#xb1; 4.3 vs 4.8 minutes &#xb1; 7.0, P = .50). During lung biopsies, the robot-assisted approach yielded fewer complications compared with the freehand approach (4.88% [four of 82] vs 16.87% [14 of 83], P = .014). Conclusion Compared with the freehand approach, robot-assisted biopsy yielded greater precision and reduced adjustments and complications while demonstrating noninferior diagnostic efficacy and comparable duration. Keywords: Robotic Needle Insertion, Biopsy, Thoracoabdominal Lesions, Robot-assisted Biopsy, CT-guided Intervention, Percutaneous Needle Biopsy, Randomized Controlled Trial, Algorithm Development, CT, Clinical Testing, Interventional-Body, Biopsy/Needle Aspiration, Percutaneous, Thorax, Abdomen/GI, Liver, Lung, Kidney &#xa9;RSNA, 2026.

Humans

Repeated low-level red-light therapy for improving asthenopic symptoms and accommodation in presbyopia.

BACKGROUND: To assess the short-term effectiveness of repeated low-level red light (RLRL) therapy in relieving asthenopia and enhancing accommodation in presbyopia. METHODS: This randomized, parallel-group, double-masked clinical trial enrolled adults with presbyopia and self-reported asthenopia. Participants were allocated using computer-generated randomization and randomly assigned at a 1:1 ratio to RLRL or sham groups. Blinding included participants, examiners, assessors, and statisticians. The primary outcome was the change from baseline in the Computer Vision Syndrome Questionnaire (CVS-Q) score at day 31. Secondary outcomes were the change in accommodative amplitude (AA), Near Activity Visual Questionnaire (NAVQ) score, habitual near visual acuity, near-addition power, accommodative facility, positive and negative relative accommodation, binocular cross-cylinder response, and accommodative convergence-to-accommodation ratio. Continuous outcomes were analyzed using linear mixed-effects models. RESULTS: Sixty-four of 66 randomized participants (aged 41-62&#x2009;years) completed the 1-month trial. At day 31, RLRL showed greater improvement than sham in CVS-Q score (adjusted mean difference, -1.75 points; 95% CI, -3.10 to -0.39), binocular AA (1.09 D; 95% CI, 0.37 to 1.82), and NAVQ score (-8.07 points; 95% CI, -14.17 to -1.97). The effect on AA was most pronounced in a subgroup of eyes with baseline amplitude >2.0&#x2009;D (adjusted mean difference 1.33&#x2009;D; 95% CI 0.32-2.34). Other measures did not differ between groups at each visit. No treatment-related adverse events were reported. Adherence was similar between groups (mean compliance: 98.2% vs 97.5%). CONCLUSIONS: Short-term treatment with RLRL significantly reduced asthenopic symptoms and improved accommodative amplitude in individuals with presbyopia.Trial registration: NCT06745661 (registered December 8, 2024).

Humans

Future promise, current clinical ambiguity: a systematic review of machine learning algorithm outputs predicting risk of cardiovascular disease.

OBJECTIVE: To examine whether the outputs of machine learning algorithms designed to predict risk of cardiovascular disease (CVD) address known deficiencies of the Framingham Risk Score (FRS) and improve risk estimates. METHODS: For this critical review, Medline, Embase and IEEE were searched from inception to 1 January 2025. Included were studies describing machine learning algorithms designed to specifically compare output of cardiovascular risk assessment with the FRS. Commentaries, letters, unpublished work or non-peer-reviewed papers were excluded.Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, two reviewers screened titles and abstracts independently, then populated a purpose-built data extraction form. A subsequent qualitative thematic analysis focused on algorithms' strengths, added value, potential harms, unintended consequences and equity implications.The main outcome assessed was whether, among healthy adults, the algorithm improved CVD risk prediction relative to the FRS. RESULTS: Of 707 studies retrieved, 29 met inclusion criteria. 23 reported improved predictive ability relative to the FRS. Most datasets and/or medical records used included sociodemographic predictors of CVD not included among FRS inputs. Some added costly diagnostic tests like CT angiography to FRS screening indicators. When they were defined, inputs and outcomes such as hypertension or myocardial infarction did not always adhere to FRS values. Statistical significance was generally taken as a proxy for clinical significance. Some algorithms overestimated the number at risk compared with the FRS without discussing whether that larger proportion might be at risk of overdiagnosis rather than CVD, while a few decreased the proportion found to be at risk. CONCLUSIONS: Use of artificial intelligence to improve accuracy of risk assessment for CVD demonstrates the technological capacity to merge known sociodemographic predictors with biologic variables and examine non-linear interactions among these. Still needed to achieve patient benefit is clinical insight, adherence to screening principles and cost-benefit assessment of inputs selected.

Humans

Alternative genetic codes in bacteria and archaea identified with a fast k-mer-based algorithm.

The genetic code is conserved across all domains of life and is often described as universal. Nevertheless, many exceptions to the "universal" code have now been documented, most of these through manual or semiautomated inspection of highly conserved genes. Modern bioinformatics tools improved our ability to find alternative genetic codes but remain computationally expensive, preventing widespread use on thousands of new species identified by sequencing environmental samples. Here, I report a >100-fold accelerated method for inferring the genetic code directly from assembled genomes and apply it to thousands of previously uncharacterized assemblies from archaea and bacteria. I describe three candidate genetic code variations, one of which, an alternative genetic code used by a family of Asgard archaea, is a unique example of sense codon reassignments for this domain. Identifying genetic code variations is important for understanding evolution of the standard code and improving accuracy of protein databases and open reading frame identification.

Genetic Code

Meniscal preservation in the age of biologics: toward a quantitative decision algorithm for personalized repair.

BACKGROUND: Despite advances in arthroscopic repair and biologic augmentation, surgical indication for meniscal tears remains heterogeneous. No standardized framework currently integrates biomechanical, clinical, and biological determinants to guide repair versus resection. PURPOSE: To develop a quantitative decision model-the Meniscal Preservation Score (MPS)-that unifies biomechanical and biological evidence to stratify reparability potential and standardize treatment selection in meniscal surgery. METHODS: A systematic evidence synthesis conducted in accordance with PRISMA 2020 reporting standards of studies published from 2000 to 2025 in PubMed, Embase, and Scopus identified key determinants of meniscal healing. Five consistent predictors-patient age, vascularity, tear morphology, associated pathology, and activity profile-were weighted through a two-round modified Delphi consensus among ten experienced knee surgeons. The resulting 0-9-point MPS was incorporated into a stepwise decision tree linking lesion morphology, biological context, and surgical strategy. Conceptual validation used 50 simulated cases and a retrospective cohort of 45 patients to test agreement between algorithm recommendations and expert surgical decisions. RESULTS: The MPS achieved 86% concordance with expert judgment in simulation and 84% agreement in clinical validation. In this retrospective exploratory cohort, cases in which surgical management was concordant with MPS recommendations demonstrated higher mean IKDC scores at 24&#xa0;months and lower observed reoperation rates. These findings should be interpreted as associative rather than causal, as treatment allocation was not controlled and discordant cases may have represented inherently more complex pathology. CONCLUSION: The MPS represents an evidence-informed decision-support framework designed to systematize reparability assessment. While exploratory analyses suggest structural coherence with expert reasoning, prospective implementation and external validation are required before clinical adoption as a predictive tool. LEVEL OF EVIDENCE: conceptual model with exploratory validation.

Humans

Duration and characteristics of hypoglycemia with once-weekly insulin efsitora alfa versus once-daily basal insulins in adults with type 2 diabetes: Exploratory safety analysis of QWINT 2-4.

AIMS: Efsitora is a novel once-weekly basal insulin. Efsitora demonstrated similar efficacy and safety compared with once-daily basal insulin comparators across four phase 3 clinical trials for type 2 diabetes. This exploratory safety analysis further characterizes hypoglycemia events in the efsitora and once-daily treatment groups in three of these trials (QWINT-2, -3, and -4). METHODS: Median duration of hypoglycemia events was assessed with masked continuous glucose monitoring. Incidence of persistent-recurrent (PR) hypoglycemia was assessed by investigators and by a pre-specified algorithm using SMBG e-diary data. Factors contributing to hypoglycemia, characteristics of hypoglycemia, and treatment methods were reported by participants and assessed across groups. RESULTS: Across the three trials, durations of hypoglycemic events for efsitora vs once-daily basal insulin comparators were: Level 1 and 2 [<70&#xa0;mg/dL]: 40-42.5 vs 40&#xa0;min; Level 2 [<54&#xa0;mg/dL]: 35-39.9 vs 35&#xa0;min. Few incidences of PR hypoglycemia were reported for efsitora or once-daily comparators. No major descriptive differences were observed between hypoglycemia contributing factors, characteristics, or treatment methods in efsitora and once-daily treatment groups. CONCLUSIONS: No clinically relevant differences were observed between the duration or characteristics of hypoglycemic events in efsitora and once-daily treatment groups in the QWINT-2, -3, and -4 trials.

Adult

Meta-PseU: A meta-classifier for robust prediction of RNA pseudouridine modification sites from long sequences.

BACKGROUND AND OBJECTIVES: Pseudouridine (&#x3a8;) represents one of the most abundant and conserved RNA modifications. &#x3a8; provides an additional hydrogen-bond donor that enhances RNA structural stability and modulates translation. It participates in diverse biological processes, including RNA-protein interactions, splicing, translational control, and stress responses. Aberrant pseudouridylation is implicated in cancer, neurodegenerative disorders, and autoimmune diseases. Despite its biological importance, experimental identification of &#x3a8; sites remains time-consuming and costly, limiting the feasibility of transcriptome-wide profiling. Computational approaches have therefore become essential complements to experimental techniques. However, state-of-the-art machine-learning and deep-learning predictors often suffer from limited generalizability due to small training datasets. To overcome these issues, we aim at constructing new long-sequence datasets and developing a novel &#x3a8; site predictor. METHODS: New long-sequence datasets were constructed as benchmarks for RNA &#x3a8;-site prediction. The &#x3a8; modification sites in RMBase 3.0 were mapped to the reference genomes across three species of human, mouse, and yeast, and the RNA sequences with a length of 201 were generated by extending the upstream and downstream from the mapped, central sites. To eliminate sequence redundancy, the sequences were clustered using CD-HIT with a 70% sequence identity threshold. We developed Meta-PseU, a logistic regression-based meta-classifier that considered 118 machine learning and deep learning classifiers. The datasets and programs are freely accessible at https://github.com/kuratahiroyuki/MetaPseU. RESULTS: By optimizing model configuration, we proposed the Meta-PseU model stacking 32 machine learning and deep learning classifiers out of 118 classifiers. Meta-PseU substantially improved model generalizability, overcoming a key limitation of existing approaches. It greatly outperformed state-of-the-art predictors and achieved increasing accuracy with increasing sequence length. CONCLUSIONS: Long-sequence datasets were newly constructed as benchmarks for RNA &#x3a8;-site prediction. Meta-PseU offers a new framework for robust &#x3a8;-site identification by using long sequences.

Pseudouridine

Machine learning vs. traditional methods for predicting postoperative cardiac complications after non-cardiac surgery: a systematic review and Bayesian network meta-analysis.

INTRODUCTION: Accurate prediction of peri-operative cardiac complications is critical to optimise pre-operative decision-making. Traditional risk prediction scores, such as the Revised Cardiac Risk Index, show only modest discrimination. Machine learning can model complex, non-linear relationships but their predictive performance compared with traditional scores remains unclear. METHODS: We performed a systematic review and Bayesian network meta-analysis. The primary outcome was postoperative adverse cardiac events following non-cardiac surgery. Prediction models were assessed relative to the Revised Cardiac Risk Index. As many studies evaluated multiple versions of each model type, the highest performing ('best version') and lowest performing ('worst version') results were analysed. Models were ranked using the surface under the cumulative ranking curve (SUCRA). RESULTS: Thirteen studies evaluating 54 models and 927,113 patients were included. Machine learning approaches generally outperformed traditional risk scores. Automated machine learning ranked highest (SUCRA 96.6) showed the greatest improvement in the best version analysis (mean difference (MD) 0.28 (95%CrI 0.16-0.40)) and remained superior in the sensitivity analysis (MD 0.30 (95%CrI 0.14-0.45)). Gradient boosting models showed superior performance over the Revised Cardiac Risk Index across analysis (best version: MD 0.20 (95%CrI 0.14-0.26), worst version: MD 0.18 (95%CrI 0.12-0.25), SUCRA 82.4). The Gupta Perioperative Risk for Myocardial Infarction or Cardiac Arrest score outperformed the Revised Cardiac Risk Index in the best version analysis (MD 0.16 (95%CrI 0.01-0.32)). Between-study heterogeneity was low. None of the included studies externally validated their machine learning models and only six were judged to be at low risk of bias. DISCUSSION: Most machine learning models showed better discrimination than traditional risk scores, with automated machine learning and gradient boosting models ranking highest. However, study quality, calibration reporting and absence of external validation limit immediate clinical adoption. Prospective, multicentre evaluation is required before integration of these models into peri-operative practice.

Humans

Artificial intelligence in treatment prediction for skeletal Class III malocclusion: A systematic review.

In skeletal Class III patients, treatment options range from orthodontics to orthognathic surgery. Choosing the optimal approach requires a comprehensive clinical evaluation, which may be supported by AI tools. The aim of this study was to assess the performance of AI models in predicting the need for orthognathic surgery and in identifying predictors influencing treatment decisions. A PRISMA-guided electronic database search (PubMed, Web of Science; 2009-2024; English/French) was performed to identify studies using machine learning (ML) or deep learning (DL) on cephalometric and clinical data. After screening and assessment for eligibility, 15 studies were critically appraised. Model performance was summarized using accuracy, sensitivity, specificity, and the area under the curve (AUC). ML algorithms (particularly Random Forest and XGBoost) and DL models (ResNet-based convolutional neural networks (CNNs)) achieved high accuracy for predicting surgical need. Frequently selected predictors included Wits appraisal, ANB angle, the maxillomandibular ratio (Mx/Md), overjet, and the divergence of the lower gonial angle. AI methods show promise for assisting treatment decisions in Class III malocclusion, with Random Forest and XGBoost performing well on tabular cephalometric data and CNNs on imaging. Larger, multicentre datasets and external validation are needed to improve reliability, address bias, and support clinical implementation.

Humans

Reinforcement learning-based dynamic ensemble for missense variant effect prediction and tiered prioritization of VUS.

BACKGROUND: Accurate classification of missense variants remains a challenging task despite major advances in genomics. Numerous computational models have been developed to assist in variant classification, but often require repeated integration and benchmarking efforts. Ensemble methods have been proposed to overcome the limitations of single predictors, but mostly rely on fixed, predefined weights that constrain their ability to capture interactions among predictive signals. METHODS: We present GenixRL, a dynamic ensemble framework that reformulates model fusion as a reinforcement learning optimization problem. GenixRL uses a Q-learning agent to learn a policy that dynamically weights the probabilistic outputs of complementary predictors, including BayesDel (addAF and noAF), ClinPred, and MetaRNN. Replacing static weighting with policy learning allows GenixRL to adaptively identify optimal weightings and substantially improve classification accuracy. RESULTS: In benchmark evaluation against 25 state-of-the-art predictors, GenixRL achieved an AUROC of 0.9644 on an independent ClinVar dataset. On saturation genome editing assays for BRCA1 and BRCA2, GenixRL achieved the best performance and ranked highest on 14 of 17 clinically significant genes in a zero-shot evaluation. Applied to uncertain and conflicting ClinVar variants, GenixRL enabled tiered, evidence-based prioritization of hundreds of thousands of variants as likely pathogenic or pathogenic with high confidence, supported by orthogonal population evidence from gnomAD. CONCLUSION: GenixRL advances pathogenicity prediction for missense variants and provides an adaptive ensemble that sorts variants of uncertain significance into tiered candidates for expert curation and functional validation.

Mutation, Missense

Strategies for mosaic variant calling in brain disorders.

The human brain is a genomic mosaic, where postzygotic mutations arising from embryogenesis to senescence drive diverse neurodevelopmental and neurodegenerative diseases. Because of numerous sequencing artifacts at ultralow variant allele frequencies (VAFs), detecting these variants remains a significant analytical challenge. This review focuses on single-nucleotide variants and small indels, summarizing current strategies for aligning sampling methods, including bulk, laser capture microdissection, and single-cell genomics, with the expected clonal architecture of the brain. It emphasizes that mosaic detection sensitivity is fundamentally constrained by sequencing depth, since even the most advanced algorithms cannot identify variants not physically represented in the sequencing library. The review further recommends the selection of variant calling algorithms based on validated VAF detection performance, matching tools like MuTect2 and MosaicForecast to their optimal performance ranges. Furthermore, we discuss how multitissue sampling, as emphasized by the SMaHT project, addresses the matched-control dilemma and supports accurate variant classification via cross-tissue VAF gradients. Integrating these established pipelines with multiomics modalities, including transcriptomic and epigenetic data, could advance the field toward a functional understanding of how the somatic genome impacts human brain health and disease.

Humans

Predicting ACL injury risk in athletes: A systematic review of machine learning-based models.

BACKGROUND: Early ACL injury risk identification in athletes is essential. This systematic review examines machine learning (ML) models for predicting ACL injuries, evaluating their methodological quality, performance, and reliability. METHOD: A comprehensive electronic search was conducted across PubMed, Scopus, Web of Science, and IEEE Xplore databases, supplemented by Google Scholar for grey literature, covering articles published between January 1, 2015, and August 30, 2025. Eligible studies were appraised using the Prediction Model Study Risk of Bias Assessment Tool (PROBAST) for methodological quality and risk of bias, and the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) guidelines for quality of evidence. RESULTS: Ten studies were included. PROBAST showed eight studies had moderate risk of bias and two low risk. TRIPOD found only two studies met quality criteria. ML models included logistic regression (n&#xa0;=&#xa0;5), support vector machines (n&#xa0;=&#xa0;4), k-nearest neighbor (n&#xa0;=&#xa0;3), decision trees (n&#xa0;=&#xa0;3), random forests (n&#xa0;=&#xa0;5), neural networks (n&#xa0;=&#xa0;2), linear discriminant analysis (n&#xa0;=&#xa0;1), and pre-trained CNNs (n&#xa0;=&#xa0;1). AUC ranged from 0.63 to 0.98. Accuracy (reported in six studies) ranged from 26% to 95%; however, these values should be interpreted with caution due to the absence of confidence intervals, lack of class imbalance handling, and limited external validation across studies. Tree-based ensemble methods such as random forest achieved competitive accuracy (74-86%), while SVM, a non-ensemble classifier, reported accuracy ranging from 71% to 95%; however, the highest values were obtained in studies with notably small sample sizes (n&#xa0;=&#xa0;12 to n&#xa0;=&#xa0;39), raising concerns about overfitting and generalizability. CONCLUSION: Current ML algorithms show promise for identifying athletes at high ACL injury risk and detecting relevant risk factors. Although study quality was generally satisfactory, future research should prioritize external validation and model interpretability to support clinical translation.

Humans