Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “learning curve”

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.

233 records · Page 2Linked to original sources
PubMed2026

Risk Factors and Predictive Model for Postoperative High Myopia in Children Undergoing Congenital Cataract Surgery With Intraocular Lens Implantation.

PURPOSE: To identify risk factors associated with the development of high myopia following congenital cataract surgery and to establish a robust predictive model. DESIGN: Retrospective clinical cohort study. SUBJECTS: This retrospective study included 106 pediatric patients who underwent congenital cataract surgery with primary IOL implantation (mean follow-up 8.19 years). The model was externally validated in an independent cohort of 72 patients with a mean follow-up of 7.83 years. METHODS: Preoperative and postoperative ocular biometric parameters were collected. Risk factors for postoperative high myopia were analyzed using Cox proportional hazards regression, which served as the basis for model construction. The predictive performance of the model was rigorously evaluated for discrimination and calibration. Discriminative ability was quantified using Harrell's C-index and the area under the receiver operating characteristic curve (AUC). Model calibration was assessed via calibration plots by comparing predicted probabilities with actual observed outcomes. Internal validation was performed using a bootstrapping method (500 iterations) to ensure model stability and adjust for potential overfitting. RESULTS: An initial postoperative refraction of <+0.75D, and a higher IOL Power to Axial length Ratio (IOL/AL ratio) were identified as significant risk factors for the development of postoperative high myopia. Shorter preoperative axial length was associated with a greater magnitude of postoperative myopic shift. The predictive model demonstrated robust performance, achieving a C-index of 0.711 (internal validation C-index: 0.713). The area under the receiver operating characteristic curve (AUC) values for predicting high myopia at 5 and 10 years were 0.858 and 0.745, respectively. Furthermore, calibration curves demonstrated excellent agreement between the predicted and observed outcomes throughout the follow-up period. In external validation, the model achieved a C-index of 0.825, 5-year AUC of 0.833, and 10-year AUC of 0.713. CONCLUSIONS: Our analysis established that initial postoperative refraction <+0.75D, and an elevated IOL/AL ratio are key determinants of high myopia risk following surgery. Shorter preoperative axial length was associated with a greater magnitude of postoperative myopic shift. This predictive framework provides clinicians with a practical tool to optimize preoperative IOL selection and identify high-risk infants who require vigilant myopia prevention and balanced amblyopia management.

Humans↗
PubMed2026

ORBIT: Oncogenic Representation Learning via Bi-Prototype Contrastive Learning in Hyperbolic Space for cancer driver gene identification.

Accurate identification of cancer driver genes is crucial for precision oncology but remains challenging due to the complexity of integrating heterogeneous data and modeling dynamic biological systems. To address these limitations, we propose ORBIT (Oncogenic Representation Learning via Bi-Prototype Contrastive Learning in Hyperbolic Space). Our framework synergistically fuses multi-omics profiles with functional network data using a context-adaptive graph reweighting mechanism to capture cancer-specific dynamics. The model employs a bi-prototype contrastive learning strategy within hyperbolic space, which aligns gene representations around distinct driver and non-driver semantic anchors while preserving the intrinsic hierarchy of biological networks. Comprehensive evaluations demonstrate that ORBIT achieves highly competitive stability in pan-cancer analysis while consistently outperforming state-of-the-art methods in cancer-specific predictions. Furthermore, functional enrichment analysis confirms that the model effectively segregates core cancer pathways, and drug sensitivity profiling validates the clinical relevance of the identified drivers. By integrating hyperbolic geometry with context-adaptive learning, ORBIT offers a robust and interpretable paradigm for precision medicine. The source codes and datasets are publicly accessible at https://github.com/spcho-dev/ORBIT.

Humans↗
PubMed2026

Advancing nursing education through social and emotional learning: A systematic review guided by the Collaborative for Academic, Social, and Emotional Learning framework.

BACKGROUND: With Generation Z entering the nursing workforce in growing numbers, strengthening social and emotional learning is critical for academic success, professional adaptation, and safe practice. However, the existing evidence remains fragmented because of varied interventions and inconsistent approaches. OBJECTIVES: This systematic review examined (1) the social and emotional learning essential for nursing students and nurses within the Collaborative for Academic, Social, and Emotional Learning framework, (2) their impact on educational and clinical outcomes, and (3) implications for advancing nursing education and practice. METHODS: Following Joanna Briggs Institute methodology and Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, five international (PubMed, EMBASE, CINAHL, PsycINFO, Cochrane) and three Korean (RISS, KoreaMed, KMBASE) databases were searched up to June 2025. Eighteen studies involving 2,952 participants met the inclusion criteria, including quasi-experimental quantitative studies, descriptive quantitative studies, qualitative studies, and mixed-methods studies. The methodological quality of the included studies was appraised using the Mixed Methods Appraisal Tool. RESULTS: Within the Collaborative for Academic, Social, and Emotional Learning framework, relationship skills and self-management were the most frequently studied competencies, emphasizing teamwork, communication, and stress regulation. Self-awareness and social awareness were underexplored, despite their importance in empathy, resilience, and reflective practice. Responsible decision-making was the least studied competency, despite its importance in ethical reasoning. Social and emotional learning was consistently associated with enhanced adaptation, communication, leadership, relationships, and clinical performance. Effective strategies included blended learning, simulation, reflective activities, and mentorship, which are aligned with Generation Z's learning preferences. CONCLUSION: Although social and emotional learning integration is associated with improvements in educational and clinical outcomes in nursing, current research has largely centered on relational and stress-related competencies while underrepresenting responsible decision-making. To cultivate reflective, empathetic, and ethically grounded nurses, curricula should integrate social and emotional learning through a balanced and structured approach. REGISTRATION: This study was registered on PROSPERO (ID: CRD420251005683).

Humans↗
PubMed2026

Implicit and explicit statistical learning in reading: Evidence from a randomized controlled-learning study and computational modeling.

A key challenge in reading acquisition is understanding how learners extract the complex probabilistic mappings between print, meaning, and sound. Statistical learning (SL) theory offers a mechanistic account of how such mappings are acquired, whether implicitly through exposure or explicitly through instruction. We conducted a randomized controlled-learning study in Chinese, a writing system characterized by multiple sub-lexical regularities linking orthography, semantics, and phonology. Ninety-five 2nd-3rd graders with or at risk for dyslexia were randomly assigned to one of three groups: an implicit-SL training group exposed to repeated lexical and sublexical orthography-semantics-phonology associations, an explicit-SL training group receiving the same input plus explicit instruction on the sublexical print-sound mapping, and a no-SL control group. Both SL groups outperformed controls on the characters they were trained on, as well as on untrained characters that required generalization. However, only the explicit group demonstrated abstraction of print-sound mapping to novel items. Neural network simulations further revealed distinct mechanisms supporting implicit and explicit SL, consistent with a dual-system account of reading acquisition. Together, these findings (1) clarify how implicit and explicit learning distinctly support the discovery of statistical structure in written language and (2) underscore the implicit-explicit dual learning mechanism underlying reading acquisition.

Humans↗
PubMed2026

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↗
PubMed2026

A Meta-learning-driven strategy for adulteration detection in sweet potato starch and vermicelli using Raman spectroscopy.

To address the widespread adulteration of sweet potato starch and its vermicelli with cheaper starches and overcome conventional supervised learning's dependency on large labeled datasets, this study developed a few-shot discrimination method integrating Raman spectroscopy with meta-learning. We constructed a meta-learning framework using cassava- and wheat-adulterated sweet potato starch as the source domain for training, with potato-adulterated sweet potato starch and cassava-adulterated sweet potato vermicelli as two target domains for testing. Raman spectra showed high consistency between sweet potato vermicelli and its raw starch, laying the foundation for cross-domain detection. Testing yielded comprehensive classification accuracies of 95.33% and 98.00% for the two target domains, significantly outperforming SVM, RF, and CNN (max. 85.24%). This approach effectively identifies subtle starch variety differences in complex adulteration, providing novel food quality inspection solutions and verifying the feasibility of raw material-to-finished product cross-domain detection.

Ipomoea batatas↗
PubMed2026

A multi-scale fusion model based on multi-phase contrast-enhanced CT for predicting pancreatic cancer resectability.

Purpose.Develop a multi-scale fusion model (MSFM) based on multi-phase contrast-enhanced computed tomography (CECT) to predict pancreatic cancer (PC) resectability, thereby assisting expert decision-making.Methods.This retrospective study enrolled 280 patients with PC from four institutions, which were randomly divided into a training cohort (202 patients) and an independent test cohort (78 patients). Three-phase CECT images (arterial, venous, and delayed phases) were used for modeling. The MSFM comprises two sub-networks: (1) a multi-phase fusion network for extracting cross-phase shared fusion features, (2) a phase-specific branch network for capturing phase-specific features; and a post-fusion strategy to generate the final predictive score by integrating the shared fusion features and three groups of phase-specific features. Additionally, a human-machine fusion deep learning model (HMfDL) was constructed by fusing the predictive score of the MSFM with expert assessments.Results.In the independent test, the MSFM achieved an AUC (area under the receiver operating characteristic curve) of 0.8385 (95% CI: 0.7521-0.9249), accuracy of 84.62%, sensitivity of 72.00%, and specificity of 90.57%. This performance outperformed single-phase models (AUC range: 0.7638-0.7781), two-phase models (AUC range: 0.7826-0.7864), and ten states-of-the-art classifiers (AUC range: 0.7404-0.7796). The HMfDL further improved the performance, reaching an AUC of 0.8626 (95% CI: 0.7853-0.9400), accuracy of 91.03%, sensitivity of 80.00%, and specificity of 96.23%. Notably, the HMfDL corrected 58.82% of misdiagnosis made by experts.Conclusions. The MSFM effectively fuses multi-phase CECT to enable highly accurate predictions of PC resectability, and provides valuable support for expert decision-making through HMfDL.

Humans↗
PubMed2026

Analysis of deep learning techniques in computer-aided diagnosis for meniscus injuries: a systematic literature review.

Meniscus informatics is a growing subject of study in the healthcare industry. One of the major hindrances to the healthcare system's transformation is obtaining knowledge and meaningful information from complicated, high-dimensional and diverse sources. Modern biomedical research, for instance, has seen an increase in the use of complex, dissimilar, poorly documented, and generally unstructured electronic health records, imaging, sensor data and text, even after many current techniques have been used to extract more robust and useful elements from the data for analysis. New efficient standards for building end-to-end learning models from complex data are therefore needed. Therefore, the current study aims to examine the most recent research on the use of deep learning techniques for diagnosing meniscus tears and recommend creating comprehensive and meaningful interpretable structures that might benefit the healthcare industry. We also draw attention to shortcomings and the need for better technique development, and we provide new perspectives about this exciting new development in the field.

Humans↗
PubMed2026

Whole-Genome Deep Learning Predicts Chemotherapy Response in Colorectal Cancer.

Chemotherapy response in colorectal cancer (CRC) exhibits significant heterogeneity, with current clinical predictors failing to capture complex genomic determinants of resistance. We developed a hybrid deep learning framework integrating convolutional neural networks (CNNs) and bidirectional long short-term memory (BiLSTM) networks to analyze whole-genome somatic mutations, evolutionary conservation, chromatin accessibility, and 3D genome architecture in 2,546 TCGA patients. An attention mechanism identified predictive genomic regions. The model achieved an AUC of 0.92 (95% CI: 0.89-0.94) in cross-validation and 0.88 (95% CI: 0.85-0.91) in independent validation, outperforming clinical models (&#x394;AUC = +0.18, p < 0.001). Key predictors included non-coding variants in TP53, KRAS, and PIK3CA regulatory regions. Triple-positive patients (mutations in all 3 regions) had significantly worse progression-free survival (HR = 4.7, p < 0.001). Our framework enables accurate chemotherapy response prediction and reveals novel non-coding resistance mechanisms, advancing precision oncology in CRC.

Humans↗
PubMed2026

Systematic review of machine learning approaches for predicting sickle cell crisis and mortality risk at the climate-health nexus.

BACKGROUND: Sickle cell anemia (SCA) is a severe genetic blood disorder characterized by recurrent vaso-occlusive crises and increased mortality, with the greatest burden occurring in low- and middle-income countries. Climatic and environmental conditions, including temperature variability, humidity, rainfall, air pollution, and seasonal changes, have been associated with disease exacerbation. However, the extent to which these factors have been incorporated into predictive models remains unclear. This study systematically reviews the application of machine learning (ML) models for predicting SCA crises and mortality in relation to climate and environmental factors. METHODOLOGY: The PRISMA guidelines were used, and 34 peer-reviewed studies published between 2005 and 2026 were analyzed to identify the climate variables, ML approaches employed, and predictive performance. The reviewed studies applied a range of ML techniques, including artificial neural networks, random forests, support vector machines, decision trees, logistic regression, and deep learning models. Temperature, humidity, rainfall, wind speed, air quality indicators, and seasonal patterns were the most frequently examined environmental variables. RESULTS: The findings indicate that most existing models rely predominantly on clinical and demographic data, with limited integration of climate information and inadequate representation of high-burden regions, especially Sub-Saharan Africa. Studies incorporating environmental variables reported improved predictive performance and highlighted the potential of climate-informed early warning systems for SCA management. CONCLUSION: The review recommends development of interdisciplinary, climate-aware ML frameworks, expansion of longitudinal environmental datasets, and increased research in underrepresented regions to support climate-resilient and patient-centered SCA care.

Humans↗
PubMed2026

A systematic literature review of cultural concepts taught by pharmacist preceptors during pharmacy student experiential placements.

BACKGROUND: Cultural concepts such as cultural intelligence, awareness, competency and safety are essential in guiding culturally responsive care in health professional practice. Pharmacist preceptors play a pivotal role in sharing both clinical and cultural safe practice with pharmacy students. Culturally responsive care can contribute to achieving health equity, which is especially important for Indigenous communities. AIM: To review literature on cultural concepts in pharmacist preceptorship practices, and how these concepts are taught and communicated to pharmacy students during experiential learning. METHOD: The systematic review followed the PRISMA 2020 guideline. Scopus, PubMed, and Google Scholar were used to identify articles specific to pharmacist preceptors and pharmacy students published between 2015 and 2025, and available in English. RESULTS: Three full-text articles met the inclusion criteria. Major themes and subthemes were identified; pharmacist preceptors lacked preparedness to teach cultural concepts, resulting in variability in preceptors' understanding of cultural concepts and confidence in fulfilling preceptor responsibilities, underutilised structured frameworks to guide students' learning, challenges with preceptorship due to limited resources and support, and the influence of preceptorship on student learning, which impacted students' learning and competency. CONCLUSION: Pharmacy students had minimal exposure to culturally informed pharmacist preceptorship. It is likely that pharmacist preceptors require country-specific educational resources to support culturally safe preceptorship. Future research is required to substantiate these findings, and to guide culturally responsive practice and promote equitable health outcomes in diverse populations.

Humans↗
PubMed2026

Systematic evaluation of one-dimensional-to-two-dimensional near-infrared spectroscopy transformations with deep learning for quantifying coconut sap adulteration.

Near-infrared (NIR) spectroscopy have limitations when combined with deep learning (DL) algorithms because they rely on low-dimensional datasets. Therefore, we investigated the potential of transforming one-dimensional (1D) NIR spectra into two-dimensional (2D) spectrograms using synchronous and asynchronous techniques and the continuous wavelet transform (CWT) and their effectiveness by integrating with DL for detecting adulteration in coconut sap. NIR spectra (12,500-4000&#xa0;cm-1) were collected from binary mixtures (0%-100%;w/w). The performance of all DL (convolutional neural networks-CNN, AlexNet and ResNet) models was compared with that of partial least squares (PLS). The models were ranked in the mentioned order based on their performances: 2D-CWT&#xa0;>&#xa0;2D-asynchronous > 2D-synchronous > 1D/2D-PLS. The important features of the best model can be explained and visualized using gradient-weighted-class-activation-mapping. The findings highlight that the 1D-to-2D NIR data transformation combined with DL is a highly robust approach because it addresses the feature representation gap in NIR data and effectively captures the spatial-spectral correlations.

Spectroscopy, Near-Infrared↗
PubMed2026

A systematic review of human avoidance learning: Cognition, computation, and methods.

Avoidance behaviour is fundamental for survival but can become maladaptive in clinical conditions. A large body of literature has accumulated on the dynamics of human avoidance learning. However, current theories and overviews do not provide an exhaustive account of this evidence. In this systematic review, we identify N = 116 studies on human avoidance learning. We analyse these studies with the goal of distilling robust empirical phenomena as a basis for theory-building, and examine their diagnostic value in differentiating between competing theories. We find that the evidence is difficult to reconcile with foundational two-factor and classical safety-signal accounts, and most strongly supports expectancy- and inference-based views, in which avoidance responses are selected with respect to represented consequences. At the same time, no current framework provides a complete account of the evidence: several findings point to an additional role for operant valuation, Pavlovian influences, and contextual or latent-state control over the expression of avoidance. Methodologically, we observe that the problem setting in the most common experimental paradigms is radically simpler than real-world avoidance and therefore unlikely to expose the limits of inferential or reflective mechanisms. Consequently, we argue that paradigms with greater computational demands and more realistic action affordances are required to identify the mechanisms underlying avoidance learning. Collectively, these insights provide a foundation for theoretical refinement, computational modelling, and methodological innovation, with implications for advancing interventions targeting maladaptive avoidance.

Humans↗
PubMed2026

Quantitative assessment of the fingerprint evidential value using machine learning.

Fingerprints as physical evidence have long supported criminal investigation and adjudication. In practice, however, fingerprint identification relies mainly on examiners' experience. Furthermore, expert opinions tend to be categorical, even though the opinions with the same conclusion could differ substantially in evidential strength. To quantitatively assess fingerprint evidential value, this study proposes a machine learning-based framework as an interpretable decision-support tool. A lightweight residual one-dimensional convolutional neural network was constructed, incorporating channel recalibration and a similarity-driven attention mechanism to learn adaptive contribution weights for different matched minutiae (minutiae for short). Controlled experiments revealed that the predicted evidential value increased with the number of minutiae and was significantly influenced by the quality of minutiae. With 10 minutiae, the mean predicted scores were 4.49, 7.00, and 9.09 for blurred, moderately blurred, and clear minutiae, respectively. Multiple regression analysis indicated that replacing a pair of blurred minutiae with a pair of clear minutiae increased the score by 0.492, whereas replacing it with a pair of moderately blurred minutiae increased the score by only 0.216. By mapping predicted scores to graded levels of evidential strength, the framework contributes to a paradigm shift from categorical expert opinions to graded ones, helping courts evaluate fingerprint evidence more scientifically.

Humans↗
PubMed2026

Machine learning-assisted Mn-N-C nanozyme colorimetric sensor array for trace-level detection of biogenic amines in meat.

Accurate detection of biogenic amines (BAs) in meat remains challenging due to their high structural similarity and co-occurrence. Herein, an Mn-N-C nanozyme was synthesized via a metal-organic framework confined pyrolysis strategy, possessing excellent oxidase (OXD)- and peroxidase (POD)-like activities. The dual enzyme-like activity showed Km values of 0.1584&#xa0;mM (OXD) and 0.1498&#xa0;mM (POD), respectively, in detection system. Leveraging these properties, a colorimetric sensor array was constructed, enabling the detection of four representative BAs within a concentration range of 2-10&#xa0;ppm with 100% classification accuracy. In addition, a concentration independent recognition model based on an artificial neural network was developed to address signal nonlinearity interference in meat. The integrated system achieved accurate trace-level identification of BAs in perishable fish, pork, and chicken, demonstrating its applicability for early-stage BAs monitoring and quality deterioration warning during storage and transportation.

Biogenic Amines↗
PubMed2026

Machine learning-based prediction of unplanned readmission and construction of an online calculator for elderly patients with mild ischemic stroke.

OBJECTIVE: To screen for independent risk factors for unplanned readmission in elderly patients with mild ischemic stroke, and to construct and validate an online risk prediction calculator based on an interpretable machine learning model, thereby providing a promising practical tool for accurate clinical assessment of 30&#x2011;day all&#x2011;cause unplanned readmission risk in this population. METHODS: A prospective cohort study was conducted, including 1050 patients aged&#xa0;&#x2265;&#xa0;60&#xa0;years with mild ischemic stroke admitted between August 2023 and September 2024. Participants were randomly divided into a training set (840 cases) and a test set (210 cases) at a ratio of 8:2. Risk factors were screened by univariate analysis and multivariable Logistic regression. Four machine learning models, namely LightGBM, XGBoost, Random Forest, and K&#x2011;Nearest Neighbors (KNN), were developed and their performance was evaluated using AUC, accuracy, sensitivity, and specificity as metrics. The SHAP framework was used for interpretability analysis, and an online calculator was subsequently developed based on the optimal model. RESULTS: Univariate analysis showed significant differences (P&#xa0;<&#xa0;0.05) in 13 factors including age, smoking, AIP, TyG index, HALP score, etc. Multivariable Logistic regression identified age (OR&#xa0;=&#xa0;9.752), smoking (OR&#xa0;=&#xa0;5.171), AIP (OR&#xa0;=&#xa0;6.691), TyG index (OR&#xa0;=&#xa0;4.393), HALP score (OR&#xa0;=&#xa0;2.831), and&#xa0;&#x2265;&#xa0;2 comorbidities (OR&#xa0;=&#xa0;3.664) as independent risk factors. All four machine learning models demonstrated good predictive performance. Based on a comprehensive evaluation of multiple metrics and computational efficiency, the LightGBM model exhibited the best predictive performance (AUC&#xa0;=&#xa0;0.884, accuracy&#xa0;=&#xa0;0.829, sensitivity&#xa0;=&#xa0;0.812, specificity&#xa0;=&#xa0;0.875). SHAP analysis showed that age, AIP, TyG index, smoking, and HALP score were key predictors. An online calculator developed based on this model enables individualized risk predictions. CONCLUSION: Key risk factors associated with 30&#x2011;day unplanned readmission in elderly patients with mild ischemic stroke were identified. The LightGBM model demonstrated high predictive accuracy, and together with the interpretability analysis and online calculator, offers a practical tool to support clinical risk assessment. However, this tool requires future external validation.

Humans↗
PubMed2026

Diagnostic performance of machine learning models for malignant and non-malignant pleural effusion: Systematic review and meta-analysis.

BACKGROUND: Accurately distinguishing malignant pleural effusion (MPE) from non-malignant pleural effusion is clinically important, but the generalisability and methodological quality of machine-learning (ML) models remain uncertain. METHODS: We searched eight databases to 23 April 2026. Diagnostic performance was pooled using random-effects and Reitsma bivariate models, and study quality was assessed using PROBAST+AI. RESULTS: Forty-two studies were included; 17 contributed to the AUC meta-analysis and 14 to the bivariate analysis. The pooled AUC was 0.90 (95&#xa0;% CI 0.85-0.94; 95&#xa0;% prediction interval 0.62-0.98), with sensitivity of 0.80 (95&#xa0;% CI 0.77-0.83) and specificity of 0.87 (95&#xa0;% CI 0.79-0.92). Only nine studies reported external, temporal or independent validation. Externally validated studies had a lower pooled AUC than studies without external validation (0.83 vs 0.92), with lower specificity observed in the two externally validated studies contributing sensitivity and specificity data. All 42 development assessments had high overall quality concerns, and all 42 model evaluations were judged at high risk of bias. CONCLUSIONS: ML models showed good apparent accuracy for distinguishing MPE from non-MPE, but the evidence was limited by substantial heterogeneity, high risk of bias and scarce external validation. The pooled estimates reflect the average performance of different selected models rather than the expected accuracy of a single clinical test. ML models should be regarded as adjuncts to existing diagnostic pathways until they are confirmed by rigorous multicentre prospective external validation and clinical-impact studies.

Humans↗