Training Set Research Articles

Comprehensive multicentre retrospective analysis for predicting isocitrate dehydrogenase‐mutant lower‐grade gliomas

AbstractObjectiveTo differentiate glioma grading and determine isocitrate dehydrogenase (IDH) mutation status, which are crucial for prognosis assessment and treatment planning in glioma patients.MethodsThis retrospective study included patients diagnosed with adult diffuse glioma from 1 January, 2018 to 31 July, 2023 in two independent institutions. It documented and analysed clinical and radiographic features. A nomogram model was constructed using stepwise regression to predict lower‐grade gliomas and IDH mutation status.ResultsA total of 383 adult patients with diffuse glioma were included in the study, with Cohort A (297 patients) serving as the training set and Cohort B (86 patients) serving as the validation cohort. Consistent with previous reports, the Hyper fluid‐attenuated inversion recovery (FLAIR) rim sign exhibited higher sensitivity in lower‐grade gliomas for IDH mutant gliomas compared with the T2‐FLAIR mismatch sign. However, the Hyper FLAIR rim sign was also present in Grade 4 gliomas, and thus, the T2‐FLAIR mismatch sign exhibited better clinical efficacy in predicting glioma grade and IDH mutation compared with the Hyper FLAIR rim sign in clinical applications. Meanwhile, preoperative magnetic resonance spectroscopy (MRS) indicators, particularly the Cho/Cr ratio, have shown excellent performance in predicting glioma grade and IDH mutation status. The nomogram developed through stepwise regression demonstrated excellent predictive capabilities in distinguishing glioma grade and IDH mutation status.InterpretationCombining imaging and molecular features, the predictive model established in this study offers a reliable non‐invasive tool for predicting glioma grading and IDH mutation status, aiding the clinical decision‐making process and improving patient management.

A simplified pyrazinamidase test for Mycobacterium tuberculosis pyrazinamide antimicrobial susceptibility testing

ABSTRACT Pyrazinamide (PZA) is an important first-line drug for tuberculosis (TB) treatment by eradicating the persisting Mycobacterium tuberculosis complex (MTBC). Due to cost and technical challenges, end TB strategies are hampered by the lack of a simple and reliable culture-based PZA antimicrobial susceptibility testing (AST) for routine use. We initially developed a simplified chromogenic pyrazinamidase (PZase) test in the TB reference laboratory using a training set MTBC isolates with various drug-resistant profiles, and validated its performance using consecutive BACTEC MGIT 960 (MGIT)-culture-positive culture in 10 clinical laboratories. The pncA gene Sanger sequencing results were used as the reference, and compared to the MGIT-PZA AST. Differential diagnosis of Mycobacterium bovis was conducted using patented in-house real-time PCR. Of the 106 training isolates, the PZase test and MGIT-PZA AST showed 100.0% and 99.1% concordance as compared to Sanger sequencing, respectively. We found 32.1% (34/106) isolates harbored pncA mutations, including one isolate with silent mutation S65S. For validation, 1,793 clinical isolates were tested including 150 duplicate isolates from specimens of the same cases and 16 isolates with uncharacterized drug resistance (UDR)-associated mutations. Excluding duplicated and UDR isolates, we identified 2.6% (43/1,627) PZA-resistant isolates, including 1.3% (21/1,627) M . bovis isolates. The kappa values were 0.851–1.000. In addition, the accuracy of the PZase test conducted by 10 laboratories was 98.5%–100.0%. Our simplified PZase test demonstrated high concordance with Sanger sequencing and MGIT-PZA AST. Integrating the PZase test into routine first-line AST is effortless and represents an improvement in laboratory services for ending TB. IMPORTANCE We developed and validated a simple pyrazinamidase (PZase) test for pyrazinamide (PZA) antimicrobial susceptibility testing (AST). Our results demonstrated that the PZase test had high agreement with the pncA gene sequencing and MGIT-PZA AST. Integrating PZase test into routine AST is effortless and represents an improvement in laboratory services for ending TB.

Constructing training sets for genomic selection to identify superior genotypes in candidate populations.

Approaches for constructing training sets in genomic selection are proposed to efficiently identify top-performing genotypes from a breeding population. Identifying superior genotypes from a candidate population is a key objective in plant breeding programs. This study evaluates various methods for the training set optimization in genomic selection, with the goal of enhancing efficiency in discovering top-performing genotypes from a breeding population. Additionally, two approaches, inspired by classical optimal design criteria, are proposed to expand the search space for the best genotypes and compared with methods focusing on maximizing accuracy in breeding value prediction. Evaluation metrics such as normalized discounted cumulative gain, Spearman's rank correlation, and Pearson's correlation are employed to assess performance in both simulation studies and real trait analyses. Overall, for candidate populations lacking a strong subpopulation structure, a ridge regression-based method, referred to as is recommended. For candidate populations with a strong subpopulation structure, a heuristic-based version of generalized coefficient of determination and a D-optimality-like method that maximizes overall genomic variation are preferred approaches for the primary objective of plant breeding. For populations with a large number of candidates, a proposed ranking method ( ) can first be used to down-scale the candidate population, after which a heuristic-based method is employed to identify the best genotypes. Notably, the proposed has been verified to be equivalent to the original version, known as , but its implementation is much more computationally efficient.

Estimating hair density with XGBoost

AbstractObjectivesHair density estimation is crucial in dermatology and trichology; however, manual counting is time‐consuming and error‐prone. Although automated approaches have been developed using image processing, neural networks, and deep learning, creating a robust and widely applicable method remains challenging. This study explored the use of XGBoost to estimate hair density with the aim of developing a more accurate and versatile approach.MethodsThe study utilized 895 scalp images to extract features and developed an XGBoost model for hair density estimation using 745 images to train the model and testing its performance on 150 images to evaluate the accuracy, error rate, and scatter plot.ResultsThe XGBoost model outperformed previous methods, achieving 89.5% accuracy on the training set and 95.3% accuracy on the test set. This surpassed the results of Kim et al. (52.4%), Urban et al. (79.6%), and Sacha et al. (88.2%) for the test set.ConclusionThe XGBoost algorithm proved to be effective for automated hair density estimation, achieving an accuracy of 95.3% on the test set. This approach, which focusses on scalp coverage and erosion features, can streamline and improve the objectivity of clinical hair analysis.

Are Nomograms Useful for Predicting Sentinel Lymph Node Status in Melanoma Patients?

ABSTRACTBackground and ObjectivesClinical nomograms have been developed to predict sentinel lymph node (SLN) status in early‐stage melanoma patients, but the clinical utility of these tools remains debatable. We created and validated a nomogram using data from a randomized clinical trial and assessed its accuracy against the well‐validated Melanoma Institute Australia (MIA) nomogram.MethodsWe developed our model to predict SLN status using logistic regression on clinicopathological patient data from the Multicenter Selective Lymphadenectomy Trial‐I. The model was externally validated using the National Cancer Database (NCDB) data set, and its performance was compared to that of the MIA nomogram.ResultsOur model had good discrimination between positive and negative SLNs, with a training set area under the curve (AUC) of 0.706 (0.661–0.751). Our model achieved an AUC of 0.715 (0.706–0.724) compared to 0.723 (0.715–0.731) with the MIA model, using the NCDB set.ConclusionOur model performed similarly to the MIA model, confirming that despite using different clinical features and data sets, no clinical nomogram is currently accurate enough for clinical use.

Development and validation of a predictive risk tool for VTE in women with breast cancer under chemotherapy: a cohort study in China.

The incidence of venous thromboembolism (VTE) is significantly elevated in breast cancer patients, with a three-to-fourfold increase, and further escalates to sixfold in those undergoing chemotherapy. This study aims to identify the risk factors for VTE and develop a Nomogram risk prediction model distinct from the traditional Khorana score. Univariate Cox regression analysis assessed the impact of each variable on the occurrence of VTE, while stepwise multivariate Cox regression analysis identified independent predictors. Based on these results, we constructed a Nomogram model. The model's performance was validated using the C-index, receiver-operating characteristic curve (ROC), calibration curves, and decision curve analysis (DCA). Comparisons were made with the Khorana score to evaluate the practical application value. Out of the 903 patients, 108 (11.96%) developed VTE. Cox regression analysis identified that Stage, undergoing surgery, age, white blood cells (WBC), D-dimer, and carcinoembryonic antigen (CEA) were significant risk factors for VTE (P < 0.05). The Nomogram model's C-index was 0.77 (95% CI 0.72-0.83) in the training set and 0.76 (95% CI 0.69-0.84) in the validation set. The model demonstrated excellent predictive accuracy and generalizability on the receiver-operating characteristic (ROC) curves and calibration curves. Compared to the traditional Khorana score, the Nomogram model showed superior predictive accuracy and greater clinical benefit. This study established a VTE risk prediction model for breast cancer patients undergoing chemotherapy. The model is characterized by its intuitive and straightforward application, making it highly suitable for rapid VTE risk assessment in clinical practice.

Machine learning model based on SERPING1, C1QB, and C1QC: A novel diagnostic approach for latent tuberculosis infection

AbstractBackgroundLatent tuberculosis infection (LTBI) is a significant source of active tuberculosis (ATB), yet distinguishing between them is challenging because specific biomarkers are lacking.MethodsWe analyzed four microarray datasets (GSE19491, GSE37250, GSE54992, GSE28623) from the gene expression omnibus to identify differentially expressed genes (DEGs). Using protein–protein interaction (PPI) networks and LASSO‐SVM algorithms, we selected three candidate biomarkers and evaluated their diagnostic efficacy. The expression levels of core genes were validated by RNA sequencing of healthy, ATB, and LTBI groups in a real‐world cohort. We conducted Gene Ontology and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses, predicted shared upstream miRNAs, constructed miRNA–hub and transcription factor (TF)–hub gene networks, and performed immune infiltration analysis.ResultsThree hub genes (SERPING1, C1QC, C1QB) were identified from 45 DEGs by PPI networks and machine learning screening. The diagnostic model based on the three hub genes had an area under the curve (AUC) value of 0.843 in the training set GSE19491 and 0.865 in the validation set GSE28623. Real‐world transcriptome sequencing confirmed the expression trends of the hub genes across healthy, LTBI, and ATB groups. GO analysis showed that the 45 hub genes were primarily associated with immune inflammatory responses and pattern recognition receptors, whereas KEGG analysis indicated enrichment in complement and coagulation cascades. The miRNA–hub and TF–hub gene network analysis identified nine miRNAs and the zinc finger TF GATA2 as potential co‐regulators of SERPING1, C1QC, and C1QB. Immune cell infiltration analysis identified significant differences in the immune microenvironment between LTBI and ATB, with macrophages and natural killer cells showing significant correlations with tuberculosis infection.ConclusionThe diagnostic model with SERPING1, C1QC, and C1QB shows promise in distinguishing LTBI from ATB, indicating its potential as a diagnostic tool.

Research on information leakage in time series prediction based on empirical mode decomposition.

Time series analysis predicts the future based on existing historical data and has a wide range of applications in finance, economics, meteorology, biology, engineering, and other fields. Although the combination of decomposition techniques and machine learning algorithms can effectively solve the problem of predicting nonstationary sequences, this kind of decomposition-integration-prediction strategy of the prediction method has serious defects. After the decomposition of the division of the training set and the test set, the information of the test set in the process of decomposition of the information leakage ultimately shows a high accuracy of the prediction of the illusionary. This paper proposes three improvement strategies for this type of "information leakage" problem: sliding window decomposition (SW-EMD), single training and multiple decomposition (STMP-EMD), and multiple training and multiple decomposition (MTMP-EMD). They are combined with a bidirectional multiscale temporal convolutional network (MSBTCN), bidirectional long- and short-term memory network (BiLSTM), and attention mechanism (DMAttention), which introduces a dependency matrix based on cosine similarity to be applied to water quality prediction. The experimental results show that the model achieves good performance in the prediction of three water quality indicators (pH, DO and KMnO4), and the accuracies of the three models proposed in this paper are improved by 1.958% and 0.853% in terms of the RMSE and MAPE, respectively, compared with those of the mainstream LSTM models. The key contributions of this study include the following: (1) three methods are proposed to improve the class EMD decomposition, which can effectively solve the problem of "information leakage" that exists in the current models via class EMD decomposition; (2) the CEEMDAN-MSBTCN-BiLSTM-DMAttention model structure is innovated by combining improved class EMD decomposition methods; and (3) the three improved decomposition methods proposed in this paper can effectively solve the problem of "information leakage" and optimize the prediction model at the same time. This study provides an effective experimental method for water quality prediction and can effectively address the problem of "overfitting" models via class EMD decompositions during model training and testing.

Application of artificial intelligence in VSD prenatal diagnosis from fetal heart ultrasound images.

Developing a combined artificial intelligence (AI) and ultrasound imaging to provide an accurate, objective, and efficient adjunctive diagnostic approach for fetal heart ventricular septal defects (VSD). 1,451 fetal heart ultrasound images from 500 pregnant women were comprehensively analyzed between January 2016 and June 2022. The fetal heart region was manually labeled and the presence of VSD was discriminated by experts. The principle of five-fold cross-validation was followed in the training set to develop the AI model to assist in the diagnosis of VSD. The model was evaluated in the test set using metrics such as mAP@0.5, precision, recall, and F1 score. The diagnostic accuracy and inference time were also compared with junior doctors, intermediate doctors, and senior doctors. The mAP@0.5, precision, recall, and F1 scores for the AI model diagnosis of VSD were 0.926, 0.879, 0.873, and 0.88, respectively. The accuracy of junior doctors and intermediate doctors improved by 6.7% and 2.8%, respectively, with the assistance of this system. This study reports an AI-assisted diagnostic method for VSD that has a high agreement with manual recognition. It also has a low number of parameters and computational complexity, which can also improve the diagnostic accuracy and speed of some physicians for VSD.

Radiomics-based predictive model for preoperative risk classification of gastrointestinal stromal tumors using multiparametric magnetic resonance imaging: aretrospective study.

The aim of this study was to develop and assess aradiomics model utilizing multiparametric magnetic resonance imaging (MRI) for the prediction of preoperative risk assessment in gastrointestinal stromal tumors (GISTs). An analysis was performed retrospectively on agroup of 121 patients who received ahistological diagnosis of GIST. They were then divided into two sets, with85 in the training set and 36 in the validation set through random partitioning. Radiomics features from five MRI sequences, totaling 600 per patient, were extracted and subjected to feature selection utilizing arandom forest algorithm. The discriminatory efficacy of the models was evaluated through receiver operating characteristic (ROC) and precision-recall (P-R) curve analyses. Model calibration was assessed via calibration curves. Subgroup analysis was performed on GISTs with apathological maximum diameter equal to or less than 5 cm. Furtherly, Kaplan-Meier (K-M) curves and log-rank tests were used to compare the differences in survival status among different groups. Cox regression analysis was employed to identify independent prognostic factors and to construct aprognostic prediction model. The clinical model (ModelC) displayed limited predictive efficacy in the context of GIST. Conversely, aradiomics model (ModelR) incorporating five parameters exhibited robust discriminative capabilities across both the training and validation sets, yielding area under the ROC curve (AUC) values of 0.893 (95% confidence interval [CI]: 0.807-0.949) and 0.855 (95% CI: 0.732-0.978), respectively. The F1max scores derived from the P‑R curves were 0.741 and 0.842 for the training and validation sets, respectively. Noteworthy was the exclusion of the two-dimensional tumor diameter and tumor location when constructing ahybrid model (ModelCR) that amalgamated radiomics and clinical features. ModelR demonstrated asubstantially enhanced discriminative capacity in the training set compared with ModelC (p < 0.005). The net reclassification improvement (NRI) corroborated the superior performance of ModelR over ModelC, thereby enhancing diagnostic accuracy and clinical applicability. Patients in the high-risk group had significantly worse recurrence-free survival (RFS, p < 0.001) and overall survival (OS, p = 0.004), and the radiomics signature is an independent risk factor for RFS. The extended model incorporating the radiomics signature outperformed the baseline model in terms of risk assessment accuracy (p < 0.001). Our investigation underscores the value of integrating radiomics analysis in conjunction with machine learning algorithms for prognostic risk stratification in GIST, presenting promising implications for informing clinical decision-making processes as well as optimizing management strategies.

Development and validation of a nomogram for predicting early acute hydrocephalus after spontaneous intracerebral hemorrhage: a single-center retrospective study.

Acute hydrocephalus is a severe complication that may occur early after an intracerebral hemorrhage (ICH). However, clinical factors predicting the occurrence of acute hydrocephalus have rarely been studied. This study aimed to establish a nomogram model to predict early acute hydrocephalus after ICH. We retrospectively analyzed the data of 930 patients with ICH who were treated at our hospital between January 2017 and May 2024. Least absolute shrinkage and selection operator (LASSO) regression analysis was used to screen for risk factors for acute hydrocephalus, and stepwise logistic regression analysis was used to construct the prediction model, which was visualized using a nomogram. Data were randomly divided into training (n = 652) and test (n = 278) sets at a 7:3 ratio. A total of 930 patients were included, of whom 123 (13.2%) developed acute hydrocephalus within 6h of being diagnosed with ICH. Univariate analysis revealed that 11 indicators were associated with acute hydrocephalus. In the training set, LASSO and stepwise logistic regression analyses identified four independent risk factors that were used to establish a prediction model. These were the modified Graeb score, age, infratentorial hemorrhage > 15 mL, and thalamic hemorrhage > 15 mL. A graphical nomogram was then developed. The area under the receiver operating characteristic (ROC) curve was 0.974 (95% confidence interval 0.961-0.987). In the Hosmer-Lemeshow test, the p-value was 0.887. The mean absolute error of the calibration plot was 0.012. The decision curve analysis (DCA) validated the fitness and clinical application value of this nomogram. Internal validation showed the test set was in good accordance with the training set. The nomogram prediction model showed good accuracy and could be used to predict the risk of early acute hydrocephalus after ICH, thereby aiding neurologist in making rapid clinical decisions.

Immune-oncology targets and therapeutic response of cell pyroptosis-related genes with prognostic implications in neuroblastoma.

Construction of a neuroblastoma (NB) prognostic predictive model based on pyroptosis-related genes (PRGs) to improve individualized management of NB patients. The NB cohort GSE49711 was obtained from the Gene Expression Omnibus (GEO) database, and a total of 498 patients were enrolled into the study, which were randomized into a training set and a test set at a ratio of 1:1, with 250 patients in the training set and 248 patients in the test set. A risk prediction model was constructed using the training set, and the GSE49711 cohort and test set were used as internal validation to verify the reliability of the model. Independent predictors associated with prognosis were screened using univariate and multivariate COX regression analyses, and risk score models were constructed. Single-cell gene set enrichment analysis (ssGSEA) was used to assess the relationship between PRGs and the tumor immune microenvironment. Nomograms were constructed to extend the clinical usability of the model and the reliability of the model was verified using ROC curves and calibration curves. Protein interaction networks of risk genes were mapped using the String database, and the expression of PRGs in NB cell lines was staged using the CCLE database. A prognostic model was first developed with the training set: the risk score formula was (-0.30 × GSDMB) + (-0.46 × IL-18) + (-0.21 × NLRP3) + (0.56 × AIM2). Patients were categorized into high- and low-risk groups based on the median risk score value. Survival analysis showed that NB patients in the high-risk group had a significantly lower survival rate than those in the low-risk group (P < 0.001). In both the GSE49711 overall cohort and the test cohort, survival analyses showed that patients in the high-risk group had significantly lower survival than those in the low-risk group (P < 0.001). Single-cell gene set enrichment analysis was used to assess the relationship between PRGs and the tumor immune microenvironment. Time-dependent ROC curves assessed the predictive performance of the nomogram in 5-, 7.5-, and 10-year survival with areas under the curve (AUC) of 0.843, 0.802 and 0.797, respectively. The calibration curves show good clinical predictive performance for nomograms. The results suggest that PRGs may serve as a novel prognostic marker for NB patients to provide new immunotherapeutic targets for the clinical treatment of NB patients.

Magnetic Resonance Imaging Texture Analysis Based on Intraosseous and Extraosseous Lesions to Predict Prognosis in Patients with Osteosarcoma

Objectives: To construct an optimal magnetic resonance imaging (MRI) texture model to evaluate histological patterns and predict prognosis in patients with osteosarcoma (OS). Methods: Thirty-four patients underwent pretreatment MRI and were diagnosed as having OS by surgical resection or biopsy between September 2008 and June 2018. Histological patterns and 3-year survival were recorded. Manual segmentation was performed in intraosseous, extraosseous, and entire lesions on T1-weighted, T2-weighted, and contrast-enhanced T1-weighted images to extract texture features and perform principal component analysis. A support vector machine algorithm with 3-fold cross-validation was used to construct and validate the models. The area under the receiver operating characteristic curve (AUC) was calculated to evaluate diagnostic performance in evaluating histological patterns and 3-year survival. Results: Eight patients were chondroblastic and the remaining twenty-six patients were non-chondroblastic patterns. Twenty-seven patients were 3-year survivors, and the remaining seven patients were non-survivors. In discriminating chondroblastic from non-chondroblastic patterns, the model from extraosseous lesions on the T2-weighted images showed the highest diagnostic performance (AUCs of 0.94 and 0.89 in the training and validation sets). The model from intraosseous lesions on the T1-weighted images showed the highest diagnostic performance in discriminating 3-year non-survivors from survivors (AUCs of 0.99 and 0.88 in the training and validation sets) with a sensitivity, specificity, positive predictive value, and negative predictive value of 85.7%, 92.6%, 75.0%, and 96.2%, respectively. Conclusions: The texture models of extraosseous lesions on T2-weighted images can discriminate the chondroblastic pattern from non-chondroblastic patterns, while the texture models of intraosseous lesions on T1-weighted images can discriminate 3-year non-survivors from survivors.

Enhancing bone metastasis prediction in prostate cancer using quantitative mpMRI features, ISUP grade and PSA density: a machine learning approach.

Bone metastasis is a critical complication in prostate cancer, significantly impacting patient prognosis and quality of life. This study aims to enhance bone metastasis prediction using machine learning (ML) techniques by integrating dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) perfusion features, International Society of Urological Pathology (ISUP) grade, and prostate-specific antigen (PSA) density. A retrospective analysis was conducted on 122 patients with histopathologically confirmed prostate cancer who underwent multiparametric prostate magnetic resonance imaging (mpMRI). Quantitative mpMRI features, PSA density, and ISUP grades were extracted and normalized. The dataset was balanced using oversampling and divided into training (70%) and test (30%) sets. Various ML models were developed and evaluated using area under the curve (AUC) metrics. Bone metastases were present in 26 patients (21.3%) at diagnosis. IAUGC and MaxSlope showed a statistically significant association with bone metastasis (p = 0.035, p = 0.050 respectively). The optimal PSA density cut-off value of 0.24 yielded a sensitivity of 0.88, specificity of 0.60, and AUC of 0.77. Machine learning models were developed using the dataset created with IAUGC, MaxSlope, ISUP grade, and PSA density values. Among the ML models, XGBoost demonstrated superior performance with validation and test AUCs of 91.5% and 92.6%, respectively, along with high precision (93.3%) and recall (93.1%). Integrating quantitative mpMRI features, ISUP grade, and PSA density through machine learning algorithms, particularly XGBoost, significantly improves the accuracy of bone metastasis prediction in prostate cancer patients. This approach can potentially reduce the need for additional imaging modalities and associated radiation exposure.

Advancing EGFR mutation subtypes prediction in NSCLC by combining 3D pretrained ConvNeXt, radiomics, and clinical features

PurposeThe aim of this study was to develop a novel approach for predicting the expression status of Epidermal Growth Factor Receptor (EGFR) and its subtypes in patients with Non-Small Cell Lung Cancer (NSCLC) using a Three-Dimensional Convolutional Neural Network (3D-CNN) ConvNeXt, radiomics features and clinical features.Materials and methodsA total of 732 NSCLC patients with available CT imaging and EGFR expression data were included in this retrospective study. The region of interest (ROI) was manually segmented, and clinicopathological features were collected. Radiomic and deep learning features were extracted. The instances were randomly divided into training, validation, and test sets. Feature selection was performed, and XGBoost was used to create solo models and combined models to predict the presence of EGFR and subtypes mutations. The effectiveness of the models was assessed using ROC and PRC curves.ResultsWe established the following models: ModelCNN, Modelradiomic, Modelclinical, ModelCNN+radiomic, ModelCNN+clinical, Modelradiomic+clinical, and ModelCNN+radiomic+clinical, which were based on deep learning features, radiomic features, clinical data and combinations of these, respectively. In predicting EGFR mutations, ModelCNN+radiomic+clinical demonstrated superior performance compared to other prediction models, achieving an AUC of 0.801. For distinguishing between EGFR subtypes ex19del and L858R, ModelCNN+radiomic reached the highest AUC value of 0.775.ConclusionsBoth deep learning models and radiomic signature-based models offer reasonably accurate non-invasive predictions of EGFR status and its subtypes. Fusion models hold the potential to enhance noninvasive methods for predicting EGFR mutations and subtypes, presenting a more reliable prediction approach.

Segmentation of glioblastomas via 3D FusionNet

IntroductionThis study presented an end-to-end 3D deep learning model for the automatic segmentation of brain tumors.MethodsThe MRI data used in this study were obtained from a cohort of 630 GBM patients from the University of Pennsylvania Health System (UPENN-GBM). Data augmentation techniques such as flip and rotations were employed to further increase the sample size of the training set. The segmentation performance of models was evaluated by recall, precision, dice score, Lesion False Positive Rate (LFPR), Average Volume Difference (AVD) and Average Symmetric Surface Distance (ASSD).ResultsWhen applying FLAIR, T1, ceT1, and T2 MRI modalities, FusionNet-A and FusionNet-C the best-performing model overall, with FusionNet-A particularly excelling in the enhancing tumor areas, while FusionNet-C demonstrates strong performance in the necrotic core and peritumoral edema regions. FusionNet-A excels in the enhancing tumor areas across all metrics (0.75 for recall, 0.83 for precision and 0.74 for dice scores) and also performs well in the peritumoral edema regions (0.77 for recall, 0.77 for precision and 0.75 for dice scores). Combinations including FLAIR and ceT1 tend to have better segmentation performance, especially for necrotic core regions. Using only FLAIR achieves a recall of 0.73 for peritumoral edema regions. Visualization results also indicate that our model generally achieves segmentation results similar to the ground truth.DiscussionFusionNet combines the benefits of U-Net and SegNet, outperforming the tumor segmentation performance of both. Although our model effectively segments brain tumors with competitive accuracy, we plan to extend the framework to achieve even better segmentation performance.

Prediction of lateral lymph node metastasis with short diameter less than 8mm in papillary thyroid carcinoma based on radiomics.

The aim of this study was to establish an ensemble learning model based on clinicopathological parameter and ultrasound radomics for assessing the risk of lateral cervical lymph node with short diameter less than 8mm (small lymph nodes were used instead) metastasis in patients with papillary thyroid cancer (PTC), thereby guiding the selection of surgical methods. This retrospective analysis was conducted on 454 patients diagnosed with papillary thyroid carcinoma who underwent total thyroidectomy and lateral neck lymph node dissection or lymph node intraoperative frozen section biopsy at the First Hospital of China Medical University between January 2015 and April 2022. In a ratio of 8:2, 362(80%) patients were assigned to the training set and 92(20%) patients were assigned to the test set. Clinical pathological features and radomics features related to ultrasound imaging were extracted, followed by feature selection using recursive feature elimination (RFE). Based on distinct feature sets, we constructed ensemble learning models comprising random forest (RF), extreme gradient boosting (XGBoost), categorical boosting (CatBoost), gradient boosting decision tree (GBDT), and light gradient boosting machine (Lightgbm) to develop clinical models, radiomics models, and clinical-radiomic models. Through the comparison of performance metrics such as area under curve (AUC), accuracy (ACC), specificity (SPE), precision (PRE), recall rate, F1 score, mean squared error (MSE) etc., we identified the optimal model and visualized its results using shapley additive exPlanations (SHAP). In this study, a total of 454 patients were included, among whom 342 PTC patients had small lymph node metastasis in the lateral neck region, while 112 did not have any metastasis. A total of 1035 features were initially considered for inclusion in this study, which were then narrowed down to 10 clinical features, 8 radiomics features, and 17 combined clinical-omics features. Based on these three feature sets, a total of fifteen ensemble learning models were established. In the test set, RF model in the clinical model is outperforms other models (AUC = 0.72, F1 = 0.75, Jaccard = 0.60 and Recall = 0.84), while CatBoost model in the radiomics model is superior to other models (AUC = 0.91, BA = 0.83 and SPE = 0.76). Among the clinical-radiomic models, Catboost exhibits optimal performance (AUC = 0.93, ACC = 0.88, BA = 0.87, F1 = 0.91, SPE = 0.83, PRE = 0.88, Jaccard = 0.83 and Recall = 0.92). Using the SHAP algorithm to visualize the operation process of the clinical-omics CatBoost model, we found that clinical omics features such as central lymph node metastasis (CLNM), Origin_Shape_Sphericity (o_shap_sphericity), LoG-sigma3_first order_ Skewness (log-3_fo_skewness), wavelet-HH_first order_Skewness (w-HH_fo_skewness) and wavelet-HH_first order_Skewness (sqr_gldm_DNUN) had the greatest impact on predicting the presence of lateral cervical small lymph node metastasis in PTC patients. (1) In this study, among the ensemble learning models established based on clinicopathological features and radiomics features for predicting PTC lateral small lymph node metastasis, the clinical-radiomic CatBoost model has the best performance. (2) SHAP can visualize how the clinical and radiomics features affect the results and realize the interpretation of the model. (3) The combined CatBoost model can improve the diagnostic accuracy of suspicious lymph nodes with short diameter < 8mm that are difficult to obtain accurate puncture results. The combined application of radiomics features is more accurate and reasonable than the prediction of clinical data alone, which helps to accurately evaluate the surgical scope and provide support for individual clinical decision making.

Prediction model and technical and tactical decision analysis of women's badminton singles based on machine learning.

In the Paris Olympic cycle, South Korean women's athlete An Se-young rose to the top of the 2023 BWF Olympic points with a win rate of 89.5%. With An Se-young as the subject, this paper aims to carry out technical and tactical analysis of women's badminton singles and formulate a prediction model based on machine learning. Firstly, An's technical and tactical statistics are analyzed and presented in a proposed "three-stage" data classification method. Secondly, we improve our "three-stage" machine learning dataset using video analysis of 10 matches (21 point games) where An Se-young faced off against four other players ranked in the top five of the World Badminton Federation (BWF) in week 44 of 2023. Finally, we establish a prediction model for the scoring and losing of points in the women's badminton singles based on the 'Decision tree', 'Random forest', 'XGBoost', 'Support vector' and 'K-proximity' algorithms, and analyze the effectiveness of this model. The results show that the improved data classification is reasonable and can be used to predict the final score of a match. When the support vector machine uses the RBF function kernel, the accuracy reaches its highest at 87.5%, and the consistency of this prediction model is strong. An's playstyle is sustained and unified; she does not seek continuous pressure, but rather exploits and maximizes her aggression following any mistake made by her opponents, immediately utilizing assault methods such as kills or dives, often resulting in the conversion of points during the subsequent 2-3 strikes.

Human identification via digital palatal scans: a machine learning validation pilot study.

This study aims to validate a machine learning algorithm previously developed in a training population on a different randomly chosen population (i.e., test set). The discrimination potential of the palatal intraoral scan-based geometric and superimposition methods was evaluated. A total of 23 participants (16 females and seven males) from different countries underwent palatal scans using the Emerald intraoral scanner. Geometric-based identification involved measuring the height, width, and depth of the palatal vault in each scan. These parameters were then input into Fisher's linear discriminant equations with coefficients determined previously on a training set. Sensitivity and specificity were calculated. For the superimposition method, scan repeatability was compared to between-subjects differences, calculating mean absolute differences (MAD) between aligned scans. Multiple linear regression analysis determined the effects of sex, longitude, and latitude of country of origin on concordance. The geometric-based method achieved 91.2% sensitivity and 97.1% specificity, consistent with the results from the training set, showing no significant difference. Latitude and longitude did not significantly affect geometric-based matches. In the superimposition method, the between-subjects MAD range (1.068-0.214mm) and the repeatability range (0.011-0.093mm) did not overlap. MAD was minimally affected by longitude and not influenced by latitude. The sex determination function recognized females over males with 69.0% sensitivity, similar to the training set. However, the specificity (62.5%) decreased. The assessment of geometric and superimposition discrimination has unequivocally demonstrated its robust reliability, remaining impervious to population. In contrast, the distinction between sexes carries only moderate reliability. The significant correlation observed among longitude, latitude, and palatal height suggests the feasibility of a comprehensive large-scale study to determine one's country of origin. Portable intraoral scanners can aid forensic investigations as adjunct identification methods by applying the proposed discriminant function to palatal geometry without population restrictions. The Clinicatrial.gov registration number is NCT05349942 (27/04/2022).

Improved prognostication of overall survival after radiotherapy in lung cancer patients by an interpretable machine learning model integrating lung and tumor radiomics and clinical parameters.

Accurate prognostication of overall survival (OS) for non-small cell lung cancer (NSCLC) patients receiving definitive radiotherapy (RT) is crucial for developing personalized treatment strategies. This study aims to construct an interpretable prognostic model that combines radiomic features extracted from normal lung and from primary tumor with clinical parameters. Our model aimed to clarify the complex, nonlinear interactions between these variables and enhance prognostic accuracy. We included 661 stage III NSCLC patients from three multi-national datasets: a training set (N = 349), test-set-1 (N = 229), and test-set-2 (N = 83), all undergoing definitive RT. A total of 104 distinct radiomic features were separately extracted from the regions of interest in the lung and the tumor. We developed four predictive models using eXtreme gradient boosting and selected the top 10 features based on the Shapley additive explanations (SHAP) values. These models were the tumor radiomic model (Model-T), lung radiomic model (Model-L), a combined radiomic model (Model-LT), and an integrated model incorporating clinical parameters (Model-LTC). Model performance was evaluated through Harrell's concordance index, Kaplan-Meier survival curves, time-dependent area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis. Interpretability was assessed using the SHAP framework. Model-LTC exhibited superior performance, with notable predictive accuracy (C-index: training set, 0.87; test-set-2, 0.76) and time-dependent AUC above 0.75. Complex nonlinear relationships and interactions were evident among the model's variables. The integration of radiomic and clinical factors within an interpretable framework significantly improved OS prediction. The SHAP analysis provided insightful interpretability, enhancing the model's clinical applicability and potential for aiding personalized treatment decisions.