10-fold Cross-validation Research Articles

Chronic low-grade inflammation in patients with systemic sclerosis is associated with increased risk for arteriosclerotic cardiovascular disease

BackgroundVasculopathy is a hallmark of systemic sclerosis (SSc) putting patients at an increased risk of cardiovascular disease. Approximately 20–25% of all SSc patients show prolonged elevated C-reactive protein (CRP) levels and thus signs of chronic low-grade inflammation. While CRP−positivity is an independent predictor of cardiovascular disease in non-SSc populations, the relationship between CRP−positivity and cardiovascular health/atherosclerosis in SSc patients is only incompletely understood. Here, we aimed to assess (1) which general, SSc disease-specific and cardiovascular parameters are associated with CRP−positivity in a cohort of SSc patients with prolonged CRP elevations (CRP+ SSc group) relative to SSc patients without CRP elevations (CRP− SSc group). In addition (2), we aimed to investigate whether prolonged CRP−positivity in SSc patients is associated with a higher cardiovascular risk and an increased atherosclerotic burden. We also aimed to (3) identify via random forest classification modeling which combined cardiovascular and/or SSc-specific parameters could differentiate best between SSc patients with elevated CRP levels (the so-called “inflammatory SSc subtype”) and SSc patients without increased CRP levels.MethodsSixty-five SSc patients were recruited and assigned to the CRP+ SSc group (n = 20) if their CRP levels were > 5 mg/L in at least three half-yearly visits within 2 years before enrolment or to the CRP− SSc group (n = 45), respectively. All patients underwent an anamnesis, physical examination, blood draw, and bilateral carotid ultrasound in order to assess arteriosclerotic burden including the presence, number and height of plaques, and carotid intima–media thickness (CIMT) as well as lipid profiles. 10-year ASCVD risk was estimated via the ASCVD risk estimator plus. Statistical evaluation included Spearman’s correlations, logistic regression and random forest modeling under 5-fold cross-validation, and permutation testing to determine combinations of cardiovascular variables highly discriminatory for CRP−positivity.ResultsSSc groups showed comparable mean age, height, and extent of SSc organ involvement. Regarding cardiovascular health, CRP+ SSc patients exhibited a significantly altered HDL-, LDL-, and triglyceride profile (0.001 ≤ p ≤ 0.017) and a significantly higher 10-year ASCVD risk (p = 0.047), relative to CRP− SSc patients. Additionally, within the subgroup of CRP+ SSc patients, positive correlations between CRP levels and CIMT right (ρ = 0.657, p = 0.002) and mean CIMT left and right (ρ = 0.497, p = 0.026) were seen. Combined ROC models identified the four lipid components (HDL, LDL, total cholesterol, and triglycerides) or the SSc duration and ASCVD category to differentiate with high cross-validated ROC-AUCs (AUC: 0.83 ± 0.15, and AUC: 0.86 ± 0.09, p < 0.001) for prolonged CRP−positivity among SSc patients.ConclusionOur data indicate that persistent CRP−positivity and thus chronic low-grade inflammation in SSc patients enhance the risk for arteriosclerotic-cardiovascular disease significantly beyond the ASCVD risk observed for our SSc patients without CRP elevations. It seems to be along with a disrupted lipid profile the hallmark of a distinct “inflammatory” subgroup of SSc patients. However, large population-based studies and clinical trials in patients with SSc are needed to validate our findings in a prospective or interventional setting.

An automated pheochromocytoma and paraganglioma lesion segmentation AI-model at whole-body 68Ga- DOTATATE PET/CT.

Somatostatin receptor (SSR) targeting radiotracer 68Ga-DOTATATE is used for Positron Emission Tomography (PET)/Computed Tomography (CT) imaging to assess patients with Pheochromocytoma and paraganglioma (PPGL), rare types of Neuroendocrine tumor (NET) which can metastasize thereby becoming difficult to quantify. The goal of this study is to develop an artificial intelligence (AI) model for automated lesion segmentation on whole-body 3D DOTATATE-PET/CT and to automate the tumor burden calculation. 132 68Ga-DOTATATE PET/CT scans from 38 patients with metastatic and inoperable PPGL, were split into 70, and 62 scans, from 20, and 18 patients for training, and test sets, respectively. The training set was further divided into patient-stratified 5 folds for cross-validation. 3D-full resolution nnUNet configuration was trained with 5-fold cross-validation. The model's detection performance was evaluated at both scan and lesion levels for the PPGL test set and two other clinical cohorts with NET (n = 9) and olfactory neuroblastoma (ONB, n = 5). Additionally, quantitative statistical analysis of PET parameters including SUVmax, total lesion uptake (TLU), and total tumor volume (TTV), was conducted. The nnUNet AI model achieved an average 5-fold validation dice similarity coefficient of 0.84 at the scan level. The model achieved dice similarity coefficients (DSC) of 0.88, 0.6, and 0.67 at the scan level, the sensitivity of 86%, 61.13%, and 61.64%, and a positive predictive value of 89%, 74%, and 86.54% at the lesion level for the PPGL test, NET and ONB cohorts, respectively. For PPGL cohorts, smaller lesions with low uptake were missed by the AI model (p < 0.001). Anatomical region-based failure analysis showed most of the false negative and false positive lesions within the liver for all the cohorts, mainly due to the high physiologic liver background activity and image noise on 68Ga- DOTATATE PET scans. The developed deep learning-based AI model showed reliable performance for automated segmentation of metastatic PPGL lesions on whole-body 68Ga-DOTATATE-PET/CT images, which may be beneficial for tumor burden estimation for objective evaluation during therapy follow-up. https://www. gov/study/NCT03206060 , https://www. gov/study/NCT04086485 , https://www. gov/study/NCT05012098 .

A Deep Learning Model to Predict Breast Implant Texture Types Using Ultrasonography Images: Feasibility Development Study.

Breast implants, including textured variants, have been widely used in aesthetic and reconstructive mammoplasty. However, the textured type, which is one of the shell texture types of breast implants, has been identified as a possible etiologic factor for lymphoma, specifically breast implant-associated anaplastic large cell lymphoma (BIA-ALCL). Identifying the shell texture type of the implant is critical to diagnosing BIA-ALCL. However, distinguishing the shell texture type can be difficult due to the loss of human memory and medical history. An alternative approach is to use ultrasonography, but this method also has limitations in quantitative assessment. This study aims to determine the feasibility of using a deep learning model to classify the shell texture type of breast implants and make robust predictions from ultrasonography images from heterogeneous sources. A total of 19,502 breast implant images were retrospectively collected from heterogeneous sources, including images captured from both Canon and GE devices, images of ruptured implants, and images without implants, as well as publicly available images. The Canon images were trained using ResNet-50. The model's performance on the Canon dataset was evaluated using stratified 5-fold cross-validation. Additionally, external validation was conducted using the GE and publicly available datasets. The area under the receiver operating characteristic curve (AUROC) and the area under the precision-recall curve (PRAUC) were calculated based on the contribution of the pixels with Gradient-weighted Class Activation Mapping (Grad-CAM). To identify the significant pixels for classification, we masked the pixels that contributed less than 10%, up to a maximum of 100%. To assess the model's robustness to uncertainty, Shannon entropy was calculated for 4 image groups: Canon, GE, ruptured implants, and without implants. The deep learning model achieved an average AUROC of 0.98 and a PRAUC of 0.88 in the Canon dataset. The model achieved an AUROC of 0.985 and a PRAUC of 0.748 for images captured with GE devices. Additionally, the model predicted an AUROC of 0.909 and a PRAUC of 0.958 for the publicly available dataset. This model maintained the PRAUC values for quantitative validation when masking up to 90% of the least-contributing pixels and the remnant pixels in breast shell layers. Furthermore, the prediction uncertainty increased in the following order: Canon (0.066), GE (0072), ruptured implants (0.371), and no implants (0.777). We have demonstrated the feasibility of using deep learning to predict the shell texture type of breast implants. This approach quantifies the shell texture types of breast implants, supporting the first step in the diagnosis of BIA-ALCL.

Common laboratory results-based artificial intelligence analysis achieves accurate classification of plasma cell dyscrasias

Background Plasma cell dyscrasias encompass a diverse set of disorders, where early and precise diagnosis is essential for optimizing patient outcomes. Despite advancements, current diagnostic methodologies remain underutilized in applying artificial intelligence (AI) to routine laboratory data. This study seeks to construct an AI-driven model leveraging standard laboratory parameters to enhance diagnostic accuracy and classification efficiency in plasma cell dyscrasias. Methods Data from 1,188 participants (609 with plasma cell dyscrasias and 579 controls) collected between 2018 and 2023 were analyzed. Initial variable selection employed Kruskal-Wallis and Wilcoxon tests, followed by dimensionality reduction and variable prioritization using the Shapley Additive Explanations (SHAP) approach. Nine pivotal variables were identified, including hemoglobin (HGB), serum creatinine, and β2-microglobulin. Utilizing these, four machine learning models (gradient boosting decision tree (GBDT), support vector machine (SVM), deep neural network (DNN), and decision tree (DT) were developed and evaluated, with performance metrics such as accuracy, recall, and area under the curve (AUC) assessed through 5-fold cross-validation. A subtype classification model was also developed, analyzing data from 380 cases to classify disorders such as multiple myeloma (MM) and monoclonal gammopathy of undetermined significance (MGUS). Results 1. Variable selection: The SHAP method pinpointed nine critical variables, including hemoglobin (HGB), serum creatinine, erythrocyte sedimentation rate (ESR), and β2-microglobulin. 2. Diagnostic model performance: The GBDT model exhibited superior diagnostic performance for plasma cell dyscrasias, achieving 93.5% accuracy, 98.1% recall, and an AUC of 0.987. External validation reinforced its robustness, with 100% accuracy and an F1 score of 98.5%. 3. Subtype Classification: The DNN model excelled in classifying multiple myeloma, MGUS, and light-chain myeloma, demonstrating sensitivity and specificity above 90% across all subtypes. Conclusions AI models based on routine laboratory results significantly enhance the precision of diagnosing and classifying plasma cell dyscrasias, presenting a promising avenue for early detection and individualized treatment strategies.

Prediction and assessment of meteorological drought in southwest China using long short-term memory model

Abstract Drought prediction is crucial for mitigating risks and designing measures to alleviate its impact. Machine learning models have been widely applied in the field of drought prediction in recent years. This study concentrated on predicting meteorological droughts in southwest China, a region prone to frequent and severe droughts, particularly in areas with sparse meteorological station coverage. The long short-term memory (LSTM) predictive model, which is a deep learning model, was constructed by calculating standardized precipitation evapotranspiration index (SPEI) values based on 144 weather station observations from 1980 to 2020. The 5-fold cross-validation method was used for the hyperparameter optimization of the model. The LSTM model underwent comprehensive assessment and validation through multiple methods. This included the use of several accuracy assessment indicators and a comparison of results. The comparison covered different drought characteristics among the LSTM predictive model, the benchmark random forest (RF) predictive model, the historical drought situations, and the calculated SPEI values based on observations from 144 weather stations. The results showed that the training results of the LSTM predictive model basically agreed with the SPEI values calculated from weather station observations. The model-predicted variation trend of SPEI values for 2020 was similar to the variation in SPEI values calculated based on weather station observations. On the test set, the coefficient of determination (R 2), the root mean square error, the explained variance score, the Nash–Sutcliffe efficiency, and the Kling–Gupta efficiency were 0.757, 0.210, 0.802, 0.761, and 0.212, respectively. The total consistency rate of the drought grade was 59.26%. The spatial correlation distribution of SPEI values between LSTM model prediction and calculation from meteorological stations in 2020 was more than 0.5 for most regions. The correlation coefficients exceeded 0.6 in western Tibet and Chengdu Plains. Compared to the RF model, the LSTM model excelled in all five performance evaluation metrics and demonstrated a higher overall consistency rate for drought categories. The Kruskal–Wallis test for both the LSTM and RF models all indicated no significant difference in the distributions between the predicted and observed data. Scatter plots revealed that the prediction accuracy for both models in 2020 was suboptimal, with the SPEI showing a comparatively narrow range of values. Nonetheless, the LSTM model significantly outperformed the RF model in terms of prediction accuracy. In summary, the LSTM model demonstrated good overall performance, accuracy, and applicability. It has the potential to enhance dynamic drought prediction in regions with complex terrain, diverse climatic factors, and sparse weather station networks.

Predictors and Profile of Severe Infectious Complications in Multiple Myeloma Patients Treated with Daratumumab-Based Regimens: A Machine Learning Model for Pneumonia Risk

Background: Daratumumab (Dara) is the first monoclonal antibody introduced into clinical practice to treat multiple myeloma (MM). It currently forms the backbone of therapy regimens in both newly diagnosed (ND) and relapsed/refractory (RR) patients. However, previous reports indicated an increased risk of infectious complications (ICs) during Dara-based treatment. In this study, we aimed to determine the profile of ICs in MM patients treated with Dara-based regimens and establish predictors of their occurrence. Methods: This retrospective, real-life study included MM patients treated with Dara-based regimens between July 2019 and March 2024 at our institution. Infectious events were evaluated using the Terminology Criteria for Adverse Events (CTCAE) version 5.0. Results: The study group consisted of a total of 139 patients, including 49 NDMM and 90 RRMM. In the RR setting, the majority (60.0%) of patients received the Dara, bortezomib, and dexamethasone (DVd) regimen, whereas ND patients were predominantly (98%) treated with the Dara, bortezomib, thalidomide, and dexamethasone (DVTd) regimen. Overall, 55 patients (39.6%) experienced ICs. The most common IC was pneumonia (37.5%), followed by upper respiratory tract infections (26.8%). Finally, twenty-five patients had severe ICs (grade ≥ 3) and required hospitalization, and eight patients died due to ICs. In the final multivariable model adjusted for setting (ND/RR) and age, hemoglobin level (OR 0.77, 95% CI: 0.61–0.96, p = 0.0037), and Eastern Cooperative Oncology Group (ECOG) &gt;1 (OR 4.46, 95% CI: 1.63–12.26, p = 0.0037) were significant factors influencing severe IC occurrence. Additionally, we developed predictive models using the J48 decision tree, gradient boosting, and random forest algorithms. After conducting 10-fold cross-validation, these models demonstrated strong performance in predicting the occurrence of pneumonia during treatment with daratumumab-based regimens. Conclusions: Simple clinical and laboratory assessments, including hemoglobin level and ECOG scale, can be valuable in identifying patients vulnerable to infections during Dara-based regimens, facilitating personalized prophylactic strategies.

Machine learning prediction of mechanical properties of bamboo by hemicelluloses removal

The use of bamboo as a sustainable material is becoming increasingly prevalent; however, the optimisation of its mechanical properties remains a significant challenge. In this study, different concentrations of sodium hydroxide (NaOH) were used to remove the hemicellulose of bamboo and the effect of mechanical properties such as modulus of elasticity, flexural strength and elastic limit were evaluated. A total of 90 samples of data were collected and five machine learning algorithms including Principal Component Regression (PCR), Partial Least Squares Regression (PLSR), Ridge Regression (RR), Lasso Regression (LR) and Elastic Network Regression (ENR) were employed to build predictive models based on these properties. The models were trained and tested using 5-fold cross validation. The results showed that the mechanical properties of elastic modulus and elastic limit of bamboo medium were enhanced to 9.63 GPa and 66.44 MPa treated by 10 % NaOH, while the flexural strength of bamboo medium was up to 147.29 GPa with 5 % NaOH. The Ridge Regression algorithm was the best compared to other four algorithms. This methodological approach is not only offers an efficient way to optimize the mechanical properties of bamboo, but also enhances sustainable practices in the environmental sector.

Machine Learning Models for Predicting Dysphonia Following Anterior Cervical Discectomy and Fusion – A Swedish Registry Study

BACKGROUNDDysphonia is one of the more common complications following anterior cervical discectomy and fusion (ACDF). ACDF is the gold standard for treating degenerative cervical spine disorders, and identifying high-risk patients is therefore crucial. PURPOSEThis study aimed to evaluate different machine learning models to predict persistent dysphonia after ACDF. STUDY DESIGNA retrospective review of the nationwide Swedish spine registry (Swespine) PATIENT SAMPLEAll adults in the Swespine registry who underwent elective ACDF between 2006 and 2020. OUTCOME MEASURESThe primary outcome was self-reported dysphonia lasting at least one month after surgery. Predictive performance was assessed using discrimination and calibration metrics. METHODS: Patients with missing dysphonia data at the one-year follow-up were excluded. Data preprocessing involved one-hot encoding categorical variables, scaling continuous variables, and imputing missing values. Four machine learning models (logistic regression, random forest (RF), gradient boosting, K-nearest neighbor) were employed. The models were trained and tested using an 80:20 data split and 5-fold cross-validation, with performance metrics guiding the selection of the best model for predicting persistent dysphonia. RESULTSIn total, 2,708 were included in the study. Twelve key predictors were identified. Four machine learning models were tested, with the RF model achieving the best performance (AUC = 0.794). The most significant predictors across models included preoperative NDI, EQ5Dindex, preoperative neurology, number of operated levels, and use of a fusion cage. The RF model, chosen for its superior performance, showed high sensitivity and consistent accuracy, but a low specificity and positive predictive value. CONCLUSIONSIn this study, machine learning models were employed to identify predictors of persistent dysphonia following ACDF. Among the models tested, the RF classifier demonstrated superior performance, with an AUC value of 0.790. The RF model identified NDI, EQ5Dindex, and number of fused vertebrae as key variables. These findings underscore the potential of machine learning models in identifying patients at increased risk for dysphonia persisting for more than one month after surgery.

Split tensile strength of fiber-reinforced coral aggregate concrete: Deep learning model and experimental validation

Machine learning methods are increasingly becoming an important research field for solving the complex mechanical behavior of concrete materials. A reliable deep learning model was developed based on a dataset consisting of 319 experimental data points to predict the split tensile strength of fiber-reinforced coral aggregate concrete (FRCAC). The feature selection was carried out using correlation analysis, and the neural network structure was optimized by adjusting hyperparameters based on the 10-fold cross-validation (CV). Subsequently, six statistical indicators were employed to evaluate the accuracy and generalization ability of the model. The deep neural network (DNN) model demonstrates excellent performance, with R2, MAE, and MAPE of 0.98, 0.181 MPa, and 0.037 for the training dataset and 0.95, 0.286 MPa, and 0.074 for the test dataset, respectively. Additionally, 24 sets of concrete proportions are designed using the trained DNN model, and the results indicate that the absolute error and relative error between predicted and experimental values are less than 0.4 MPa and 6.8 %, respectively. The feature analysis based on SHapley Additive exPlanations (SHAP) reveals that volume fraction, curing age, water-binder ratio, and the tensile strength of fiber have the highest contribution in estimating the split tensile strength of FRCAC. This research can provide a reference for optimizing the design of FRCAC.

Impact of Data Pre-Processing Techniques on XGBoost Model Performance for Predicting All-Cause Readmission and Mortality Among Patients with Heart Failure

Background: Heart failure poses a significant global health challenge, with high rates of readmission and mortality. Accurate models to predict these outcomes are essential for effective patient management. This study investigates the impact of data pre-processing techniques on XGBoost model performance in predicting all-cause readmission and mortality among heart failure patients. Methods: A dataset of 168 features from 2008 heart failure patients was used. Pre-processing included handling missing values, categorical encoding, and standardization. Four imputation techniques were compared: Mean, Multivariate Imputation by Chained Equations (MICEs), k-nearest Neighbors (kNNs), and Random Forest (RF). XGBoost models were evaluated using accuracy, recall, F1-score, and Area Under the Curve (AUC). Robustness was assessed through 10-fold cross-validation. Results: The XGBoost model with kNN imputation, one-hot encoding, and standardization outperformed others, with an accuracy of 0.614, recall of 0.551, and F1-score of 0.476. The MICE-based model achieved the highest AUC (0.647) and mean AUC (0.65 ± 0.04) in cross-validation. All pre-processed models outperformed the default XGBoost model (AUC: 0.60). Conclusions: Data pre-processing, especially MICE with one-hot encoding and standardization, improves XGBoost performance in heart failure prediction. However, moderate AUC scores suggest further steps are needed to enhance predictive accuracy.

Optimizing classification models for medical image diagnosis: a comparative analysis on multi-class datasets

The surge in machine learning (ML) and artificial intelligence has revolutionized medical diagnosis, utilizing data from chest ct-scans, COVID-19, lung cancer, brain tumor, and alzheimer parkinson diseases. However, the intricate nature of medical data necessitates robust classification models. This study compares support vector machine (SVM), naïve Bayes, k-nearest neighbors (K-NN), artificial neural networks (ANN), and stochastic gradient descent on multi-class medical datasets, employing data collection, Canny image segmentation, hu moment feature extraction, and oversampling/under-sampling for data balancing. Classification algorithms are assessed via 5-fold cross-validation for accuracy, precision, recall, and F-measure. Results indicate variable model performance depending on datasets and sampling strategies. SVM, K-NN, ANN, and SGD demonstrate superior performance on specific datasets, achieving accuracies between 0.49 to 0.57. Conversely, naïve Bayes exhibits limitations, achieving precision levels of 0.46 to 0.47 on certain datasets. The efficacy of oversampling and under-sampling techniques in improving classification accuracy varies inconsistently. These findings aid medical practitioners and researchers in selecting suitable models for diagnostic applications.

Antibiotic profile classification of Proteus mirabilis using Machine Learning: An Investigation into Multidimensional Radiomics Features

Antimicrobial resistance (AMR) presents a significant threat to global healthcare. Proteus mirabilis causes catheter-associated urinary tract infections (CAUTIs) and exhibits increased antibiotic resistance. Traditional diagnostics still rely on culture-based approaches, which remain time-consuming. Here, we study the use of machine learning (ML) to classify bacterial resistance profiles using straightforward microscopic imaging of P. mirabilis for resistance classification integrated with radiomics feature analysis and ML models.From 150 P. mirabilis strains isolated from catheters of patients diagnosed with a CAUTI, 30 % displayed multidrug resistance using the standardized disk diffusion method, and 60% showed strong biofilm activity in microtiter plate assays. As a more rapid alternative, we introduce wavelet-based and regular microscopy imaging with feature extraction/selection, following image preprocessing steps (image denoising, normalization, and mask creation). These features enable training and testing different ML models with 5-fold cross-validation for P. mirabilis resistance classification. From these models, the Random Forest (RF) algorithm exhibited the highest performance with ACC=0.95, specificity=0.97, sensitivity=0.88, and AUC=0.98 among the other ML algorithms considered in this study for P. mirabilis resistance classification.This successful application of wavelet-based feature Radiomics analysis with RF model represents a crucial step towards a precise, rapid, and cost-effective method to distinguish antibiotic resistant P. mirabilis strains.

Categorical classification of skin cancer using a weighted ensemble of transfer learning with test time augmentation

Skin cancer is the abnormal development of cells on the surface of the skin and is one of the most fatal diseases in humans. It usually appears in locations that are exposed to the sun, but can also appear in areas that are not regularly exposed to the sun. Due to the striking similarities between benign and malignant lesions, skin cancer detection remains a problem, even for expert dermatologists. Considering the inability of dermatologists to diagnose skin cancer accurately, a convolutional neural network (CNN) approach was used for skin cancer diagnosis. However, the CNN model requires a significant number of image datasets for better performance; thus, image augmentation and transfer learning techniques have been used in this study to boost the number of images and the performance of the model, because there are a limited number of medical images. This study proposes an ensemble transfer-learning-based model that can efficiently classify skin lesions into one of seven categories to aid dermatologists in skin cancer detection: (i) actinic keratoses, (ii) basal cell carcinoma, (iii) benign keratosis, (iv) dermatofibroma, (v) melanocytic nevi, (vi) melanoma, and (vii) vascular skin lesions. Five transfer learning models were used as the basis of the ensemble: MobileNet, EfficientNetV2B2, Xception, ResNext101, and DenseNet201. In addition to the stratified 10-fold cross-validation, the results of each individual model were fused to achieve greater classification accuracy. An annealing learning rate scheduler and test-time augmentation (TTA) were also used to increase the performance of the model during the training and testing stages. A total of 10,015 publicly available dermoscopy images from the HAM10000 (Human Against Machine) dataset, which contained samples from the seven common skin lesion categories, were used to train and evaluate the models. The proposed technique attained 94.49% accuracy on the dataset. These results suggest that this strategy can be useful for improving the accuracy of skin cancer classification. However, the weighted average of f1-score, recall, and precision were obtained to be 94.68%, 94.49%, and 95.07%, respectively.

Application of K-Nearest Neighbor Algorithm for Consumer Behaviour Identification and Product Personalisation Based on Big Data Analysis

The rapid digital transformation has fundamentally altered the retail industry, presenting challenges such as shifting consumer behavior and intensified market competition. This research explores the application of the K-Nearest Neighbor (KNN) algorithm for identifying consumer behavior and developing product personalization systems based on big data insights. Utilizing a dataset comprising 10,000 transaction records from January to December 2023 and 5,302 product types, we implemented the KNN algorithm to predict consumer purchases. The data was processed into 89,908 distinct transaction records. Our evaluation, using 5-fold cross-validation, demonstrated that the optimal performance of the KNN model was achieved with 𝑘=10, yielding a precision of 0.8319, recall of 0.8311, and an F1-score of 0.8312. These findings highlight the effectiveness of KNN in enhancing consumer satisfaction through precise product recommendations. This study provides strategic insights for modern retailers aiming to leverage AI and big data to remain competitive and meet evolving consumer expectations.

An Artificial Intelligence-Based Hybrid Approach to Detect the Type of Buried Objects with Broad Frequency Band Antenna Systems

Knowing the type of buried object before excavation prevents unnecessary excavation. Moreover, it saves time and money. In this study, an experiment set was prepared for the detection of buried objects. The experimental set was composed of an antenna that sends and receives electromagnetic waves in a wide frequency band, software that records and processes reflections, and a sandbox. In the study, metallic and non-metallic objects with different depths, sizes and shapes were buried in this sand pool and measurements were taken along a profile. 2D images were created from the measurements and image processing techniques were applied to these images. Classification algorithms were used to detect the type of bruied object from processed images. To increase the success of the algorithms, correlation-based attribute selection (CFS) and Principal Component Analysis (PCA) were used as attribute selection techniques. Genetic algorithm (GA), Particle Swarm Optimization (PSO), Harmony search (HA), and Evolutionary search (EA), which are among the metaheuristic optimization algorithms, were preferred as search methods in attribute selection with CFS. The performance of the algorithms was analyzed using the 10-fold cross-validation method. As a result, it was understood that the use of the PCA algorithm in attribute selection increases the classification success more than metaheuristic algorithms. The most successful among the classification algorithms used is the Random tree algorithm. After PCA, the accuracy value of this algorithm was 95.8 Therefore, a hybrid approach is proposed in which PCA and Random tree algorithms are used in the software embedded in the measurement system.

How do we safely preserve ovaries in patients with cervical adenocarcinoma: risk factors and predictive models

ObjectiveTo study and predict the risk of ovarian metastasis (OM) in patients with cervical adenocarcinoma (ADC).MethodsPatients with ADC who received surgical treatment from January 2015 to December 2022 in the Obstetrics and Gynecology Hospital of Fudan University were included in the study. Patients were further divided into OP (ovaries were preserved in surgery) and BSO (bilateral salpingo-oophorectomy) groups. For the patients who accepted BSO, 60% of the patients were randomly grouped into a training cohort, and predictive prognostic models were constructed with 10-fold cross-validation through random forest, LASSO, stepwise, and optimum subset analysis. The model with the highest area under receiver operator curve (AUC) was screened out in the testing cohort. The nomogram and its calibration curve presented the possibility of OM in future patients. The prognoses between the different populations were compared using Kaplan–Meier analysis. All data processing was carried out in R 4.2.0 software.ResultsA total of 934 patients were enrolled in our cohort; 266 patients had their ovaries preserved and 668 patients had BSO according to the previous criteria reported The clinical safety with these criteria was secured, while the 5-year overall survival had no significant difference between the BSO and OP groups (p = 0.069), which suggested that the current criteria could be extended and are more precise. Four predictive models for ovarian metastasis by machine learning were constructed in our study, and the random forest model that obtained the highest AUC in both training and testing sets (0.971 for training and 0.962 for testing set) was taken as the best model. The optimal cut-off point of the ROC curve (specificity 99.5% and 90% sensitivity) was utilized to stratify the patients into high- and low-risk OM. Further comparing the survival curves of the high and low-OM risk groups, it was found that both DFS and OS were significantly prolonged in the low-risk group (p &amp;lt; 0.01). On the basis of this random forest model, a nomogram was used to calculate the OM risk, and the results were validated with calibration. The predictive model was further applied to the whole cohort (934 patients), and we identified the OM low-risk population (n = 822) and the patients with high risk who should be recommended for BSO (n = 112). No significant difference was found in the 5-year survival between the low-risk group with our model and the patients who already had ovaries preserved according to the previous criteria (n = 266).ConclusionThe predictive model constructed in our study could identify the low-risk population of OM in patients with ADC, which remarkably extended the number with the previous criteria, for whom we could potentially preserve ovaries to help improve their life quality.

Graph-Root: Prediction of Root-Associated Proteins in Maize, Sorghum, And Soybean Based on Graph Convolutional Network and Network Embedding Method

Background: The root system plays an irreplaceable role in plant growth. Its improvement can increase crop productivity. However, such a system is still mysterious for us. The underlying mechanism has not been fully uncovered. The investigation on proteins related to the root system is an important means to complete this task. In the previous time, lack of root-related proteins makes it impossible to adopt machine learning methods for designing efficient models for the discovery of novel root-related proteins. Recently, a public database on root-related proteins was set up and machine learning methods can be applied in this field. Objective: The purpose of this study was to design an efficient computational method to predict root-associated proteins in three plants: maize, sorghum, and soybean. Method: In this study, we proposed a machine learning based model, named Graph-Root, for the identification of root-related proteins in maize, sorghum, and soybean. The features derived from protein sequences, functional domains, and one network were extracted, where the first type of features were processed by graph convolutional neural network and multi-head attention, the second type of features reflected the essential functions of proteins, and the third type of features abstracted the linkage between proteins. These features were fed into the fully connected layer to make predictions. Results: The 5-fold cross-validation and independent tests suggested its acceptable performance. It also outperformed the only previous model, SVM-Root. Furthermore, the importance of each feature type and component in the proposed model was investigated. Conclusion: Graph-Root had a good performance and can be a useful tool to identify novel rootrelated proteins. BLOSUM62 features were found to be important in determining root-related proteins.

A novel bi-level optimal strategy-assisted design boosting discovery of specific surface area of perovskite oxide for effective photocatalysis

Due to the inherent property of materials, photocatalysis plays a crucial role in diverse applications, such as the degradation of organic pollutants, fuel production, and nitrogen fixation. Perovskites, in particular, are remarkable for their high light absorption rate and energy conversion efficiency within the visible light range, thereby positioning them as promising photocatalysts. Given that specific surface area (SSA) is a critical metric for evaluating photocatalytic performance, accurately and quickly predicting the SSA of perovskites has emerged as a key area of technical research. Therefore, it is essential to explore the relationship between potential materials descriptors and SSA, which can significantly expedite the discovery of photocatalyzed perovskite materials with high SSA. In this study, we proposed a bi-level optimization method that combines descriptors identified strategy and model optimization strategy for predicting the SSA of unknown perovskite candidates, thereby elucidating the relationship between material features and SSA. Initially, at the upper level, the Gradient Boosting Regression combined with Recursive Feature Elimination (RFE-GBR) method was employed to identify crucial features that are closely correlated with the SSA of photocatalyzed perovskites. Subsequently, the Sure Independence Screening.Sparsifying Operator (SISSO) was utilized to develop new descriptors that exhibited strong correlations with SSA. Based on refined descriptors identifiied at the upper level, at the lower level, the optimal model strategy was obtained by Improved Osprey Optimization Algorithm (IOOA) using golden sine strategy and Gradient Boosting Regression (GBR) regression to tune the key parameters of GBR for optimizing SSA. The optimal designing method achieved remarkable predictive performance with R2 (R-squared) score of 0.90 and MAE (Mean Absolute Error) of 3.97 m2/g, along with 5-fold cross-validation. Then this method was applied to predict the SSA of 300 unidentified ABO3 perovskites. Comparison with experimental data revealed that our approach significantly accelerated the development of perovskite materials with superior photocatalytic properties, showing the potential of our model in advancing the field of photocatalysis.

Tracer-Separator: A Deep Learning Model for Brain PET Dual-Tracer (18F-FDG and Amyloid) Separation.

Multiplexed PET imaging revolutionized clinical decision-making by simultaneously capturing various radiotracer data in a single scan, enhancing diagnostic accuracy and patient comfort. Through a transformer-based deep learning, this study underscores the potential of advanced imaging techniques to streamline diagnosis and improve patient outcomes. The research cohort consisted of 120 patients spanning from cognitively unimpaired individuals to those with mild cognitive impairment, dementia, and other mental disorders. Patients underwent various imaging assessments, including 3D T1-weighted MRI, amyloid PET scans using either 18F-florbetapir (FBP) or 18F-flutemetamol (FMM), and 18F-FDG PET. Summed images of FMM/FBP and FDG were used as proxy for simultaneous scanning of 2 different tracers. A SwinUNETR model, a convolution-free transformer architecture, was trained for image translation. The model was trained using mean square error loss function and 5-fold cross-validation. Visual evaluation involved assessing image similarity and amyloid status, comparing synthesized images with actual ones. Statistical analysis was conducted to determine the significance of differences. Visual inspection of synthesized images revealed remarkable similarity to reference images across various clinical statuses. The mean centiloid bias for dementia, mild cognitive impairment, and healthy control subjects and for FBP tracers is 15.70 ± 29.78, 0.35 ± 33.68, and 6.52 ± 25.19, respectively, whereas for FMM, it is -6.85 ± 25.02, 4.23 ± 23.78, and 5.71 ± 21.72, respectively. Clinical evaluation by 2 readers further confirmed the model's efficiency, with 97 FBP/FMM and 63 FDG synthesized images (from 120 subjects) found similar to ground truth diagnoses (rank 3), whereas 3 FBP/FMM and 15 FDG synthesized images were considered nonsimilar (rank 1). Promising sensitivity, specificity, and accuracy were achieved in amyloid status assessment based on synthesized images, with an average sensitivity of 95 ± 2.5, specificity of 72.5 ± 12.5, and accuracy of 87.5 ± 2.5. Error distribution analyses provided valuable insights into error levels across brain regions, with most falling between -0.1 and +0.2 SUV ratio. Correlation analyses demonstrated strong associations between actual and synthesized images, particularly for FMM images (FBP: Y = 0.72X + 20.95, R2 = 0.54; FMM: Y = 0.65X + 22.77, R2 = 0.77). This study demonstrated the potential of a novel convolution-free transformer architecture, SwinUNETR, for synthesizing realistic FDG and FBP/FMM images from summation scans mimicking simultaneous dual-tracer imaging.

From Data to Diagnosis: Effective Machine Learning-based Heart Disease Prediction

Heart disease is a leading cause of mortality worldwide, contributing to nearly 18 million deaths annually. Early detection is critical but remains a significant challenge due to the limitations of traditional diagnostic methods, which can be prone to human error. This study aims to enhance heart disease prediction using machine learning (ML) by comparing the performance of three ML models: Support Vector Machine (SVM), Random Forest (RF), and XGBoost. A dataset containing 12 features from 918 patients was used, with preprocessing steps such as one-hot encoding for categorical variables and MinMax scaling for numerical features. The models were trained and evaluated using 5-fold cross-validation to ensure robustness. Random Forest demonstrated the highest accuracy at 82.78%, followed closely by SVM (82.67%) and XGBoost (81.58%). The feature importance analysis identified ST_Slope as the most significant predictor of heart disease, providing important insights into which features are most influential in the diagnosis process. While the Random Forest model outperformed the others, this study also highlights the need for better interpretability in ML models, especially in medical applications where understanding the relationships between features is crucial. Future research should focus on improving model transparency to bridge the gap between accuracy and practical application in clinical settings.