K-fold Cross-validation Method Research Articles

Machining Quality Prediction of Marine Diesel Engine Block Based on Error Transmission Network

In view of the high precision requirement of the marine diesel engine body and the difficulty of quality control, a quality prediction method of the body, based on a process error transfer network, is proposed. First, according to the processing information of the body, the network nodes and edges are abstracted to establish the process error transfer network of the body. Then, the key quality control points and key quality features of the diesel engine body are determined by the PageRank and node degree. The key quality features obtained from the network analysis are taken as the output, and the corresponding process errors and process parameters are taken as the input. Finally, the quality prediction model of the body is established based on SVR algorithm, and the C, g parameters of SVR algorithm are optimized by the K-fold cross-validation method and grid search method to improve the prediction accuracy of the body processing quality.

Sequential Backward Feature Selection Method Based on Stacking Framework

Aiming at the problems of noise and irrelevant feature filtering in data processing, a feature selection method based on the stacking framework is proposed. Use the K-Fold cross-validation method to train and save DNN and SVM-based learners. The prediction results of the base learners are used as the input of the meta-learner, and the logistic regression learning model is trained and saved; comprehensively analyze the correlation coefficients of the fully connected neural network weight matrix and support vector machine, According to the learning results of the meta-learner model, assign different weights to each base learner, calculate the influence factors of each feature, and call the sequence backward search algorithm (SBS) to generate the optimal feature subset. In the experimental stage, a disease diagnosis model was constructed based on the open data set of heart disease research on the Kaggle website, and Stacking-SBS was called to generate the optimal feature subset in the feature space, and the performance comparison experiment of the diagnostic model before and after feature selection was performed, and the method was improved with information. (IG), Chi-square test (Chi) and correlation-based feature selection method (CFS) are compared. The results show that the application of this method can not only reduce model training time, but also significantly improve the model's recall rate and F1 value. In addition, this method is significantly better than the other three feature selection methods in terms of performance improvement. Finally, the open data set of cardiovascular research on the Kaggle website is used to verify the generalization ability of Stacking-SBS. The experimental results show that this method can also significantly improve the performance of the disease diagnosis model.

Analisis Perbandingan Kinerja Algoritma Klasifikasi dengan Menggunakan Metode K-Fold Cross Validation

This study aims to compare the performance of two classification data mining algorithms, namely the K-Nearest Neighbor algorithm, and C4.5 using the K-fold cross validation method. The data used in this study are iris public data with a total of 150 data and 3 label target classes, namely iris-setosa, iris-versicolor, and iris-virginica. The training data used is 97% or 145 data from 150 data, and the testing data used is 3% or 5 data, and the number of K in the K-fold cross validation is 30 or 30 times the experimental stage. The results showed that the performance of the K-Nearest Neighbor algorithm was 95.33%, recall was 95.33%, and precision was 96.27%. While the C4.5 algorithm obtained an accuracy of 96.00%, recall of 94.44%, and precision of 93.52%.

A weight optimization-based transfer learning approach for plant disease detection of New Zealand vegetables.

Deep learning (DL) is an effective approach to identifying plant diseases. Among several DL-based techniques, transfer learning (TL) produces significant results in terms of improved accuracy. However, the usefulness of TL has not yet been explored using weights optimized from agricultural datasets. Furthermore, the detection of plant diseases in different organs of various vegetables has not yet been performed using a trained/optimized DL model. Moreover, the presence/detection of multiple diseases in vegetable organs has not yet been investigated. To address these research gaps, a new dataset named NZDLPlantDisease-v2 has been collected for New Zealand vegetables. The dataset includes 28 healthy and defective organs of beans, broccoli, cabbage, cauliflower, kumara, peas, potato, and tomato. This paper presents a transfer learning method that optimizes weights obtained through agricultural datasets for better outcomes in plant disease identification. First, several DL architectures are compared to obtain the best-suited model, and then, data augmentation techniques are applied. The Faster Region-based Convolutional Neural Network (RCNN) Inception ResNet-v2 attained the highest mean average precision (mAP) compared to the other DL models including different versions of Faster RCNN, Single-Shot Multibox Detector (SSD), Region-based Fully Convolutional Networks (RFCN), RetinaNet, and EfficientDet. Next, weight optimization is performed on datasets including PlantVillage, NZDLPlantDisease-v1, and DeepWeeds using image resizers, interpolators, initializers, batch normalization, and DL optimizers. Updated/optimized weights are then used to retrain the Faster RCNN Inception ResNet-v2 model on the proposed dataset. Finally, the results are compared with the model trained/optimized using a large dataset, such as Common Objects in Context (COCO). The final mAP improves by 9.25% and is found to be 91.33%. Moreover, the robustness of the methodology is demonstrated by testing the final model on an external dataset and using the stratified k-fold cross-validation method.

Performance evaluation and comparison of machine learning algorithms for prediction of electrodeposited copper ions

ABSTRACT Machine-learning prediction algorithms have been used in different scenarios by earlier researchers. This article found the application of these algorithms for predictions of the material deposition rate of copper ions through high-speed selective jet electrodeposition (HSSJED). The presented article emphasises the selection of the best suitable machine-learning prediction algorithms among Gaussian Process Regression (GPR), Support Vector Machine (SVM) regression and Linear Regression (LR) for prediction and compares them with each other for selection of the best algorithm which agrees with experimental data. The comparison of performances of these machine-learning algorithms has been done taking root-mean-square error (RMSE), R2, mean-squared error (MSE) and mean absolute error (MAE) as a basic measure for the same, also various graphs have been plotted and discussed for better understanding. The data set was collected by performing the HSSJED experiment on various parameters like electrolyte composition, electrode gap and applied dc potential. K-fold cross-validation method has been adopted for validation. GPR predictions were found better than the other two which also agree with the experimental output as the quantitative comparison has been done and results are discussed. The average values of RMSE, R2, MSE and MAE were 0.057668, 0.96, 0.0032432 and 0.045733 for GPR.

Ensemble Machine Learning approach for evaluating the material characterization of carbon nanotube-reinforced cementitious composites

Time and cost-efficient techniques are essential to avoid extra conventional experimental studies with large data-set for material characterization of composite materials. This study is aimed at providing a correlation between the structural performance and mechanical properties of carbon nano-tubes reinforced cementitious composites through efficient predictive Machine Learning (ML) models. The Flexural (FS) and Compressive (CS) Strength of Carbon Nanotube (CNT)-reinforced composites were predicted based on the data-rich framework provided in the literature. Two different ensembled ML methods including Random Forest (RF) and Gradient Boosting Machine (GBM) were implemented on those data for predicting the CNT-reinforced cement-based composites. Data-set were utilized for training the proposed models through employing SciKit-Learn library in Python, followed by hyper-parameter tuning and k-fold cross-validation method for obtaining an optimum model to predict the target values. It was shown that the CS values predicted by the proposed models were more accurate than the FS counterparts and the developed model of GBM has less sensitivity to the alteration of test data than the proposed RF model. Finally, sensitivity analysis was conducted through Sobol algorithm and the parameters with highest contribution were identified.

Design and development of hybrid optimization enabled deep learning model for COVID-19 detection with comparative analysis with DCNN, BIAT-GRU, XGBoost

The recent investigation has started for evaluating the human respiratory sounds, like voice recorded, cough, and breathing from hospital confirmed Covid-19 tools, which differs from healthy person's sound. The cough-based detection of Covid-19 also considered with non-respiratory and respiratory sounds data related with all declared situations. Covid-19 is respiratory disease, which is usually produced by Severe Acute Respiratory Syndrome Coronavirus-2 (SARS-CoV-2). However, it is more indispensable to detect the positive cases for reducing further spread of virus, and former treatment of affected patients. With constant rise in the COVID-19 cases, there has been a constant rise in the need of efficient and safe ways to detect an infected individual. With the cases multiplying constantly, the current detecting devices like RT-PCR and fast testing kits have become short in supply. An effectual Covid-19 detection model using devised hybrid Honey Badger Optimization-based Deep Neuro Fuzzy Network (HBO-DNFN) is developed in this paper. Here, the audio signal is considered as input for detecting Covid-19. The gaussian filter is applied to input signal for removing the noises and then feature extraction is performed. The substantial features, like spectral roll-off, spectral bandwidth, Mel frequency cepstral coefficients (MFCC), spectral flatness, zero crossing rate, spectral centroid, mean square energy and spectral contract are extracted for further processing. Finally, DNFN is applied for detecting Covid-19 and the deep leaning model is trained by designed hybrid HBO algorithm. Accordingly, the developed Hybrid HBO method is newly designed by incorporating Honey Badger optimization Algorithm (HBA) and Jaya algorithm. The performance of developed Covid-19 detection model is evaluated using three metrics, like testing accuracy, sensitivity and specificity. The developed Hybrid HBO-based DNFN is outpaced than other existing approaches in terms of testing accuracy, sensitivity and specificity of “0.9176, 0.9218 and 0. 9219”. All the test results are validated with the k-fold cross validation method in order to make an assessment of the generalizability of these results. When k-fold value is 9, sensitivity of existing techniques and developed JHBO-based DNFN is 0.8982, 0.8816, 0.8938, and 0.9207. The sensitivity of developed approach is improved by means of gaussian filtering model. The specificity of DCNN is 0.9125, BI-AT-GRU is 0.8926, and XGBoost is 0.9014, while developed JHBO-based DNFN is 0.9219 in k-fold value 9.

A Parallel Classification Model for Marine Mammal Sounds Based on Multi-Dimensional Feature Extraction and Data Augmentation.

Due to the poor visibility of the deep-sea environment, acoustic signals are often collected and analyzed to explore the behavior of marine species. With the progress of underwater signal-acquisition technology, the amount of acoustic data obtained from the ocean has exceeded the limit that human can process manually, so designing efficient marine-mammal classification algorithms has become a research hotspot. In this paper, we design a classification model based on a multi-channel parallel structure, which can process multi-dimensional acoustic features extracted from audio samples, and fuse the prediction results of different channels through a trainable full connection layer. It uses transfer learning to obtain faster convergence speed, and introduces data augmentation to improve the classification accuracy. The k-fold cross-validation method was used to segment the data set to comprehensively evaluate the prediction accuracy and robustness of the model. The evaluation results showed that the model can achieve a mean accuracy of 95.21% while maintaining a standard deviation of 0.65%. There was excellent consistency in performance over multiple tests.

Application of pest detection on vegetable crops using the cnn algorithm as a smart farm innovation to realize food security in the 4.0 era

Pests and diseases are one of the factors that become obstacles in the cultivation of vegetables because they can cause a decrease in the quality and quantity of production. The more varied types of pests have different impacts on crops, so if farmers incorrectly identify the class of pests, the treatment will be ineffective. Therefore, we need a technology that can classify the types of pests on vegetable crops to maintain the quality and quality of the product as well as the abundant harvest. The classification model of pests on vegetables using the deep learning method using the Convolutional Neural Network (CNN) algorithm with a high level of accuracy is the solution to this problem. The application of artificial intelligence in the agricultural sector also supports smart agriculture in Indonesia. Based on the research that has been carried out, the application of pest classification on vegetable crops made by applying the CNN model using the Inception V3 - k-fold cross-validation method has a test accuracy rate of 99%, meaning that the application can perform pest classification correctly.

Extreme Gradient Boosting Regressor Solution for Defy in Drilling of Materials

Drilling is a quite common operation being performed in the manufacturing of components. Instrumental response in drilling is geometrical accuracy and surface integrity of the drilled parts. For the application where geometrical tolerance is very small, an operation is to be carried out very carefully. If not, rejection of drilled samples will be higher and consequently production loss will be higher. The use of prediction model in this scenario is much more appropriate and cost-effective. This research aimed to apply extreme gradient boosting (XGBoost) regressor to develop a drilling prediction model. Drilling experiments were conducted after developing design of experiments with twenty-seven unique sets. Experimental data analysis was then carried out on experimental data sets that have features such as speed, feed, angle, hole length, and surface roughness. After correlation analysis, the k-fold cross validation method was applied for parameterisation. Hyperparameters estimated from the k-fold cross validation were then applied to train and test the XGBoost regressor-based machine learning (ML) model. It is concluded from the model evaluation metric (R2) that the XGBoost regressor model has resulted 0.89 before tuning and 0.94 after tuning of the model, which is higher than the polynomial regressor and support vector regressor.

New SHapley Additive ExPlanations (SHAP) Approach to Evaluate the Raw Materials Interactions of Steel-Fiber-Reinforced Concrete.

Recently, artificial intelligence (AI) approaches have gained the attention of researchers in the civil engineering field for estimating the mechanical characteristics of concrete to save the effort, time, and cost of researchers. Consequently, the current research focuses on assessing steel-fiber-reinforced concrete (SFRC) in terms of flexural strength (FS) prediction by employing delicate AI techniques as well as to predict the raw material interaction that is still a research gap. In this study, the FS of SFRC is estimated by deploying supervised machine learning (ML) techniques, such as DT-Gradient Boosting, DT-XG Boost, DT-AdaBoost, and DT-Bagging. In addition to that, the performance model is also evaluated by using R2, root mean square error (RMSE), and mean absolute error (MAE). Furthermore, the k-fold cross-validation method is also applied to validate the model’s performance. It is observed that DT-Bagging with an R2 value of 0.95 is superior to DT-XG Boost, DT-Gradient Boosting, and DT-AdaBoost. Lesser error MAE and RMSE and higher R2 values for the DT-Bagging model show the enhanced performance of the model compared to the other ensembled approaches. Considerable conservation of time, effort, and cost can be made by applying ML techniques to predict concrete properties. The evaluation of the outcome depicts that the estimated results of DT-Bagging are closer to the experimental results, indicating the accurate estimation of SFRC flexural strength. It is further revealed from the SHapley Additive exPlanations (SHAP) study that the volumetric content of steel fiber highly and positively influences the FS of SFRC.

Machine learning approach for classification of mangifera indica leaves using digital image analysis

ABSTRACT There is a wide range of horticulture farming in Asia. Mangifera Indica belongs to the species of flowering plant, also publicly recognized as mango. It has a significant local demand as well as a broad export marketplace throughout the world, and is considered as ‘King of Fruits.’ There are many mango varieties and each has its own business market. Efficient identification of the mango varieties is still difficult because of untrained growers and obsolete farming culture, especially in remote areas of the Asia. The primary purpose of this research study was to discriminate mango varieties with the potential of machine learning techniques by analyzing their leaves. For the purpose, we selected leaves of eight mango varieties, namely: Anwar-Ratul (AR), Chaunsa (CHAUN), Langra (LANG), Sindhri (SIND), Saroli (SARO), Fajri (FAJ), Desi (DESI), Alo-Marghan (ALM). A digital cell phone camera captured these datasets in open atmosphere without any well-equipped lab and infrastructure. Binary, histogram, RST, spectral, and texture features were employed for machine learning (ML)-based mango leaf image discrimination. A k-fold (k = 10) cross-validation method was used for ML classification. The k nearest neighbors (KNN) classifier achieved maximum overall classification accuracy (OCA) from 88.33% to 97%.

Landslide susceptibility analysis based on a PSO-DBN prediction model in an earthquake-stricken area

In recent years, the major geological hazard of landslides has greatly impact normal human life. Deep belief networks (DBN) is a commonly used deep learning model, and the DBN hyperparameter determination problem is the key to its application. To improve the accuracy of regional landslide susceptibility prediction, this paper introduces the particle swarm algorithm (PSO) to determine the hyperparameters of the DBN; this is applied to regional landslide susceptibility prediction. Firstly, PSO is used to optimize the hyperparameters of the DBN and obtain a set of hyperparameters with the optimal fitness function. A landslide susceptibility prediction model based on PSO-DBN is then constructed and the K-fold cross-validation method is used to determine its accuracy. The model is applied to landslide susceptibility prediction in the most impacted area of the Wenchuan earthquake to analyze the model’s accuracy. Finally, model susceptibility analysis is performed. The research results show that the final optimal model accuracy of the PSO-DBN model is 95.52%, which is approximately 28.31% and 15.35% higher than that of the logistic regression (LR) model and the common DBN model, respectively. The Kappa coefficient is 0.883, which is higher than that of the LR model. Compared with the LR model and the common DBN model, Kappa coefficient is improved by approximately 0.542 and 0.269 respectively; the area under the curve (AUC) is 0.951, which is improved by approximately 0.201 and 0.080 compared to the LR model and the common DBN model. The susceptibility of the model to the inertia factor is low, the average change in model accuracy (when the inertia factor changes by 0.1) is approximately 0.1%, and the overall stability of the model is high. The landslide susceptibility level is very high. The area includes 219 landslide points, which account for 39.2% of total landslide points. In the area with a high level of landslide susceptibility are 191 landslide points, accounting for 34.2% of total landslide points. Together, the two contain approximately 73.4% of the landslide points. This indicates that the model prediction results agree well with the spatial distribution characteristics of the landslide.

SKCV: Stratified K-fold cross-validation on ML classifiers for predicting cervical cancer

Cancer is the unregulated development of abnormal cells in the human body system. Cervical cancer, also known as cervix cancer, develops on the cervix’s surface. This causes an overabundance of cells to build up, eventually forming a lump or tumour. As a result, early detection is essential to determine what effective treatment we can take to overcome it. Therefore, the novel Machine Learning (ML) techniques come to a place that predicts cervical cancer before it becomes too serious. Furthermore, four common diagnosis testing namely, Hinselmann, Schiller, Cytology, and Biopsy have been compared and predicted with four common ML models, namely Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbors (K-NNs), and Extreme Gradient Boosting (XGB). Additionally, to enhance the better performance of ML models, the Stratified k-fold cross-validation (SKCV) method has been implemented over here. The findings of the experiments demonstrate that utilizing an RF classifier for analyzing the cervical cancer risk, could be a good alternative for assisting clinical specialists in classifying this disease in advance.

Predicting protection capacities of pyrimidine-based corrosion inhibitors for mild steel/HCl interface using linear and nonlinear QSPR models.

Pyrimidine compounds have proven to be effective and efficient additives capable of protecting mild steel in acidic media. This class of organic compounds often functions as adsorption-type inhibitors of corrosion by forming a protective layer on the metallic substrate. The present study reports a computational study of forty pyrimidine compounds that have been investigated as sustainable inhibitors of mild steel corrosion in molar HCl solution. Quantitative structure property relationship was conducted using linear (multiple linear regression) and nonlinear (artificial neural network) models. Standardization method was employed in variable selection yielding five top chemical descriptors utilized for model development along with the inhibitor concentration. Multiple linear regression model yielded a fair predictive model. Artificial neural network model developed using k-fold cross-validation method provided a comprehensive insight into the corrosion protection mechanism of studied pyrimidine-based corrosion inhibitors. Using a multilayer perceptron with Levenberg-Marquardt algorithm, the study obtained the optimal model having a MSE of 8.479, RMSE of 2.912, MAD of 1.791, and MAPE of 2.648. The optimal neural network model was further utilized to forecast the protection capacities of nine non-synthesized pyrimidine derivatives. The predicted inhibition efficiencies ranged from 89 to 98%, revealing the significance of the considered chemical descriptors, the predictive capacity of the developed model, and the potency of the theoretical inhibitors.

Use of Artificial Intelligence for Predicting Parameters of Sustainable Concrete and Raw Ingredient Effects and Interactions.

Incorporating waste material, such as recycled coarse aggregate concrete (RCAC), into construction material can reduce environmental pollution. It is also well-known that the inferior properties of recycled aggregates (RAs), when incorporated into concrete, can impact its mechanical properties, and it is necessary to evaluate the optimal performance. Accordingly, artificial intelligence has been used recently to evaluate the performance of concrete compressive behaviour for different types of construction material. Therefore, supervised machine learning techniques, i.e., DT-XG Boost, DT-Gradient Boosting, SVM-Bagging, and SVM-Adaboost, are executed in the current study to predict RCAC’s compressive strength. Additionally, SHapley Additive exPlanations (SHAP) analysis shows the influence of input parameters on the compressive strength of RCAC and the interactions between them. The correlation coefficient (R2), root mean square error (RMSE), and mean absolute error (MAE) are used to assess the model’s performance. Subsequently, the k-fold cross-validation method is executed to validate the model’s performance. The R2 value of 0.98 from DT-Gradient Boosting supersedes those of the other methods, i.e., DT- XG Boost, SVM-Bagging, and SVM-Adaboost. The DT-Gradient Boosting model, with a higher R2 value and lower error (i.e., MAE, RMSE) values, had a better performance than the other ensemble techniques. The application of machine learning techniques for the prediction of concrete properties would consume fewer resources and take less time and effort for scholars in the respective engineering field. The forecasting of the proposed DT-Gradient Boosting models is in close agreement with the actual experimental results, as indicated by the assessment output showing the improved estimation of RCAC’s compressive strength.

Learning to Utilize Curiosity: A New Approach of Automatic Curriculum Learning for Deep RL

In recent years, reinforcement learning algorithms based on automatic curriculum learning have been increasingly applied to multi-agent system problems. However, in the sparse reward environment, the reinforcement learning agents get almost no feedback from the environment during the whole training process, which leads to a decrease in the convergence speed and learning efficiency of the curriculum reinforcement learning algorithm. Based on the automatic curriculum learning algorithm, this paper proposes a curriculum reinforcement learning method based on the curiosity model (CMCL). The method divides the curriculum sorting criteria into temporal-difference error and curiosity reward, uses the K-fold cross validation method to evaluate the difficulty priority of task samples, uses the Intrinsic Curiosity Module (ICM) to evaluate the curiosity priority of the task samples, and uses the curriculum factor to adjust the learning probability of the task samples. This study compares the CMCL algorithm with other baseline algorithms in cooperative-competitive environments, and the experimental simulation results show that the CMCL method can improve the training performance and robustness of multi-agent deep reinforcement learning algorithms.

Sentiment Analysis About COVID-19 Booster Vaccine on Twitter Using Deep Learning

The rapid spread of COVID-19 cases to various countries has made the COVID-19 outbreak a global pandemic by the World Health Organization (WHO). The effect of the designation of COVID-19 as a pandemic has prompted the government to take preventive action against vaccination, as well as the WHO which has asked the public to immediately get a third or booster dose of vaccine. Various responses regarding the COVID-19 booster vaccine continue to emerge on social media such as Twitter. Twitter is often used by its users to express emotions about something either positive or negative. People tend to believe what they find on social networks, which makes them vulnerable to rumors and fake news. Sentiment analysis or opinion mining is one solution to overcome the problem of automatically classifying opinions or reviews into positive or negative opinions. In this study, the Deep Learning algorithm was used to analyze public opinion sentiment regarding the COVID-19 booster vaccine on Twitter. The data collection method used is crawling data using an access token obtained from the Twitter API. Meanwhile, to evaluate the model, the K-fold Cross-Validation method is used. The results of testing the model obtained the highest accuracy value at iterations = 10, which is 82.78% with AUC value = 0.836, precision = 83.33% and recall = 95.89%.

Sparse representation of brain signals offers effective computation of cortico-muscular coupling value to predict the task-related and non-task sEMG channels: A joint hdEEG-sEMG study.

Cortico-muscular interactions play important role in sensorimotor control during motor task and are commonly studied by cortico-muscular coherence (CMC) method using joint electroencephalogram-surface electromyogram (EEG-sEMG) signals. As noise and time delay between the two signals weaken the CMC value, coupling difference between non-task sEMG channels is often undetectable. We used sparse representation of EEG channels to compute CMC and detect coupling for task-related and non-task sEMG signals. High-density joint EEG-sEMG (53 EEG channels, 4 sEMG bipolar channels) signals were acquired from 15 subjects (30.26 ± 4.96 years) during four specific hand and foot contraction tasks (2 dynamic and 2 static contraction). Sparse representations method was applied to detect projection of EEG signals on each sEMG channel. Bayesian optimization was employed to select best-fitted method with tuned hyperparameters on the input feeding data while using 80% data as the train set and 20% as test set. K-fold (K = 5) cross-validation method was used for evaluation of trained model. Two models were trained separately, one for CMC data and the other from sparse representation of EEG channels on each sEMG channel. Sensitivity, specificity, and accuracy criteria were obtained for test dataset to evaluate the performance of task-related and non-task sEMG channels detection. Coupling values were significantly different between grand average of task-related compared to the non-task sEMG channels (Z = -6.33, p< 0.001, task-related median = 2.011, non-task median = 0.112). Strong coupling index was found even in single trial analysis. Sparse representation approach (best fitted model: SVM, Accuracy = 88.12%, Sensitivity = 83.85%, Specificity = 92.45%) outperformed CMC method (best fitted model: KNN, Accuracy = 50.83%, Sensitivity = 52.17%, Specificity = 49.47%). Sparse representation approach offers high performance to detect CMC for discerning the EMG channels involved in the contraction tasks and non-tasks.

Clinical Characteristics, Prognostic Factor and a Novel Dynamic Prediction Model for Overall Survival of Elderly Patients With Chondrosarcoma: A Population-Based Study.

BackgroundChondrosarcoma is the most common primary bone sarcoma among elderly population. This study aims to explore independent prognostic factors and develop prediction model in elderly patients with CHS.MethodsThis study retrospectively analyzed the clinical data of elderly patients diagnosed as CHS between 2004 and 2018 from the Surveillance, Epidemiology, and End Results (SEER) database. We randomly divided enrolled patients into training and validation group, univariate and multivariate Cox regression analyses were used to determine independent prognostic factors. Based on the identified variables, the nomogram was developed and verified to predict the 12-, 24-, and 36-month overall survival (OS) of elderly patients with CHS. A k-fold cross-validation method (k=10) was performed to validate the newly proposed model. The discrimination, calibration and clinical utility of the nomogram were assessed using the Harrells concordance index (C-index), receiver operating characteristic (ROC) curve and the area under the curve (AUC), calibration curve, decision curve analysis (DCA), the integrated discrimination improvement (IDI) and net reclassification index (NRI). Furthermore, a web-based survival calculator was developed based on the nomogram.ResultsThe study finally included 595 elderly patients with CHS and randomized them into the training group (419 cases) and validation group (176 cases) at a ratio of 7:3. Age, sex, grade, histology, M stage, surgery and tumor size were identified as independent prognostic factors of this population. The novel nomogram displayed excellent predictive performance, which can be accessible by https://nomoresearch.shinyapps.io/elderlywithCHS/, with a C-index of 0.800 for the training group and 0.789 for the validation group. The value AUC values at 12-, 24-, and 36-month of 0.866, 0.855, and 0.860 in the training group and of 0.839, 0.856, and 0.840 in the validation group, respectively. The calibration curves exhibited good concordance from the predicted survival probabilities to actual observation. The ROC curves, IDI, NRI, and DCA showed the nomogram was superior to the existing AJCC staging system.ConclusionThis study developed a novel web-based nomogram for accurately predicting probabilities of OS in elderly patients with CHS, which will contribute to personalized survival assessment and clinical management for elderly patients with CHS.