Predicting Classification Accuracy of Unlabeled Datasets Using Multiple Deep Neural Networks

This study aims to provide useful information for the conservation and management of Brachydiplax chalybea flavovittata (Ris, 1911) (Insecta: Odonata), which is a climate-sensitive biological indicator species, and to develop a more accurate and optimal distribution model that aligns with the research objectives. To achieve this, we intend to evaluate two factors that influence the prediction accuracy of the distribution model. In order to compare the prediction accuracy of the models based on biased location data in a specific region and based on the multicollinearity of environmental variables known as bioclimatic variables, the confirmed habitat locations of B. chalybea flavovittata were collected from the GBIF database, and 19 biological climate variables used as environmental variables were collected from the Worldclim database. The bias of the location data was examined using four types of data selected with a certain distance between the location data. The multicollinearity of the environmental variables was assessed using the Variation Inflation Factor (VIF). Preprocessed data were applied to seven species distribution models, and the prediction accuracy of the models was compared and analyzed using three evaluation metrics (Kappa, TSS, AUC). Among machine learning and regression models, seven distribution models were used for comparison. The machine learning models included Boosted Regression Trees (BRT), Random Forest (RF), Support Vector Machine (SVM), Classification and Regression Trees (CART), and MaxEnt, while the regression models included Generalized Linear Models (GLM) and Multivariate Adaptive Regression Splines (MARS). Overall, the prediction accuracy of the models was better explained by the Random Forest (RF) model, which is a machine learning method that better captures the complex relationship between species and environmental variables, compared to other models. The bias of location data, which is more commonly surveyed in easily accessible areas, affected the prediction accuracy in all models, while multicollinearity varied depending on the model. This study is expected to serve as fundamental guidance for developing an optimal species distribution model that aligns with the research objectives of species conservation and it can be utilized as foundational data for effective management of wildlife habitat in national parks.

The influencing factors of community risk are complex. For the low accuracy of traditional prediction model, a multichannel convolutional neural network community risk prediction model is proposed by improving convolutional neural network of deep learning. First of all, in the community risk prediction model, the structure of multichannel input convolutional neural network is selected. Then, add it into the full connection layer. Subsequently, the DenseNet layer is added to establish connections between different network layers. Finally, the receptive field is improved, and the gradient disappearance is solved. Thus, the prediction accuracy of model is improved. Compared with the traditional model, the proposed multichannel convolutional neural network model has better prediction accuracy. In addition, it performs better on the three indicators, namely, correlation coefficient R , coefficient of determination R 2 , and mean square root error RMSE. Compared with the commonly used LSTM model and logic regression model, the proposed model also has certain advantages, which is more suitable for community risk prediction.

AimTo investigate the optimal training dataset size (TDS) for respiration prediction accuracy using a long short-term memory (LSTM) model.MethodsThe respiratory signals of 151 patients acquired with the real-time position management system were retrospectively included in this study. Among the dataset, 101 respiratory signals were utilized to evaluate the impact of the TDS on prediction accuracy, while the remaining 50 signals were employed for setting the default hyperparameters. The prediction accuracy of the LSTM model using eight different TDSs (10 s, 20 s, 30 s, 60 s, 90 s, 110 s, 130 s, and 150 s) was examined and evaluated by the root mean square error (RMSE) between the real and predicted respiratory signals. The interplay effects of the main hyperparameters, the ahead time and the different testing data lengths using different TDSs were also measured.ResultsFor the 520 ms ahead time, the root mean square error values of the LSTM model using the eight different training data sizes listed above were 0.146 cm, 0.137 cm, 0.134 cm, 0.125 cm, 0.120 cm, 0.121 cm, 0.121 cm, and 0.119 cm, respectively. The LSTM model achieved the highest prediction accuracy when the TDS was 150 s. The prediction accuracy was stable when the TDS exceeded 90 s.ConclusionsTDS selection could influence the respiration signal prediction accuracy of the LSTM model. The relationship between TDS and the prediction accuracy of the LSTM model was not linear. The 90 s seemed to be an optimal TDS for the respiration signal prediction tasks using the LSTM model, as it was the shortest time at which a favorable prediction accuracy was maintained in this study.

Objective: To compare the performance of multiple machine learning algorithms in predicting recurrence after resection of early-stage hepatocellular carcinoma(HCC). Methods: Clinical data of 882 early-stage HCC patients who were admitted to the First Affiliated Hospital of Nanjing Medical University from May 2009 to December 2019 and treated with curative surgical resection were retrospectively collected. There were 701 males and 181 females,with an age of (57.3±10.5)years(range:21 to 86 years). All patients were randomly assigned in a 2∶1 ratio, the training dataset consisted of 588 patients and the test dataset consisted of 294 patients. The construction of machine learning-based prediction models included random survival forest(RSF),gradient boosting machine,elastic net regression and Cox regression model. The prediction accuracy of the model was measured by the concordance index(C-index). The prediction error of the model was measured by the integrated Brier score. Model fit was assessed by the calibration plot. The performance of machine learning models with that of rival model and HCC staging systems was compared. All models were validated in the independent test dataset. Results: Median recurrence-free survival was 61.7 months in the training dataset while median recurrence-free survival was 61.9 months in the validation dataset, there was no significant difference between two datasets in terms of recurrence-free survival(χ²=0.029,P=0.865). The RSF model consisted of 5 commonly used clinicopathological characteristics, including albumin-bilirubin grade,serum alpha fetoprotein,tumor number,type of hepatectomy and microvascular invasion. In both training and test datasets,the RSF model provided the best prediction accuracy,with respective C-index of 0.758(95%CI:0.725 to 0.791) and 0.749(95%CI:0.700 to 0.797),and the lowest prediction error,with respective integrated Brier score of 0.171 and 0.151. The prediction accuracy of RSF model for recurrence after resection of early-stage HCC was superior to that of other machine learning models,rival model(ERASL model) as well as HCC staging systems(BCLC,CNLC and TNM staging),with statistically significant difference(P<0.01). Calibration curves demonstrated good agreement between RSF model-predicted probabilities and observed outcomes.All patients could be stratified into low-risk,intermediate-risk or high-risk group based on RSF model;statistically significant differences among three risk groups were observed in both training and test datasets(all P<0.01). The risk stratification of RSF model was superior to that of TNM staging. Conclusion: The proposed RSF model assembled with 5 commonly used clinicopathological characteristics in this study can predict the recurrence risk with favorable accuracy that may facilitate clinical decision-support for patients with early-stage HCC.

Laser-induced breakdown spectroscopy (LIBS) has been applied to many fields for the quantitative analysis of diverse materials. Improving the prediction accuracy of LIBS regression models is still of great significance for the Mars exploration in the near future. In this study, we explored the quantitative analysis of LIBS for the one-dimensional ChemCam (an instrument containing a LIBS spectrometer and a Remote Micro-Imager) spectral data whose spectra are produced by the ChemCam team using LIBS under the Mars-like atmospheric conditions. We constructed a convolutional neural network (CNN) regression model with unified parameters for all oxides, which is efficient and concise. CNN that has the excellent capability of feature extraction can effectively overcome the chemical matrix effects that impede the prediction accuracy of regression models. Firstly, we explored the effects of four activation functions on the performance of the CNN model. The results show that the CNN model with the hyperbolic tangent (tanh) function outperforms the CNN models with the other activation functions (the rectified linear unit function, the linear function and the Sigmoid function). Secondly, we compared the performance among the CNN models using different optimization methods. The CNN model with the stochastic gradient descent optimization and the initial learning rate = 0.0005 achieves satisfactory performance compared to the other CNN models. Finally, we compared the performance of the CNN model, the model based on support vector regression (SVR) and the model based on partial least square regression (PLSR). The results exhibit the CNN model is superior to the SVR model and the PLSR model for all oxides. Based on the above analysis, we conclude the CNN regression model can effectively improve the prediction accuracy of LIBS.

One of the hottest topics in current research is the use of different kinds of deep learning models for the identification and forecasting of important indicators in the sintering process. The model's poor prediction accuracy results from the fact that current research approaches mostly focus on chemical composition analysis or single sintering picture data analysis, making it challenging to properly investigate the complementary information between several forms of heterogeneous data. Using a convolutional neural network (CNN) framework, this study builds a Fuzzy Kolmogorov-Arnold Networks (KAN) + Transformer (KanFuzz-Trans) model that offers a precise and effective way to estimate the FeO concentration of sintered ore intelligently. To accurately depict the complex non-linear relationship between FeO content and various influencing factors, the model fully utilises the spatial feature extraction capability of CNN, the network adaptive feature optimisation capability of fuzzy KAN, and the multi-attention mechanism of Transformer. These models can capture the global feature dependency relationship, connection between several influencing elements and FeO content. The prediction accuracy is greatly increased by the model's ability to fully take into account numerous feature information by combining multimodal input, such as text and images, throughout the sintering process. Furthermore, by combining characteristics from several sources of heterogeneous data, this model significantly improves prediction accuracy, as evidenced by its improved performance metrics ( R 2 = 0.93, mean squared error = 0.046, root mean squared error = 0.068, mean absolute error = 0.054). The new model presented in this article overcomes the main drawbacks of multimodal information utilisation and prediction accuracy when compared to previous methods. It also offers a fresh concept for achieving intelligent prediction and online detection of FeO content in sintered ores.

We evaluated a physics-based model for planning for magnetic resonance-guided laser interstitial thermal therapy for focal brain lesions. Linear superposition of analytical point source solutions to the steady-state Pennes bioheat transfer equation simulates laser-induced heating in brain tissue. The line integral of the photon attenuation from the laser source enables computation of the laser interaction with heterogeneous tissue. Magnetic resonance thermometry data sets (n = 31) were used to calibrate and retrospectively validate the model’s thermal ablation prediction accuracy, which was quantified by the Dice similarity coefficient (DSC) between model-predicted and measured ablation regions (T > 57 °C). A Gaussian mixture model was used to identify independent tissue labels on pre-treatment anatomical magnetic resonance images. The tissue-dependent optical attenuation coefficients within these labels were calibrated using an interior point method that maximises DSC agreement with thermometry. The distribution of calibrated tissue properties formed a population model for our patient cohort. Model prediction accuracy was cross-validated using the population mean of the calibrated tissue properties. A homogeneous tissue model was used as a reference control. The median DSC values in cross-validation were 0.829 for the homogeneous model and 0.840 for the heterogeneous model. In cross-validation, the heterogeneous model produced a DSC higher than that produced by the homogeneous model in 23 of the 31 brain lesion ablations. Results of a paired, two-tailed Wilcoxon signed-rank test indicated that the performance improvement of the heterogeneous model over that of the homogeneous model was statistically significant (p < 0.01).

Abstract. According to the empirical orthogonal function (EOF), the non-stationary time series data are decomposed into time function and space function, so this mathematical method can simplify the non-stationary time series and eliminate redundant information, thus it performs well in non-stationary time series analysis. The ionospheric Vertical Total Electron Content (VTEC) is a non-stationary time series, which has non-stationary and seasonal variation and the activity of VTEC is more active in low latitudes. Guangxi is located in the middle and low latitudes of the Northern Hemisphere with abundant sunshine in summer and autumn. The energy released by solar radiation makes the ionospheric activity in this region more complex than that in the high latitudes. However, no expert or scholar has used EOF analysis method to conduct a comprehensive study of the low latitudes. The International GNSS Service (IGS) provided by high precision Global Ionospheric Maps (GIM) center in Guangxi are used in the modeling data, the GIM data of the first 10 days of different seasons are decomposed by EOF, and then the time function is predicted by ARIMA model. VTEC values for the next five days are obtained through reconstruction, and relative accuracy and standard deviation are used as accuracy evaluation criteria. The results of EOF-ARIMA model are compared with those of ARIMA model, and the prediction accuracy of EOF-ARIMA model at the equatorial anomaly is analyzed in order to explore the reliability of the model in the more complex region of ionospheric activity. The results show that the average relative precision of EOF-ARIMA model is 84.0, the average standard deviation is 7.45TECu, the average relative precision of ARIMA model is 81.5, the average standard deviation is 8.29TECu, and the precision of EOF-ARIMA model is higher than that of ARIMA model.; There is no significant seasonal difference in the prediction accuracy of EOF-ARIMA model, and the prediction accuracy of ARIMA model in autumn is lower than that of other seasons, which indicates that the prediction results of EOF-ARIMA model are more reliable; The prediction accuracy of the EOF-ARIMA model at the equatorial anomaly is not affected, and it is consistent with the accuracy of the high latitude area in Guangxi. It is shown that the EOF-ARIMA model has high accuracy and stability in the short-term ionospheric prediction in Guangxi at low latitudes of China, and provides a new and reliable method for ionospheric prediction at low latitudes.

Automatic identification of consistently defined body regions in medical images is vital in many applications. In this paper, we describe a method to automatically demarcate the superior and inferior boundaries for neck, thorax, abdomen, and pelvis body regions in computed tomography (CT) images. For any three-dimensional (3D) CT image I, following precise anatomic definitions, we denote the superior and inferior axial boundary slices of the neck, thorax, abdomen, and pelvis body regions by NS(I), NI(I), TS(I), TI(I), AS(I), AI(I), PS(I), and PI(I), respectively. Of these, by definition, AI(I)=PS(I), and so the problem reduces to demarcating seven body region boundaries. Our method consists of a two-step approach. In the first step, a convolutional neural network (CNN) is trained to classify each axial slice in I into one of nine categories: the seven body region boundaries, plus legs (defined as all axial slices inferior to PI(I)), and the none-of-the-above category. This CNN uses a multichannel approach to exploit the interslice contrast, providing the neural network with additional visual context at the body region boundaries. In the second step, to improve the predictions for body region boundaries that are very subtle and that exhibit low contrast, a recurrent neural network (RNN) is trained on features extracted by CNN, limited to a flexible window about the predictions from the CNN. The method is evaluated on low-dose CT images from 442 patient scans, divided into training and testing sets with a ratio of 70:30. Using only the CNN, overall absolute localization error for NS(I), NI(I), TS(I), TI(I), AS(I), AI(I), and PI(I) expressed in terms of number of slices (nS) is (mean±SD): 0.61±0.58, 1.05±1.13, 0.31±0.46, 1.85±1.96, 0.57±2.44, 3.42±3.16, and 0.50±0.50, respectively. Using the RNN to refine the CNN's predictions for select classes improved the accuracy of TI(I) and AI(I) to: 1.35±1.71 and 2.83±2.75, respectively. This model outperforms the results achieved in our previous work by 2.4, 1.7, 3.1, 1.1, and 2 slices, respectively for TS(I), TI(I), AS(I), AI(I)=PS(I), and PI(I) classes with statistical significance. The model trained on low-dose CT images was also tested on diagnostic CT images for NS(I), NI(I), and TS(I) classes; the resulting errors were: 1.48±1.33, 2.56±2.05, and 0.58±0.71, respectively. Standardized body region definitions are a prerequisite for effective implementation of quantitative radiology, but the literature is severely lacking in the precise identification of body regions. The method presented in this paper significantly outperforms earlier works by a large margin, and the deviations of our results from ground truth are comparable to variations observed in manual labeling by experts. The solution presented in this work is critical to the adoption and employment of the idea of standardized body regions, and clears the path for development of applications requiring accurate demarcations of body regions. The work is indispensable for automatic anatomy recognition, delineation, and contouring for radiation therapy planning, as it not only automates an essential part of the process, but also removes the dependency on experts for accurately demarcating body regions in a study.

The availability of data from electronic health records facilitates the development and evaluation of risk-prediction models, but estimation of prediction accuracy could be limited by outcome misclassification, which can arise if events are not captured. We evaluate the robustness of prediction accuracy summaries, obtained from receiver operating characteristic curves and risk-reclassification methods, if events are not captured (i.e., "false negatives"). We derive estimators for sensitivity and specificity if misclassification is independent of marker values. In simulation studies, we quantify the potential for bias in prediction accuracy summaries if misclassification depends on marker values. We compare the accuracy of alternative prognostic models for 30-day all-cause hospital readmission among 4548 patients discharged from the University of Pennsylvania Health System with a primary diagnosis of heart failure. Simulation studies indicate that if misclassification depends on marker values, then the estimated accuracy improvement is also biased, but the direction of the bias depends on the direction of the association between markers and the probability of misclassification. In our application, 29% of the 1143 readmitted patients were readmitted to a hospital elsewhere in Pennsylvania, which reduced prediction accuracy. Outcome misclassification can result in erroneous conclusions regarding the accuracy of risk-prediction models.

Due to the high pour point of the oil products transported in the long-distance high wax crude oil pipeline, in order to ensure the operation safety, it is necessary to adopt heating transmission technology, so as to ensure that the oil temperature along the pipeline is 3–5 °C higher than the pour point, that is to say, the oil temperature is the most important operation parameter of the long-distance hot oil pipeline, and the accurate prediction and control of the oil temperature is the premise of the pipeline safety optimization. Aiming at the problems of large prediction error and poor applicability of the previous theoretical formula, this paper studies the establishment of oil temperature prediction model by using data mining algorithms such as Back Propagation (BP) neural network, and improves the prediction efficiency and accuracy of the model by using Genetic Algorithm (GA) optimization. The correlation coefficient formula is used to calculate the influence coefficient of oil temperature, ground temperature, pipeline transportation and other parameters on the inlet oil temperature of the downstream station, so as to obtain the input parameters of the model. The actual production data training model is downloaded through SCADA system, and the prediction accuracy of the control model is ±0.5 °C. Compared with BP model and other theoretical formulas, the accuracy and efficiency of GA-BP oil temperature prediction model are greatly improved, and the adaptability is better. The GA-BP oil temperature prediction model trained according to the actual production data can be effectively applied to the future pipeline big data platform, which lays a theoretical foundation for the intelligent control of the pipeline.

A. Purpose: Predicting recurrence and survival in individual patients with localized breast cancer will identify circumstance-specific events and interventions that either predispose or prevent recurrence. B. Methods and Procedures: We used the SEER-Medicare linked dataset to investigate women diagnosed with stage I-IV breast cancer who were enrolled at 65 years or older for age eligibility. We collected time-fixed data on patients and cancers at diagnosis and time-varying covariates after diagnosis (e.g., treatments, comorbidities, age, frailty index, adverse events, anxiety, and depression). Our longitudinal data comprised hundreds of thousands of patients with hundreds of millions of records spanning over 20 years. We identified the recurrence of stage I-III disease by documenting new diagnoses of recurrent, contralateral, new chemo-, bio, hormone, or radiotherapy after 4 months following completion of initial therapy, and for all patients, recorded date of death. We have previously demonstrated that combining time-varying with time-fixed covariates into DL modeling of survival for individual patients at all stages results in significant improvements in the model's prediction accuracy (from around 65% to &amp;gt;90%) for stage I-IV patients and considers the impact of subsequent lines of potential individually tailored therapy on survival of stage IV patients. In an earlier study, we extended four deep learning models (DL) to deal with combined time-fixed and time-varying patient covariates to predict survival for individual patients at all stages. The results showed improvements in the model's prediction accuracy (from around 65% to &amp;gt;90%). In this study, we applied four deep learning models (DL) to predict individual patients' recurrence-free survival probabilities in different patient and cancer categories distributed according to race, stage, and hormone receptor status. Results: The predictive accuracy of the models was greater than 95%. Patient recurrence-free survival curves generated by the DL models reveal a vast variability in predicted survival within each broad patient grouping (stage, race, hormone status). Our results show that approximately 36% of the patient population had a diagnosis of anxiety and/or depression, with a higher prevalence in White (W) patients. Our results also demonstrate that the adrenergic stressors, anxiety and depression, previously suspected factors in recurrence, increase the population recurrence rate of dormant breast cancer by 27%. Conclusions: Our modeling confirms the exceptional circumstance-specific variability in interpatient recurrence and survival probabilities. It demonstrates the capacity to model individual patient recurrence and the impact of anxiety/depression as proof of the principle of life events that can affect recurrence from localized breast cancer. The application of these models will serve as a vital tool for testing relevant hypotheses for recurrence-inducing or recurrence-preventing events in individual circumstances that can be tested in clinical trials with a high likelihood of success. Citation Format: Nabil R. Adam, Tarek R. Adam, Robert Wieder. Cancer Recurrence and Anxiety/Depression: High Accuracy Artificial Intelligence Predictive Modeling [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Artificial Intelligence and Machine Learning; 2025 Jul 10-12; Montreal, QC, Canada. Philadelphia (PA): AACR; Clin Cancer Res 2025;31(13_Suppl):Abstract nr B059.

Prediction of species distribution and its changes play an increasingly important role in the fields of ecological protection and application as well as global climate changes. It is impracticable to survey species distribution in large area, especially for rare species; therefore species distribution models are very useful for predicting species distribution. Sampling size has an essential influence on the cost of actual survey and the accuracy of model prediction. Generally, the larger sample size, the higher accuracy of models, but with the more cost of survey. It is necessary to research species distribution models to get the highest accuracy but with the least sample size. Taking 31 different sample sizes (5, 10, 15, 20, 25, 30, 40, 50, 60, 70, 80, 90, 100, 120, 150, 180, 200, 220, 250, 300, 350, 400, 450, 500, 550, 600, 650, 700, 800, 900, 1000 and 1200) of 4 species as an example, this study analyzed the influence of different sample sizes on accuracy and stability of both MaxEnt (Maximum Entropy Species Prediction Model) and GARP(Genetic Algorithm for Rule-set Production). The results showed that both the accuracy and stability of two models increased with the sample size. And the two models had different predictive results to different species. The omission values of GARP were always smaller than that of MaxEnt at the same sample size to all species.

This paper proposes a neural network model based on deep learning, which can better analyze the impact of short video live broadcast on sports machinery and equipment. Firstly, this paper proposes a U-Net-based convolutional neural network as the backbone network of this paper, which mainly realizes the impact of short video live broadcast on sales. Secondly, this paper proposes a dense residual module based on the Transformer lightweight module, which can effectively improve the global modeling ability of the network model and improve the prediction accuracy of the network model. Finally, through a large number of experiments, it is proved that the convolutional neural network based on U-Net proposed in this paper can be better used for the task of short video live broadcast for the sales of sports machinery and equipment, and achieves better prediction accuracy and reasoning speed.

The reasonable pricing of options can effectively help investors avoid risks and obtain benefits, which plays a very important role in the stability of the financial market. The traditional single option pricing model often fails to meet the ideal expectations due to its limited conditions. Combining an economic model with a deep learning model to establish a hybrid model provides a new method to improve the prediction accuracy of the pricing model. This includes the usage of real historical data of about 10,000 sets of CSI 300 ETF options from January to December 2020 for experimental analysis. Aiming at the prediction problem of CSI 300ETF option pricing, based on the importance of random forest features, the Convolutional Neural Network and Long Short-Term Memory model (CNN-LSTM) in deep learning is combined with a typical stochastic volatility Heston model and stochastic interests CIR model in parameter models. The dual hybrid pricing model of the call option and the put option of CSI 300ETF is established. The dual-hybrid model and the reference model are integrated with ridge regression to further improve the forecasting effect. The results show that the dual-hybrid pricing model proposed in this paper has high accuracy, and the prediction accuracy is tens to hundreds of times higher than the reference model; moreover, MSE can be as low as 0.0003. The article provides an alternative method for the pricing of financial derivatives.