Lower Prediction Error Research Articles

Prediction of evapotranspiration variance in the Budyko framework with the incorporation of soil storage and runoff

Water availability needs to be accurately assessed to understand and effectively manage hydrologic environments. However, the estimation of evapotranspiration (ET) is prone to errors due to the complex interactions that occur between the atmosphere, the Earth's surface, and vegetation cover. This paper proposes a novel approach for analyzing the sources of inaccuracy in estimating the annual ET using the Budyko framework (BF), particularly temporal variability in precipitation (P), potential evapotranspiration (EP), runoff (R), and the change in soil storage (ΔS). Error decomposition is employed to determine the individual contributions of P, R, EP, and ΔS to the ET error variance at 12 locations in the state of Illinois using a dataset covering a 22-year period. To the best of our knowledge, this study represents the first BF-based investigation that considers R in the error decomposition of the predicted ET variance. The ET error variance increases with the variance in the P and R in Illinois and decreases with the covariance between these two variables. In addition, when accounting for ΔS in the BF, the scenario in which ΔS affects the total available water (i.e., P) is reliable, with a low prediction error and a 13.87 % lower root mean square error compared with the scenario in which the effect of ΔS is negligible. We thus recommend the inclusion of ΔS and R as key variables in the BF to improve water budget estimations.

Web search volume as a near-real-time complementary surveillance tool of tick-borne encephalitis (TBE) in Italy

The Internet is an important gateway for accessing health-related information, and data generated through web queries have been increasingly used as a complementary source for monitoring and forecasting of infectious diseases and they may partially address the issue of underreporting. In this study, we assessed whether tick-borne encephalitis (TBE)-related Internet search volume may be useful as a complementary tool for TBE surveillance in Italy. Monthly Google Trends (GT) data for TBE-related information were extracted for the period between January 2017 and September 2022, corresponding to the available time series of TBE notifications in Italy. Time series modeling was performed by applying seasonal autoregressive integrated moving average (SARIMA) models with or without GT data. The search terms relative to tick bites reflected best the observed temporal distribution of TBE cases, showing a correlation coefficient of 0.81 (95 % CI: 0.71–0.88). Particularly, both the reported number of TBE cases and GT searches occurred mainly during the summer. The peak of disease notifications coincided with that of Google searches in 4 of 6 years. Once calibrated, SARIMA models with or without GT data were applied to a validation set. Retrospective forecast made by the model with GT data was associated with a lower prediction error and accurately predicted the peak timing. By contrast, the traditional SARIMA model underestimated the actual number of TBE notifications by 65 %. Timeliness, easy availability, low cost and transparency make monitoring of the TBE-related Internet search queries a promising addition to the traditional methods of TBE surveillance in Italy.

A complete artificial intelligence pipeline for radio frequency energy prediction in cellular bands for energy harvesting systems

Radio Frequency (RF) energy harvesting has been used to power wireless and low-powered devices. However, RF energy harvesting has limitations in terms of the amount of power that can be collected based on signal availability. Hence, energy prediction is essential to improve energy harvesting circuits' performance. Previous research has mainly focused on improving power harvesting policies or theoretically estimating the harvested energy. Very few works have considered the prediction of the RF signal as time series data using real RF measurements. Moreover, challenges such as the power consumed by the circuit's harvesting decisions and the impact of outliers on the model performance haven't been addressed yet. This paper presents a complete pipeline for developing the best predictive model for RF energy in cellular frequency bands. Real-time measurements are taken in different frequency bands using software-defined radio technology. We use four artificial intelligence techniques to model the RF energy signal. Additionally, we propose an optimized model with an enhanced loss function, which makes the model more resilient to anomalies, saving computational power and time consumed in cleaning the data. The four algorithms are investigated, and their prediction accuracies are compared. The average power of a period of 5 min is accurately forecasted. Numerical results in the 1960 MHz band show that long short-term memory has the best performance, followed by the DeepAR algorithm with prediction accuracies of 95.76% and 95.02%, respectively. Moreover, the proposed optimized model showed a 32.2% lower prediction error than the traditional models.

Error state prediction of capacitor voltage transformer based on TimesNet and gate control unit

The stability and reliability of the error state of capacitor voltage transformer (CVT) are poor, which affects the fairness of electricity trade settlement and the safe operation level of power grid. However, the CVT error data is superimposed by different periodic information such as daily period, monthly period and quarterly period, which makes the prediction of transformer error state challenging. In this paper, a CVT error state prediction method based on TimesNet and gate control unit (GRU) is proposed. The TimesNet network is employed to capture the intraperiod-variation and interperiod-variation characteristics of the ratio difference data of a-phase, b-phase and c-phase, and the feature data and GRU model are employed to predict the output. The simulation results demonstrate that the mean square error (MSE) and mean absolute error (MAE) of the proposed model are 0.0002 and 0.0101, respectively, indicating that this model has lower prediction error and higher prediction accuracy.

Creating 3D Nanoparticle Structural Space via Data Augmentation to Bidirectionally Predict Nanoparticle Mixture's Purity, Size, and Shape from Extinction Spectra.

Nanoparticle (NP) characterization is essential because diverse shapes, sizes, and morphologies inevitably occur in as-synthesized NP mixtures, profoundly impacting their properties and applications. Currently, the only technique to concurrently determine these structural parameters is electron microscopy, but it is time-intensive and tedious. Here, we create a three-dimensional (3D) NP structural space to concurrently determine the purity, size, and shape of 1000 sets of as-synthesized Ag nanocubes mixtures containing interfering nanospheres and nanowires from their extinction spectra, attaining low predictive errors at 2.7-7.9 %. We first use plasmonically-driven feature enrichment to extract localized surface plasmon resonance attributes from spectra and establish a lasso regressor (LR) model to predict purity, size, and shape. Leveraging the learned LR, we artificially generate 425,592 augmented extinction spectra to overcome data scarcity and create a comprehensive NP structural space to bidirectionally predict extinction spectra from structural parameters with <4 % error. Our interpretable NP structural space further elucidates the two higher-order combined electric dipole, quadrupole, and magnetic dipole as the critical structural parameter predictors. By incorporating other NP shapes and mixtures' extinction spectra, we anticipate our approach, especially the data augmentation, can create a fully generalizable NP structural space to drive on-demand, autonomous synthesis-characterization platforms.

Creating 3D Nanoparticle Structural Space via Data Augmentation to Bidirectionally Predict Nanoparticle Mixture's Purity, Size, and Shape from Extinction Spectra

AbstractNanoparticle (NP) characterization is essential because diverse shapes, sizes, and morphologies inevitably occur in as‐synthesized NP mixtures, profoundly impacting their properties and applications. Currently, the only technique to concurrently determine these structural parameters is electron microscopy, but it is time‐intensive and tedious. Here, we create a three‐dimensional (3D) NP structural space to concurrently determine the purity, size, and shape of 1000 sets of as‐synthesized Ag nanocubes mixtures containing interfering nanospheres and nanowires from their extinction spectra, attaining low predictive errors at 2.7–7.9 %. We first use plasmonically‐driven feature enrichment to extract localized surface plasmon resonance attributes from spectra and establish a lasso regressor (LR) model to predict purity, size, and shape. Leveraging the learned LR, we artificially generate 425,592 augmented extinction spectra to overcome data scarcity and create a comprehensive NP structural space to bidirectionally predict extinction spectra from structural parameters with &lt;4 % error. Our interpretable NP structural space further elucidates the two higher‐order combined electric dipole, quadrupole, and magnetic dipole as the critical structural parameter predictors. By incorporating other NP shapes and mixtures′ extinction spectra, we anticipate our approach, especially the data augmentation, can create a fully generalizable NP structural space to drive on‐demand, autonomous synthesis‐characterization platforms.

Use of responsible artificial intelligence to predict health insurance claims in the USA using machine learning algorithms

Aim: This study investigates the potential of artificial intelligence (AI) in revolutionizing healthcare insurance claim processing in the USA. It aims to determine the most effective machine learning (ML) model for predicting health insurance claims, leading to cost savings for insurance companies. Methods: Six ML algorithms were used to predict health insurance claims, and their performance was evaluated using various metrics. The algorithms examined include support vector machine (SVM), decision tree (DT), random forest (RF), linear regression (LR), extreme gradient boosting (XGBoost), and k-nearest neighbors (KNN). The research involves a performance assessment that encompasses key metrics. Additionally, a feature importance analysis is conducted to illuminate the critical variables that exert influence on the prediction of insurance claims. Results: The findings demonstrate that the XGBoost and RF models outperformed the other algorithms, displaying the highest R-squared values of 79% and 77% and the lowest prediction errors. The feature importance analysis underscores the pivotal role of variables such as smoking habits, body mass index (BMI), and blood pressure levels in the domain of insurance claim prediction. These results emphasize the degree to which these variables should be included in the formulation of insurance policies and pricing strategies. Conclusions: This study supports the transformative potential of AI, with specific emphasis on the XGBoost model, in extending the precision and efficiency of healthcare insurance claim processing. The identification of key variables and the mitigation of prediction errors not only signal the potential for substantial cost savings but also affirm the potential to integrate AI into healthcare insurance processes. This research supports the value of the utilization of AI as an emerging tool for process optimization and data-informed decision-making within the healthcare insurance domain.

Accurate pH Monitoring of Highly Concentrated Saline Aqueous Solutions (Seawater-like) with a pH Colorimetric Sensor Array.

A pH colorimetric sensor array (CSA) was prepared on a nitrocellulose membrane and used for accurate pH measurement in highly concentrated saline solutions. The CSAs consisted of sensing spots made of a suitable OrMoSil polymer prepared from organo-fluorinated-silane precursors and/or organosilane with tetraethyl orthosilicate hosting an acid-base indicator. Four CSAs were prepared: D, 1F, 2F, and 3F. In D, a nonfluorinated organosilane was present. From 1F to 3F, the concentration of the fluorinated organosilane increased and improved the pH measurement accuracy in highly saline concentrations. No recalibrations were required, and the analytical signal was stable in time. D, 1F, 2F, and 3F were deposited in triplicate, and they were prepared to work in the seawater pH interval (7.50-8.50). The use of fluorinated precursors led to a lower pH prediction error and tailored the interval of the CSA at more basic pH values so that the inflection points of the sigmoidal calibrations of D, 1F, 2F, and 3F moved from 6.97 to 7.98. The overall pH prediction error was 0.10 pH (1F), 0.02 pH (2F), and 0.04 pH units (3F). The CSAs were stable, reversible, reusable, and independent of salinity (S) between 20 and 40. The performances of the CSA were compared with those of a glass electrode, whose pHNIST values were converted in the pHSWS scale through a conversion equation. Being unaffected by the typical drawback of the glass electrode, the CSAs can be used directly in seawater real samples, and it validated the proposed conversion equation.

New labor curves of dilation and station to improve the accuracy of predicting labor progress

BackgroundThe diagnosis of failure to progress, the most common indication for intrapartum cesarean delivery, is based on the assessment of cervical dilation and station over time. Labor curves serve as references of expected changes in dilation and fetal descent. The labor curves of Friedman, Zhang et al and others are based on time alone and derived from mothers with spontaneous labor onset. However, labor induction is now common, and clinicians also consider other factors when assessing labor progress. Labor curves that consider the use of induction and other factors that influence labor progress have the potential to be more accurate and closer to clinical decision-making. ObjectiveTo compare the prediction errors of labor curves based on a single factor (time) or multiple clinically relevant factors using 2 modeling methods: mixed-effects regression, a standard statistical method, and Gaussian processes, a machine learning method. Study designThis was a longitudinal cohort study of changes in dilation and station based on data from 8022 births in nulliparous women with a live, singleton, vertex presenting fetus at ≥35 weeks of gestation with a vaginal delivery. New labor curves of dilation and station were generated with 10-fold cross-validation. External validation was performed using a geographically independent group. Model variables included time from the first exam in the 20 hours before delivery; dilation, effacement and station recorded at the previous examination; cumulative contraction counts; and use of epidural anesthesia and labor induction. To assess model accuracy, we calculated the differences between each model’s predicted value and its corresponding observed value. These prediction errors were summarized using mean absolute error and root mean squared error statistics. Results(1) Dilation curves based on multiple parameters were more accurate than those derived from time alone. (2) The mean absolute error with the multifactor methods were better (lower) than those from the single-factor methods [0.826 cm (95% CI, 0.820–0.832) for the multifactor machine learning and 0.893 cm (95% CI, 0.885–0.901) for the multifactor mixed-effects method and 2.122 cm (95% CI, 2.108–2.136) for the single-factor methods; P<0.0001 for both comparisons]. (3) The root mean squared errors with the multifactor methods were also better (lower) than those from the single-factor methods [1.126 cm (95% CI, 1.118–1.133) P<0.0001 for the machine learning and 1.172cm (95% CI, 1.164–1.181) for the mixed-effects method and 2.504 cm (95% CI, 2.487–2.521) for the single-factor; P<0.0001 for both comparisons]. (4) The multifactor machine learning dilation models showed small but statistically significant improvements in accuracy compared to the mixed-effects regression models (P<0.0001). (5) The multifactor machine learning method produced a curve of descent with a mean absolute error of 0.512 cm (95% CI, 0.509–0.515) and a root mean squared error of 0.660 cm (95% CI, 0.655–0.666). (6) External validation using independent data produced similar findings. Conclusions(1) Cervical dilation models based on multiple clinically relevant parameters showed improved (lower) prediction errors compared to models based on time alone; (2) the mean prediction errors were reduced by more than 50%; and (3) a more accurate assessment of departure from expected dilation and station may help clinicians optimize intrapartum management.

One-Time Prediction of Battery Capacity Fade Curve under Multiple Fast Charging Strategies

Using different fast charging strategies for lithium-ion batteries can affect the degradation rate of the batteries. In this case, predicting the capacity fade curve can facilitate the application of new batteries. Considering the impact of fast charging strategies on battery aging, a battery capacity degradation trajectory prediction method based on the TM-Seq2Seq (Trend Matching—Sequence-to-Sequence) model is proposed. This method uses data from the first 100 cycles to predict the future capacity fade curve and EOL (end of life) in one-time. First, features are extracted from the discharge voltage-capacity curve. Secondly, a sequence-to-sequence model based on CNN, SE-net, and GRU is designed. Finally, a trend matching loss function is designed based on the common characteristics of capacity fade curves to constrain the encoding features of the sequence-to-sequence model, facilitating the learning of the underlying relationship between inputs and outputs. TM-Seq2Seq model is verified on a public dataset with 132 battery cells and multiple fast charging strategies. The experimental results indicate that, compared to other popular models, the TM-Seq2Seq model has lower prediction errors.

HyFish: hydrological factor fusion for prediction of fishing effort distribution with VMS dataset

Predicting fishing effort distribution is crucial for guiding fisheries management in developing effective strategies and protecting marine ecosystems. This task requires a deep understanding of how various hydrological factors, such as water temperature, surface height, salinity, and currents influence fishing activities. However, there are significant challenges in designing the prediction model. Firstly, how hydrological factors affect fishing effort distributions remains unquantified. Secondly, the prediction model must effectively integrate the spatial and temporal dynamics of fishing behaviors, a task that shows analytical difficulties. In this study, we first quantify the correlation between hydrological factor fields and fishing effort distributions through spatiotemporal analysis. Building on the insights from this analysis, we develop a deep-learning model designed to forecast the daily distribution of fishing effort for the upcoming week. The proposed model incorporates residual networks to extract features from both the fishing effort distribution and the hydrological factor fields, thus addressing the spatial limits of fishing activity. It also employs Long Short-Term Memory (LSTM) networks to manage the temporal dynamics of fishing activity. Furthermore, an attention mechanism is included to capture the importance of various hydrological factors. We apply the approach to the VMS dataset from 1,899 trawling fishing vessels in the East China Sea from September 2015 to May 2017. The dataset from September 2015 to May 2016 is used for correlation analysis and training the prediction model, while the dataset from September 2016 to May 2017 is employed to evaluate the prediction accuracy. The prediction error ratio for each day of the upcoming week range is only 5.6% across all weeks from September 2016 to May 2017. HyFish, notable for its low prediction error ratio, will serve as a versatile tool in fisheries management for developing sustainable practices and in fisheries research for providing quantitative insights into fishing resource dynamics and assessing ecological risks related to fishing activities.

What’s in Your Fruit Juice?—Rapid Quality Screening Based on Infrared (FT-IR) Spectroscopy

Fruit juices (FJ) have gained widespread global consumption, driven by their perceived health benefits. The accuracy of nutrition information is essential for consumers assessing FJ quality, especially with increasing concerns about added sugars and obesity risk. Conversely, ascorbic acid (Vitamin C), found in nature in many fruits and vegetables, is often lost due to its susceptibility to light, air, and heat, and it undergoes fortification during FJ production. Current analytical methods for determining FJ components are time-consuming and labor-intensive, prompting the need for rapid analytical tools. This study employed a field-deployable portable FT-IR device, requiring no sample preparation, to simultaneously predict multiple quality traits in 68 FJ samples from US markets. Using partial least square regression (PLSR) models, a strong correlation (RCV ≥ 0.93) between FT-IR predictions and reference values was obtained, with a low standard error of prediction. Remarkably, 21% and 37% of FJs deviated from nutrition label values for sugars and ascorbic acid, respectively. Portable FT-IR devices offer non-destructive, simultaneous, simple, and high-throughput approaches for chemical profiling and real-time prediction of sugars and acid levels in FJs. Their handiness and ruggedness can provide food processors with a valuable “out-of-the-laboratory” analytical tool.

Age Prediction Using Resting-State Functional MRI.

The increasing lifespan and large individual differences in cognitive capability highlight the importance of comprehending the aging process of the brain. Contrary to visible signs of bodily ageing, like greying of hair and loss of muscle mass, the internal changes that occur within our brains remain less apparent until they impair function. Brain age, distinct from chronological age, reflects our brain's health status and may deviate from our actual chronological age. Notably, brain age has been associated with mortality and depression. The brain is plastic and can compensate even for severe structural damage by rewiring. Functional characterization offers insights that structural cannot provide. Contrary to the multitude of studies relying on structural magnetic resonance imaging (MRI), we utilize resting-state functional MRI (rsfMRI). We also address the issue of inclusion of subjects with abnormal brain ageing through outlier removal. In this study, we employ the Least Absolute Shrinkage and Selection Operator (LASSO) to identify the 39 most predictive correlations derived from the rsfMRI data. The data is from a cohort of 176 healthy right-handed volunteers, aged 18-78 years (95/81 male/female, mean age 48, SD 17) collected at the Mind Research Imaging Center at the National Cheng Kung University. We establish a normal reference model by excluding 68 outliers, which achieves a leave-one-out mean absolute error of 2.48 years. By asking which additional features that are needed to predict the chronological age of the outliers with a smaller error, we identify correlations predictive of abnormal aging. These are associated with the Default Mode Network (DMN). Our normal reference model has the lowest prediction error among published models evaluated on adult subjects of almost all ages and is thus a candidate for screening for abnormal brain aging that has not yet manifested in cognitive decline. This study advances our ability to predict brain aging and provides insights into potential biomarkers for assessing brain age, suggesting that the role of DMN in brain aging should be studied further.

Soft measurement model for wet gas flow rate based on ultrasonic and differential pressure sensing

Non-separation measurement of gas–liquid two-phase flow is of great significance to industrial production and theoretical research on two-phase flow. In this paper, a soft measurement method of wet gas flow rate based on ultrasonic and differential pressure (DP) sensors is proposed. A total of 129 sets of experimental data are obtained in DN50 horizontal pipe, the superficial velocity of gas and liquid ranging from 5 m s−1 to 33 m s−1 and 0.015 m s−1 to 0.6 m s−1, respectively. Ten feature parameters of ultrasonic signals such as the time difference between two echoes and kurtosis factor are proposed from time and frequency domain. As well as the power spectral density, mean value and standard deviation value are analyzed based on DP signals. Furthermore, the Spearman correlation coefficient is introduced to evaluate these feature parameters quantitatively. And nine parameters which are highly correlated to gas–liquid flow rate are selected as the inputs of the soft measurement model. Finally, Decision Tree, least square support vector machines and artificial neural network (ANN) models are established through training with the experimental data. Comparison results show that ANN has the lowest prediction errors on the test set, with the mean absolute percentage errors of 1.49% and 3.2% for gas and liquid, respectively.

Metabolic alterations of peripheral blood immune cells and heterogeneity of neutrophil in intracranial aneurysms patients.

Intracranial aneurysms (IAs) represent a severe cerebrovascular disease that can potentially lead to subarachnoid haemorrhage. Previous studies have demonstrated the involvement of peripheral immune cells in the formation and progression of IAs. Nevertheless, the impact of metabolic alterations in peripheral immune cells and changes in neutrophil heterogeneity on the occurrence and progression of IAs remains uncertain. Single-cell Cytometry by Time-of-Flight (CyTOF) technology was employed to profile the single-cell atlas of peripheral blood mononuclear cells (PBMCs) and polymorphonuclear cells (PMNs) in 72 patients with IAs. In a matched cohort, metabolic shifts in PBMC subsets of IA patients were investigated by contrasting the expression levels of key metabolic enzymes with their respective counterparts in the healthy control group. Simultaneously, compositional differences in peripheral blood PMNs subsets between the two groups were analysed to explore the impact of altered heterogeneity in neutrophils on the initiation and progression of IAs. Furthermore, integrating immune features based on CyTOF analysis and clinical characteristics, we constructed an aneurysm occurrence model and an aneurysm growth model using the random forest method in conjunction with LASSO regression. Different subsets exhibited distinct metabolic characteristics. Overall, PBMCs from patients elevated CD98 expression and increased proliferation. Conversely, CD36 was up-regulated in T cells, B cells and monocytes from the controls but down-regulated in NK and NKT cells. The comparison also revealed differences in the metabolism and function of specific subsets between the two groups. In terms of PMNs, the neutrophil landscape within patients group revealed a pronounced shift towards heightened complexity. Various neutrophil subsets from the IA group generally exhibited lower expression levels of anti-inflammatory functional molecules (IL-4 and IL-10). By integrating clinical and immune features, the constructed aneurysm occurrence model could precisely identify patients with IAs with high prediction accuracy (AUC=0.987). Furthermore, the aneurysm growth model also exhibited superiority over ELAPSS scores in predicting aneurysm growth (lower prediction errors and out-of-bag errors). These findings enhanced our understanding of peripheral immune cell participation in aneurysm formation and growth from the perspectives of immune metabolism and neutrophil heterogeneity. Moreover, the predictive model based on CyTOF features holds the potential to aid in diagnosing and monitoring the progression of human IAs.

Wind Energy Assessment in Forested Regions Based on the Combination of WRF and LSTM-Attention Models

In recent years, the energy crisis has become increasingly severe, and global attention has shifted towards the development and utilization of wind energy. The establishment of wind farms is gradually expanding to encompass forested regions. This paper aims to create a Weather Research and Forecasting (WRF) model suitable for simulating wind fields in forested terrains, combined with a long short-term time (LSTM) neural network enhanced with attention mechanisms. The simulation focuses on capturing wind characteristics at various heights, short-term wind speed prediction, and wind energy assessment in forested areas. The low-altitude observational data are obtained from the flux tower within the study area, while high-altitude data are collected using mobile radar. The research findings indicate that the WRF simulations using the YSU boundary layer scheme and MM5 surface layer scheme are applicable to forested terrains. The LSTM model with attention mechanisms exhibits low prediction errors for short-term wind speeds at different heights. Furthermore, based on the WRF simulation results, a wind energy assessment is conducted for the study area, demonstrating abundant wind energy resources at the 150 m height in forested regions. This provides valuable support for the site selection in wind farm development.

Predicting Implantable Collamer Lens Vault Using Machine Learning Based on Various Preoperative Biometric Factors.

To predict the vault size after Implantable Collamer Lens (ICL) V4c implantation using machine learning methods and to compare the predicted vault with the conventional manufacturer's nomogram. This study included 707 patients (707 eyes) who underwent ICL V4c implantation at the Department of Ophthalmology, Peking Union Medical College Hospital, from September 2019 to January 2022. Random Forest Regression (RFR), XGBoost, and linear regression (LR) were used to predict the vault size 1 week after ICL V4c implantation. The mean absolute error (MAE), median absolute error (MedAE), root mean square error (RMSE), symmetric mean absolute percentage error (SMAPE), and Bland-Altman plot were utilized to compare the prediction performance of these machine learning methods. The dataset was divided into a training set of 180 patients (180 eyes) and a test set of 527 patients (527 eyes). XGBoost had the lowest prediction error, with mean MAE, RMSE, and SMAPE values of 121.70µm, 148.87µm, and 19.13%, respectively. The Bland‒Altman plots of RFR and XGBoost showed better prediction consistency than LR. However, XGBoost showed narrower 95% limits of agreement (LoA) than RFR, ranging from -307.12 to 256.59µm. XGBoost demonstrated better predictive performance than RFR and LR, as it had the lowest prediction error and the narrowest 95% LoA. Machine learning may be applicable for vault prediction, and it might be helpful for reducing the complications and the secondary surgery rate. Using the proposed machine learning model, surgeons can consider the postoperative vault to reduce the surgical complications.

Simulasi Progres Proyek Konstruksi Time Performance Menggunakan Earned Value Management dengan Integrasi Artificial Neural Network

Digitization of construction using technologies such as AI, Big Data, Machine Learning, and Internet of Things (IoT) can help improve the productivity and efficiency of the construction industry. Machine learning methods such as Artificial Neural Network (ANN) are used to solve complex problems, including in construction projects. In addition, Earned Value Management (EVM) is used as a method to analyze and control project performance and estimate project completion time. Although EVM has the disadvantage of predicting estimated project completion times that are linear in nature, the use of non-linear methods such as Artificial Neural Network can help improve the accuracy of estimating project completion times that are complex and have a high degree of variation. This research aims to analyze the physical achievements of the work and predict estimates of work completion using Earned Value Management which is integrated with the Artificial Neural Network on the Advanced Development Project for Facilities and Infrastructure of the Sanggabuana Karawang Training Area Maintenance Detachment (Sarpras Denharrahlat) belonging to the Indonesian Ministry of Defense. By using EVM and ANN, project management can be improved in terms of cost control, scheduling, and better estimation of project completion time. The data analysis method used is EVM. The data analyzed is Week, Planned Value and Earned Value. The three data are then plotted and translated in graphical form to calculate the Schedule Variance and Schedule Performance Index to evaluate the project. After conducting an analysis using the EVM, then a predictive model is formed to calculate the estimated project completion time using the Artificial Neural Network (ANN) method. Next, optimization of the predictive model is carried out to reduce the level of prediction error. The Software used to build predictive models is RapidMiner. At week 26, work progress was 100% ahead of plan, with an Earned Schedule value of 31.00. This indicates that the project has been completed in accordance with the target time. Certain parameter modifications to the optimal model include training cycles of 100, learning rate of 0.316, and momentum of 0.316. With these parameter settings, the model produces the most accurate predictions with a low prediction error rate.

Performance improvement of underwater LIBS qualitative and quantitative analysis by irradiating with long nanosecond pulses.

Long nanosecond pulses have been proven to be efficient at enhancing underwater LIBS emission. However, the quantitative analytical capability of underwater long-pulse LIBS has yet to be further revealed. In this work, we investigated the spectral characteristics by irradiating with a laser pulse of 120 ns duration. The alkali and alkaline earth metals Li, K and Ca and the transition element Mn were selected for analysis. It is shown that obvious self-reversal structures were observed in the spectra at high concentrations, making the calibration curves saturated. Correction was performed using the approximate Voigt function fitting method, which significantly improves the linearity of the calibration curves. In addition to the target metal elements, atomic lines of the matrix elements H and O in water were also observed, which can serve as promising internal standards for quantitative analysis. A comparison of the quantification performance with and without the internal standards demonstrates that the use of the internal standards is conducive to improving the robustness of the calibration approaches with higher determination coefficients. More importantly, the underwater LIBS signal stability is improved by more than 3 times, and the prediction error for validation samples is reduced by 2-4 times. The present results suggest that long ns pulses are favorable to significantly improving the qualitative and quantitative performance of underwater single-pulse LIBS, enabling long-pulse LIBS to have great potential to be applied to underwater in situ chemical analysis.

Research on Global Climate Change Prediction based on Machine Learning Model

Climate prediction models have traditionally relied on complex physical equations to simulate the dynamics of the climate system, but these models often require significant computational resources and long computational lengths. In recent years, machine learning techniques have shown great potential for pattern recognition and prediction. Specifically, machine learning models have become a hot research direction in the field of climate science due to their advantages in processing large-scale datasets. In this work, we propose a convolutional neural network-based (CNN) model capable of processing and analysing large-scale climate datasets from satellites, including multi-dimensional data including temperature, air pressure, humidity, and CO2 concentration. The input is historical climate data, and the spatial features are extracted through the convolutional layer, and then the feature fusion and final prediction output are performed through the fully connected layer. Finally, we utilized historical climate data as the training set and tested the model on data over multiple time periods. The results show that compared with traditional physical models, CNN-based models provide higher accuracy and lower prediction errors in predicting global average temperature changes, precipitation, and extreme weather events.