Performance Metrics Research Articles

Precise and automated lung cancer cell classification using deep neural network with multiscale features and model distillation

Lung diseases globally impose a significant pathological burden and mortality rate, particularly the differential diagnosis between adenocarcinoma, squamous cell carcinoma, and small cell lung carcinoma, which is paramount in determining optimal treatment strategies and improving clinical prognoses. Faced with the challenge of improving diagnostic precision and stability, this study has developed an innovative deep learning-based model. This model employs a Feature Pyramid Network (FPN) and Squeeze-and-Excitation (SE) modules combined with a Residual Network (ResNet18), to enhance the processing capabilities for complex images and conduct multi-scale analysis of each channel's importance in classifying lung cancer. Moreover, the performance of the model is further enhanced by employing knowledge distillation from larger teacher models to more compact student models. Subjected to rigorous five-fold cross-validation, our model outperforms existing models on all performance metrics, exhibiting exceptional diagnostic accuracy. Ablation studies on various model components have verified that each addition effectively improves model performance, achieving an average accuracy of 98.84% and a Matthews Correlation Coefficient (MCC) of 98.83%. Collectively, the results indicate that our model significantly improves the accuracy of disease diagnosis, providing physicians with more precise clinical decision-making support.

Automated machine learning model for fundus image classification by health-care professionals with no coding experience

To assess the feasibility of code-free deep learning (CFDL) platforms in the prediction of binary outcomes from fundus images in ophthalmology, evaluating two distinct online-based platforms (Google Vertex and Amazon Rekognition), and two distinct datasets. Two publicly available datasets, Messidor-2 and BRSET, were utilized for model development. The Messidor-2 consists of fundus photographs from diabetic patients and the BRSET is a multi-label dataset. The CFDL platforms were used to create deep learning models, with no preprocessing of the images, by a single ophthalmologist without coding expertise. The performance metrics employed to evaluate the models were F1 score, area under curve (AUC), precision and recall. The performance metrics for referable diabetic retinopathy and macular edema were above 0.9 for both tasks and CFDL. The Google Vertex models demonstrated superior performance compared to the Amazon models, with the BRSET dataset achieving the highest accuracy (AUC of 0.994). Multi-classification tasks using only BRSET achieved similar overall performance between platforms, achieving AUC of 0.994 for laterality, 0.942 for age grouping, 0.779 for genetic sex identification, 0.857 for optic, and 0.837 for normality with Google Vertex. The study demonstrates the feasibility of using automated machine learning platforms for predicting binary outcomes from fundus images in ophthalmology. It highlights the high accuracy achieved by the models in some tasks and the potential of CFDL as an entry-friendly platform for ophthalmologists to familiarize themselves with machine learning concepts.

Diving into a pool of data: Using principal component analysis to optimize performance prediction in women’s short-course swimming

ABSTRACT This study aimed to optimise performance prediction in short-course swimming through Principal Component Analyses (PCA) and multiple regression. All women’s freestyle races at the European Short-Course Swimming Championships were analysed. Established performance metrics were obtained including start, free-swimming, and turn performance metrics. PCA were conducted to reduce redundant variables, and a multiple linear regression was performed where the criterion was swimming time. A practical tool, the Potential Predictor, was developed from regression equations to facilitate performance prediction. Bland and Altman analyses with 95% limits of agreement (95% LOA) were used to assess agreement between predicted and actual swimming performance. There was a very strong agreement between predicted and actual swimming performance. The mean bias for all race distances was less than 0.1s with wider LOAs for the 800 m (95% LOA −7.6 to + 7.7s) but tighter LOAs for the other races (95% LOAs −0.6 to + 0.6s). Free-Swimming Speed (FSS) and turn performance were identified as Key Performance Indicators (KPIs) in the longer distance races (200 m, 400 m, 800 m). Start performance emerged as a KPI in sprint races (50 m and 100 m). The successful implementation of PCA and multiple regression provides coaches with a valuable tool to uncover individual potential and empowers data-driven decision-making in athlete training.

Developing a Measurement Framework for Ethiopian Dry Port Sustainability: An Empirical Study

Developing a Measurement Framework for Ethiopian Dry Port Sustainability: An Empirical Study

Snake swarm optimization‐based deep reinforcement learning for resource allocation in edge computing environment

SummaryMobile edge computing has become popular in the past few years as a means of creating computing resources close to end‐user nodes at the network edge. Nodes—end users—demand work offloading to improve service utilization. Furthermore, when the number of users in mobile edge computing increases, the minimal resources deployed at the edge become a problem. Develop the idea of reinforcement learning using a metaheuristic technique intended to achieve effective resource allocation and resolve offloading issues to handle this issue. The ideal way to manage the implementation of mobile edge computing with a cognitive agent's help is to request compensation for all client necessities. To complete the infrastructure type for the Internet of Things (IoT), the operator information is combined with its distinctive methodology. Neural caching during task execution is provided by reinforcement learning based on snake swarm optimization (SSO). Neural caching during task execution is provided by reinforcement learning based on SSO. In the process of creating the cost mapping table and incentive factor‐based optimal resource allocation, this suggested method is applied to a contract with effective resource allocation among the end manipulators. Using performance metrics like delivery ratio, energy consumption, throughput, and delay, the suggested approach is put into practice and examined. It is also contrasted with traditional methods like Gray Wolf Optimization (GWO) ant colony optimization (ACO) and genetic algorithms (GA).

Artificial Intelligence-Driven Model for Gold Price Prediction

Artificial Intelligence-Driven Model for Gold Price Prediction

Study on the hydrodynamics of structured packing: Liquid holdup and pressure drop of a novel 3D printed packing

SpiroPak is a patented innovative structured packing and can be used in process industries for chemical separation purposes. The core of its design is to ensure an uninterrupted flow path for both liquids and gases, leading to remarkably low-pressure drops but with substantial surface areas for efficient mass transfer. Experimental studies using the wire-mesh sensor technique have showcased SpiroPak's ability to eliminate the channelling effect that is observed in commercial packings. Furthermore, its wide-ranging liquid holdup capability encompasses the functionalities of conventional packings, offering clients the feasibility of diverse applications. In this study, the pressure-drop correlations for SpiroPak in this work are based on Stichlmair empirical formulas, with the coefficients of C1, C2 and C3 tuned through the curve fitting method. The primary objective of this study was to establish packing selection criteria based on the key hydraulic performance metrics, such as pressure drop and liquid holdup. By swiftly identifying suitable packing solutions, the method accelerates the pre-selection process, significantly reducing computational complexity and time expenditures at the early stages of research and development.

Reservoir temperature prediction based on characterization of water chemistry data—case study of western Anatolia, Turkey

Reservoir temperature estimation is crucial for geothermal studies, but traditional methods are complex and uncertain. To address this, we collected 83 sets of water chemistry and reservoir temperature data and applied four machine learning algorithms. These models considered various input factors and underwent data preprocessing steps like null value imputation, normalization, and Pearson coefficient calculation. Cross-validation addressed data volume issues, and performance metrics were used for model evaluation. The results revealed that our machine learning models outperformed traditional fluid geothermometers. All machine learning models surpassed traditional methods. The XGBoost model, based on the F-3 combination, demonstrated the best prediction accuracy with an R2 of 0.9732, while the Bayesian ridge regression model using the F-4 combination had the lowest performance with an R2 of 0.8302. This study highlights the potential of machine learning for accurate reservoir temperature prediction, offering geothermal professionals a reliable tool for model selection and advancing our understanding of geothermal resources.

Comprehensive Performance Evaluation between Visual SLAM and LiDAR SLAM for Mobile Robots: Theories and Experiments

SLAM (Simultaneous Localization and Mapping), primarily relying on camera or LiDAR (Light Detection and Ranging) sensors, plays a crucial role in robotics for localization and environmental reconstruction. This paper assesses the performance of two leading methods, namely ORB-SLAM3 and SC-LeGO-LOAM, focusing on localization and mapping in both indoor and outdoor environments. The evaluation employs artificial and cost-effective datasets incorporating data from a 3D LiDAR and an RGB-D (color and depth) camera. A practical approach is introduced for calculating ground-truth trajectories and during benchmarking, reconstruction maps based on ground truth are established. To assess the performance, ATE and RPE are utilized to evaluate the accuracy of localization; standard deviation is employed to compare the stability during the localization process for different methods. While both algorithms exhibit satisfactory positioning accuracy, their performance is suboptimal in scenarios with inadequate textures. Furthermore, 3D reconstruction maps established by the two approaches are also provided for direct observation of their differences and the limitations encountered during map construction. Moreover, the research includes a comprehensive comparison of computational performance metrics, encompassing Central Processing Unit (CPU) utilization, memory usage, and an in-depth analysis. This evaluation revealed that Visual SLAM requires more CPU resources than LiDAR SLAM, primarily due to additional data storage requirements, emphasizing the impact of environmental factors on resource requirements. In conclusion, LiDAR SLAM is more suitable for the outdoors due to its comprehensive nature, while Visual SLAM excels indoors, compensating for sparse aspects in LiDAR SLAM. To facilitate further research, a technical guide was also provided for the researchers in related fields.

Enhanced performance of CMIP6 climate models in simulating historical precipitation in the Florida Peninsula

AbstractAssessing the performance of General Circulation Models (GCMs) in simulating historical regional precipitation is essential for climate assessments. This study analysed 18 GCMs from the Coupled Model Intercomparison Project Phase 5 (CMIP5) and 27 GCMs from Phase 6 (CMIP6) for the Florida Peninsula. Naïve combinations formed the CMIP5‐Ensemble and CMIP6‐Ensemble. The reference dataset consisted of global gridded precipitation data from National Oceanic and Atmospheric Administration. GCM‐simulated precipitation data were compared to the reference dataset across multiple time scales between 1951 and 2005. Most CMIP5 and CMIP6 GCMs exhibited a negative bias at the daily time scale, with an absolute value ranging between 0 and 2 mm day−1 at the median level, except for INM‐CM4 and MRI‐CGCM3 (CMIP5) and CNRM‐ESM2‐1 (CMIP6). Two performance metrics, that is, Pearson correlation and mean absolute error (MAE), were used to evaluate the GCMs. A correlation value above 0.3 is statistically significant at the significance level (α = 0.05). For the Pearson correlation between simulated and observed monthly precipitation, only 15% of CMIP5 GCMs (3 out of 19) had a median value above 0.3, whilst this increased to approximately 39% (11 out of 28) for CMIP6 GCMs. For monthly climatological precipitation, only 21% (4 out of 21) of GCMs in CMIP5 showed a correlation with a median value of 0.8 or higher. This percentage rose to 50% (14 out of 28) for CMIP6. A notable difference was observed in the MAE between CMIP5 and CMIP6 climate models. Lastly, GCM raw outputs were evaluated based on rainy season characteristics (onset, demise and duration). Significant improvements from the CMIP5 to the CMIP6 models were observed in capturing the rainy season in the Florida Peninsula. This study thoroughly compares CMIP5 and CMIP6 climate models in simulating historical precipitation at various time scales for the study region. Although the results are specific to the study area, the methods can be applied more broadly.

Exploiting building information modeling and machine learning for optimizing rooftop photovoltaic systems

The primary objective of this study is to develop a strategy to maximize the potential of Building Applied Photovoltaics (BAPV) by providing researchers and experts in the field with the appropriate tools. By utilizing operational data from an operational roof-mounted BAPV system and incorporating Building Information Modeling (BIM) to improve its design and smooth integration into built settings, the study offers novel insights.A three-phase research methodology is applied, encompassing data collection and performance assessment, BIM modeling, and machine learning algorithms for energy forecasting. The case study involves a 160 kWp photovoltaic system in Abruzzo, Italy, over a three-year period. The research employs different performance metrics recommended by IEC 61724 to compare experimental and theoretical data. BIM simulations are exploited as they are crucial for identifying the correct design process. Furthermore, four cutting-edge machine learning algorithms are used to forecast daily energy production. The random forest is identified as the best model for forecasting and the effect of tuning of hyperparameters on the efficiency of all models is also reported.The research adds to the larger conversation on sustainable energy management and solutions by providing researchers involved in the design and improvement of BAPV systems with a solid foundation for future developments.

Using decision tree models and comprehensive statewide data to predict opioid overdoses following prison release

PurposeIdentifying predictors of opioid overdose following release from prison is critical for opioid overdose prevention. MethodsWe leveraged an individually linked, state-wide database from 2015–2020 to predict the risk of opioid overdose within 90 days of release from Massachusetts state prisons. We developed two decision tree modeling schemes: a model fit on all individuals with a single weight for those that experienced an opioid overdose and models stratified by race/ethnicity. We compared the performance of each model using several performance measures and identified factors that were most predictive of opioid overdose within racial/ethnic groups and across models. ResultsWe found that out of 44,246 prison releases in Massachusetts between 2015–2020, 2237 (5.1%) resulted in opioid overdose in the 90 days following release. The performance of the two predictive models varied. The single weight model had high sensitivity (79%) and low specificity (56%) for predicting opioid overdose and was more sensitive for White non-Hispanic individuals (sensitivity = 84%) than for racial/ethnic minority individuals. ConclusionsStratified models had better balanced performance metrics for both White non-Hispanic and racial/ethnic minority groups and identified different predictors of overdose between racial/ethnic groups. Across racial/ethnic groups and models, involuntary commitment (involuntary treatment for alcohol/substance use disorder) was an important predictor of opioid overdose.

Cooling load prediction of a double-story terrace house using ensemble learning techniques and genetic programming with SHAP approach

Since the cooling systems used in buildings in hot climates account for a significant portion of the energy consumption, it is very important for both economy and environment to accurately predict the cooling load and consider it in building designs. This study aimed to maximize energy efficiency by appropriately selecting the features of a building that affect its cooling load. To this end, data-driven, accurate, and accessible tools were developed that enable the prediction of the cooling load of a building by practitioners. The study involves simulating the energy consumption of a mid-rise, double-story terrace house in Malaysia using building information modeling (BIM) and estimating the cooling load using ensemble machine learning models and genetic programming. Categorical Boosting (CatBoost), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (LightGBM), and Random Forest (RF) models have been developed and made available as an online interactive graphical user interface on the Streamlit platform. Furthermore, the symbolic regression technique has been utilized to obtain a closed-form equation that predicts the cooling load. The dataset used for training the predictive models comprised 94,310 data points with 10 input variables and the cooling load as the output variable. Performance metrics such as the coefficient of determination (R2), root mean squared error (RMSE), and mean absolute error (MAE) were used to measure the predictive model performances. The results of the machine learning models indicated successful prediction, with the CatBoost model achieving the highest score (R2 = 0.9990) among the four ensemble models and the predictive equation. The SHAP analysis determined the aspect ratio of the building as the most impactful feature of the building.

Human resource challenges in health systems: evidence from ten African countries.

Sub-Saharan Africa has fewer medical workers per capita than any region of the world, and that shortage has been highlighted consistently as a critical constraint to improving health outcomes in the region. This paper draws on newly available, systematic, comparable data from ten countries in the region to explore the dimensions of this shortage. We find wide variation in human resources performance metrics, both within and across countries. Many facilities are barely staffed, and effective staffing levels fall further when adjusted for health worker absences. However, caseloads-while also varying widely within and across countries-are also low in many settings, suggesting that even within countries, deployment rather than shortages, together with barriers to demand, may be the principal challenges. Beyond raw numbers, we observe significant proportions of health workers with very low levels of clinical knowledge on standard maternal and child health conditions. This work demonstrates that countries may need to invest broadly in health workforce deployment, improvements in capacity and performance of the health workforce, and on addressing demand constraints, rather than focusing narrowly on increases in staffing numbers.

Performance and Return to Play After Surgery for Thoracic Outlet Syndrome in Professional Baseball Players: A Matched Cohort Analysis.

Thoracic outlet syndrome (TOS) remains a rare diagnosis but is being recognized as a cause of upper extremity dysfunction in professional baseball players. The purpose was to determine performance and return-to-play (RTP) outcomes in professional baseball players after surgical treatment of TOS. The hypothesis was that there would be a high RTP rate in professional baseball players after TOS surgery with no statistical differences in performance between pitchers who had TOS surgery and matched controls. Cohort study; Level of evidence, 3. All professional baseball players who underwent surgical treatment of TOS between 2010 and 2017 were identified using the Major League Baseball Health and Injury Tracking System database. Demographic and performance data (before and after surgery) for each player were recorded. Performance variables were then compared between players who underwent TOS surgery and matched controls. The matching criteria were no history of previous surgeries on affected arm, age at time of surgery, throwing side, level of play (Major or Minor League Baseball), and years of experience playing professional baseball. Overall, 52 players underwent surgery for TOS, of whom 46 (88%) were pitchers. The type of TOS was neurogenic in 69% and venous in 29%. One player had arterial TOS. After TOS surgery, 79% of players returned to play at the same or higher level (RTSP) by 9.5 months and played ≥3 years after surgery. No differences were found in RTSP rate based on the type of TOS. No statistical difference was found in RTP rates between major and minor league players. Pitchers had a 76% RTSP, which was similar to the natural attrition for control pitchers (P = .874). After TOS surgery, pitchers saw a decline in several performance metrics, but these declines were not different from those of control pitchers, indicating that the decline in performance after TOS surgery was no faster than is seen in healthy professional pitchers as they age. The rate of RTSP after surgery for TOS in professional baseball players was 79%. There was no difference in RTP based on the type of TOS. Pitchers who underwent surgery for TOS had no significant differences in pitching performance metrics after surgery compared with matched controls.

202 Source matters: long-term effects of nursery sanitary condition and dietary Zinc source and level on growth performance, diarrhea incidence, and carcass metrics of pigs

Abstract There is growing evidence that requirements for particular nutrients increase when pigs are kept under low sanitary conditions (LSC). It may be hypothesized that dietary zinc (Zn) source can exert influence on immune status, which may influence resistance to subclinical and clinical diseases. The objective was to investigate the long-term effects of nursery sanitary condition and dietary Zn source and growth performance, diarrhea incidence, and carcass metrics of pigs. An experiment was performed with 480 mixed-sex weanling pigs (initial body weight = 6.4 ± 0.38 kg) which were assigned to 1 of 5 treatments during the nursery (42 d) as follows: 1) pigs housed in high sanitary conditions (HSC) and fed 50 mg/kg Zn from Zn-amino acid complex (ZnAA), pigs housed in LSC and fed either 2) 50 mg/kg Zn from ZnAA, 3) 100 mg/kg Zn from ZnAA, 4) 100 mg/kg Zn and 0.4 mg/kg chromium (Cr) from Zn+Cr-amino acid complex (Zn+CrAA), or 5) 100 mg/kg Zn from inorganic Zn (ZnSO4). Pigs were then placed onto a common grower (42 d) and finisher (42 d) diets and housed under similar, standard conditions and were monitored until slaughter. Growth performance, diarrhea incidence, and carcass characteristics were assessed. Pigs housed in LSC and fed 100 mg/kg from ZnSO4 had less average daily feed intake (ADFI) during wk 1 and 2 of nursery (0.29 vs. 0.33 ± 0.01 kg/d; P < 0.05) and poorer feed to gain (FGR) during wk 3 and 4 of nursery (1.76 vs. 1.54 ± 0.03 kg/kg; P < 0.05), compared with the other groups. Overall nursery FGR was poorer in pigs housed in LSC and fed 100 mg/kg from ZnSO4 compared with the other groups housed in LSC (1.63 vs. 1.54 ± 0.02 kg/kg; P < 0.05). Diarrhea incidence tended to increase in all groups housed in LSC compared with the group housed in HSC during week 1 of nursery (91.1 vs. 80.4 ± 3.11%; P < 0.10), while no effect on diarrhea incidence was observed during the subsequent week (P > 0.10). Slaughter weight was increased in pigs housed in LSC and fed 100 mg/kg Zn and 0.4 mg/kg Cr from Zn+CrAA compared with pigs housed in LSC and fed 100 mg/kg from ZnSO4 (123.4 vs. 120.1 ± 0.63 kg; P < 0.05). Housing pigs in LSC during the nursery seems to trigger negative effects on growth performance and diarrhea incidence during nursery only, but these effects do not seem to be translated into the grower and finisher phases. Feeding ZnAA and Zn+CrAA aid the pigs in responding to LSC during nursery and may be an approach to help minimize negative effects of immune system stimulation.

A Machine Learning Approach for Detection and Classification of Colon Cancer using Convolutional Neural Network Architecture

In both developing and developed countries, cancer has become one of the most fearsome health ailments in the 21st century. Among the different variants of cancer, colon cancer is the 3rd most prevalent cancer globally and the 2nd deadliest form of cancer. Early diagnosis is very critical for the successful treatment of colon cancer. In recent times, machine-learning approaches have made significant strides in IoT-assisted healthcare systems. Computer-aided diagnostic systems driven by deep learning algorithms can detect colon cancer with great accuracy, assisting the medical profession in quick diagnosis and developing quick remedies against it. In our approach, we have developed a deep learning model based on convolutional neural network architecture to accurately classify and detect colon cancer. The developed model has been trained with 10,000 histopathological images of the colon, divided into two classes: colon adenocarcinoma and colon benign tissues, each containing 5000 images. We have also employed various performance metrics in our research to monitor the performance of our machine-learning model. The proposed model has been trained for 30 epochs with a batch size of 32 and achieved an overall accuracy of 98.79%.

IMPACT OF MACROECONOMIC FACTORS ON THE INDIAN STOCK MARKET

The research paper investigates the impact of selected macroeconomic factors on the Indian stock market performance. Using a comprehensive dataset spanning key macroeconomic indicators and stock market performance metrics, the statistical analysis techniques to examine the relationship between macroeconomic variables and stock market movements is employed. The study focuses on understanding how changes in Foreign Portfolio Net Investments, Index of Industrial Production, Consumer Price Index and Exchange Rate can influence investor sentiment and trading behaviour in the Indian stock market. The findings of this research contribute to a deeper understanding of the dynamics between macroeconomic factors and stock market performance in India, providing valuable insights for investors, policymakers, and financial institutions. Keywords: Foreign Portfolio Net Investments (FPI), Index of Industrial Production (IIP), Consumer Price Index (CPI), Exchange Rate(ER), NIFTY, Indian Stock Market, Inflation.

Double decomposition with enhanced least-squares support vector machine to predict water level

Double decomposition with enhanced least-squares support vector machine to predict water level

Weather Forecasting: A Comprehensive Exploration of IoT and Machine Learning

Weather forecasting play an important role in human lives.It is impacting on various human activities such as agriculture and transportation.Changes in climate conditions often lead to unpredictable weather. Conventional forecasting models sometimes provide inaccurate predictions because they don't adapt well to the dynamic nature of climate.The integration of (IoT) and Machine Learning (ML) technologies has revolutionized weather forecasting methodologies.This paper explores how combining Machine Learning (ML) algorithms with Internet of Things (IoT) devices improves weather forecasting accuracy and efficiency. By analyzing real-time data and historical patterns, this paper shows how IoT and ML can offer more accurate and timely forecasts compare to the limitations of traditional methods.Accurate weather forecasts are really important for things like farming, transportation, saving energy.In this paper the effectiveness of the algorithm is evaluated using various performance metrics including r2 score comparison,mean square error comparison. Key Words: IOT,Machine Learning,Support Vector Machine