Microsoft Azure Machine Learning Research Articles

Physical education and sport activity assessment tool-based machine learning predictive analysis for planification of training sessions

Background and purpose The aim of this study is to incorporte machine learning techniques in physical education activities assessment so we can plan a training session and learning cycle based on predictive analyses using machine learning algorithms. Material and methods A dataset represent the collection of physical tests (as Harvard test, Vertical and Horizontal Trigger) and activities performance (as 600 m, 1000 m, 12 min cooper) of 600 students in a secondary high school, aged between 15 and 20 years old (mean:16,21, SD:0,92), during 2021-2022 scholar year and project the predicted results on the following learning cycles in the scholar year of 2022-2023. We used Microsoft Azure Machine Learning Studio to obtain the best predictive model based on R2 score as an evaluating metric. Results Even if we focus on one metric test (as a target) with numeric values in this article, the results were promising compared to the predicted values of both physical tests and athletic performances, where we noticed some students have exceeded the expected values to reach. And the predictive analysis unveiled the more important features impacting the predicted results for the physical test. Conclusions Incorporating the Machine Learning techniques may encourage the change in the way we teach physical education and sport activities; otherwise, the assessment based on ML techniques will give a different overview on how to start a learning cycle and follow it up. The obtained predictive model provides an explication of the most impacting features on students’ performance allowing any training planification to relay on their importance respectively based on their density that affects prediction.

Establishment and analysis of online energy baseline for water purification plants: A case study in Kaohsiung city, Taiwan

Water and energy are closely linked and are crucial for national security and economic development. Most water providers prioritise the stability of water supply and aim to reduce energy consumption under the premise of a stable supply. The average energy required to supply water in Taiwan in one of the lowest worldwide. In the Kaohsiung area, the average energy used by a water purification plant to provide 1 m3 of water is 0.32 kWh/m3, lower than the world average of 0.37 kWh/m3. However, the most energy-consuming plant (Weng Park water purification plant) uses eight times as much energy as the least energy-consuming plant (Pingding water purification plant). Most studies focus on the energy required to provide 1 m3 of water. This study combined attributes of four plants, such as the amount of energy consumed, quantity of water supplied, purified, and collected, and weather data. These data were used to model energy baselines for water providers. Artificial intelligence was imported into Microsoft Azure machine learning to train the model, which was verified using another Kaohsiung plant and one overseas to establish an online energy baseline modelling system that can be applied in various water purification plants.

Forecasting Order Delays with Artificial Intelligence-Based Applications

The digitalization trend in the world in the last two decades has begun to transform business processes. Departing from the traditional business approach, the use of technology in every field has changed business processes and, accordingly, the approach of managers. On the other hand, cloud computing, the internet of things, sensors and instant notification systems through these components brought by the last industrial revolution have transformed supply chains into a smart structure. Now, the information that a product sold by a retailer, which is the last link of a supply chain, is out of stock is transmitted to the raw material supplier in the first link of the chain. In this way, a perfect supply chain structure is formed and operations are carried out in lean working principles. The products prepared according to the reduction of the retailer's stocks are continuously supplied, which shows flawless operation. On the other hand, through digital components such as sensors and the Internet of Things, each step in workflows is instantly converted into data and stored in virtual environments called cloud computing. At this point, besides the problem of data storage and security, the main problem is to process the data in a meaningful way. Analyzing big data generated by AI-based components and systems is also possible with AI-based systems. Methods called machine learning have been developed for this situation and their application area is increasing day by day. In the light of this information, within the scope of the study, forecasts of order quantity delays for future periods using historical order data from customers for the products produced by an enterprise were analyzed using machine learning algorithms. The results of the analysis made through the Microsoft Azure Machine Learning Studio platform will contribute to the use of digital tools in the sector as well as increasing the application examples of machine learning. According to the results obtained as a result of the analyzes, recommendations were developed for the enterprise. Finally, suggestions for the application of machine learning in the field of business administration are presented.

Facilitating Model Training With Automated Techniques

Automating artificial intelligence (AI) model training has emerged as a significant challenge in the field of automation. The complete pipeline from raw data to model deployment poses the need to define robust processes that ensure the efficiency of the services that expose the models. This paper introduces a generic architecture for automating data preparation, training of models, selection of models, and deployment of models as web services for third-party consumption using Microsoft Azure Machine Learning’s (AzureML) CI/CD tools. We conducted a practical experiment utilizing AzureML pipelines with predefined and custom modules, demonstrating its readiness for integration into any production application. We also successfully integrated this architecture into a real-world product designed for industrial forecasting. This practical implementation demonstrates the effectiveness and adaptability of our approach, indicating its potential to address diverse training needs. Keywords and phrases: Artificial Intelligence, Automation, Optimization.

Performance Improvement of Melanoma Detection Using a Multi-Network System Based on Decision Fusion

The incidence of melanoma cases continues to rise, underscoring the critical need for early detection and treatment. Recent studies highlight the significance of deep learning in melanoma detection, leading to improved accuracy. The field of computer-assisted detection is extensively explored along all lines, especially in the medical industry, as the benefit in this field is to save hu-man lives. In this domain, this direction must be maximally exploited and introduced into routine controls to improve patient prognosis, disease prevention, reduce treatment costs, improve population management, and improve patient empowerment. All these new aspects were taken into consideration to implement an EHR system with an automated melanoma detection system. The first step, as presented in this paper, is to build a system based on the fusion of decisions from multiple neural networks, such as DarkNet-53, DenseNet-201, GoogLeNet, Inception-V3, InceptionResNet-V2, ResNet-50, ResNet-101, and compare this classifier with four other applications: Google Teachable Machine, Microsoft Azure Machine Learning, Google Vertex AI, and SalesForce Einstein Vision based on the F1 score for further integration into an EHR platform. We trained all models on two databases, ISIC 2020 and DermIS, to also test their adaptability to a wide range of images. Comparisons with state-of-the-art research and existing applications confirm the promising performance of the proposed system.

AN ADAPTIVE DECISION-MAKING SUPPORT MODEL IN THE MANAGEMETN OF ENGINEERING INFRASTRUCTURE RECONSTRUCTION PROGRAMS AND PROJECTS

Reconstruction of engineering infrastructure has become an important topic for Ukraine since the beginning of the full-scale invasion of russian federation in 2022. Standard approach to implementation of programs and projects is inefficient for turbulent environments and therefore the use adaptive approach is relevant. The concept of adaptive management has been analyzed. Means and instruments of adaptive management were analyzed. It was defined that different means and instruments of adaptive management are relevant for different program implementation phases. For the delivery phase these are learning and forecasting, for the closure phase – analysis of obtained experience for more efficient implementation of the next programs, for the definition phase – instruments that can ensure the further implementation of adaptivity to the management processes. An adaptive decision-making support model concept is proposed for adaptive engineering infrastructure reconstruction programs and projects management. The model is based on machine learning methods and can be used for program architecture and project work structure development and management. In this case the decision-making consists in choosing the optimal composition and configuration of the system that is the reconstruction object from among the available alternatives by predicting the values of the parameters of its elements in order to minimize the costs of implementing the program or the project. The model is created with Microsoft Azure Machine Learning Studio, the user interface is created in Microsoft Excel. The distinguishing features of the model are adaptivity due to the use of machine learning methods, possibility of scaling of the model to ensure its application to different system levels and presence of post-processing instruments for different use cases including calculation of additional parameter values, parameter values dependency graphs construction etc. The dataset for the model consists of several parameter categories that characterize the system modelled: technical and technological parameters, environmental parameters, energy efficiency parameters, energy security parameters, economic parameters, operational safety parameters. Keywords: adaptive program management, adaptive project management, decision support systems, machine learning.

Sentiment analysis of epidemiological surveillance reports on COVID-19 in Greece using machine learning models.

The present research deals with sentiment analysis performed with Microsoft Azure Machine Learning Studio to classify Facebook posts on the Greek National Public Health Organization (EODY) from November 2021 to January 2022 during the pandemic. Positive, negative and neutral sentiments were included after processing 300 reviews. This approach involved analyzing the words appearing in the comments and exploring the sentiments related to daily surveillance reports of COVID-19 published on the EODY Facebook page. Moreover, machine learning algorithms were implemented to predict the classification of sentiments. This research assesses the efficiency of a few popular machine learning models, which is one of the initial efforts in Greece in this domain. People have negative sentiments toward COVID surveillance reports. Words with the highest frequency of occurrence include government, vaccinated people, unvaccinated, telephone communication, health measures, virus, COVID-19 rapid/molecular tests, and of course, COVID-19. The experimental results disclose additionally that two classifiers, namely two class Neural Network and two class Bayes Point Machine, achieved high sentiment analysis accuracy and F1 score, particularly 87% and over 35%. A significant limitation of this study may be the need for more comparison with other research attempts that identified the sentiments of the EODY surveillance reports of COVID in Greece. Machine learning models can provide critical information combating public health hazards and enrich communication strategies and proactive actions in public health issues and opinion management during the COVID-19 pandemic.

Consumer sentiment toward international activist advertising

Companies have been increasingly conveying activist advertising messages to international audiences in response to persisting social ills and unsustainable business practices. Given the ambiguity surrounding the effects of international advertising, this study aims to compare the response of local consumers to a multinational brand’s standardized activist advertising strategy with creative executions adapted to the national contexts of Greece and the USA. The selected brand originates from the USA. The paper used a systematic qualitative approach and sentiment analysis using Microsoft Excel and Azure Machine Learning add-in. Analysis conducted on 1,051 user comments in January 2023 regarding two publicly accessible social media posts of a multinational haircare brand showed both similarities and differences in consumer responses. 662 comments were in English and 389 comments were in Greek. The study’s dataset was anonymized entirely and de-identified. The results indicate that consumer sentiment was largely negative in both countries. Although the relative share of negative comments was significantly higher in the USA (96.7%) than in Greece (59.4%), both groups of consumers placed their emphasis on the same themes (e.g., children and products). Therefore, they have perceived the main message similarly, irrespective of the execution differences. Overall, the findings can be attributed to strategic and tactical issues of the activist advertising campaign, as well as to sociocultural particularities of the national context. In the case of international advertising, attempts to incorporate the brand’s stance on a controversial sociopolitical issue into its advertising strategy pose significant risks to business organizations.

Perspectives of using artificial intelligence elements in bread baking

Enterprises of bread baking sector produce traditional everyday products and, at the same time, this sector remains one of the most conservative ones. There are always new types of raw materials and additives on the market, but new recipes are developed traditionally based on the results of test baking. However, a technologically justified solution is not always optimal regarding nutritional value. The purpose of this work is consideration of the possibilities and prospects of using elements of artificial intelligence (AI) in bread baking based on examples from other sectors of the food industry and study the AI’s ability to analyze and evaluate (categorise) existing recipes of bread products. The initial stage of AI use is machine learning (ML), but today there are no unified electronic databases that can be used to fulfil the task. We took approved collections of recipes as the basis for filling such databases. In addition, technologically acceptable variations of the main recipe components and their mutual substitution were carried out. The base was analyzed using Microsoft Azure Machine Learning service and Google Cloud Machine Learning Engine. Normalised database with 5,000 variants of recipes of bakery products that can be manufactured in conditions of both large industrial companies and small bakeries were created. The ML showed that the system effectively determines the main components of recipes and can independently, with high accuracy, classify recipes entered by the user into categories "bread" or "enriched bread". The work showed the possibility and viability of using AI elements in the baking industry. The database can be used to design new bakery products with specified recipe composition and (after improvement) to model applications with specified chemical composition or biological value.

Easy-to-use machine learning system for the prediction of IDH mutation and 1p/19q codeletion using MRI images of adult-type diffuse gliomas.

Adult-type diffuse gliomas are divided into Astrocytoma, IDH-mutant, Oligodendroglioma, IDH-mutant and 1p/19q-codeleted and Glioblastoma, IDH-wildtype based on the IDH mutation, and 1p/19q codeletion status. To determine the treatment strategy for these tumors, pre-operative prediction of IDH mutation and 1p/19q codeletion status might be effective. Computer-aided diagnosis (CADx) systems using machine learning have been noted as innovative diagnostic methods. However, it is difficult to promote the clinical application of machine learning systems at each institute because the support of various specialists is essential. In this study, we established an easy-to-use computer-aided diagnosis system using Microsoft Azure Machine Learning Studio (MAMLS) to predict these statuses. We constructed an analysis model using 258 adult-type diffuse glioma cases from The Cancer Genome Atlas (TCGA) cohort. Using MRI T2-weighted images, the overall accuracy, sensitivity, and specificity for the prediction of IDH mutation and 1p/19q codeletion were 86.9%, 80.9%, and 92.0%, and 94.7%, 94.1%, and 95.1%, respectively. We also constructed an reliable analysis model for the prediction of IDH mutation and 1p/19q codeletion using an independent Nagoya cohort including 202 cases. These analysis models were established within 30min. This easy-to-use CADx system might be useful for the clinical application of CADx in various institutes.

Revolutionizing Social Robotics: A Cloud-Based Framework for Enhancing the Intelligence and Autonomy of Social Robots

Social robots have the potential to revolutionize the way we interact with technology, providing a wide range of services and applications in various domains, such as healthcare, education, and entertainment. However, most existing social robotics platforms are operated based on embedded computers, which limits the robot’s capabilities to access advanced AI-based platforms available online and which are required for sophisticated physical human–robot interactions (such as Google Cloud AI, Microsoft Azure Machine Learning, IBM Watson, ChatGPT, etc.). In this research project, we introduce a cloud-based framework that utilizes the benefits of cloud computing and clustering to enhance the capabilities of social robots and overcome the limitations of current embedded platforms. The proposed framework was tested in different robots to assess the general feasibility of the solution, including a customized robot, “BuSaif”, and commercialized robots, “Husky”, “NAO”, and “Pepper”. Our findings suggest that the implementation of the proposed platform will result in more intelligent and autonomous social robots that can be utilized by a broader range of users, including those with less expertise. The present study introduces a novel methodology for augmenting the functionality of social robots, concurrently simplifying their utilization for non-experts. This approach has the potential to open up novel possibilities within the domain of social robotics.

Development and Validation of Innovative Machine Learning Models for Predicting Date Palm Mite Infestation on Fruits

The date palm mite (DPM), Oligonychus afrasiaticus (McGregor), is a key pest of unripe date fruits. The detection of this mite depends largely on the visual observations of the webs it produces on the green fruits. One of the most important problems of DPM control is the lack of an accurate decision-making approach for monitoring and predicting infestation on date fruits. Therefore, this study aimed to develop, evaluate, and validate prediction models for DPM infestation on fruits based on meteorological variables (temperature, relative humidity, wind speed, and solar radiation) and the physicochemical properties of date fruits (weight, firmness, moisture content, total soluble solids, total sugar, and tannin content) using two machine learning (ML) algorithms, i.e., linear regression (LR) and decision forest regression (DFR). The meteorological variables data in the study area were acquired using an IoT-based weather station. The physicochemical properties of two popular date palm cultivars, i.e., Khalas and Barhee, were analyzed at different fruit development stages. The development and performance of the LR and DFR prediction models were implemented using Microsoft Azure ML. The evaluation of the developed models indicated that the DFR was more accurate than the LR model in predicting the DPM based on the input variables, i.e., meteorological variables (R2 = 0.842), physicochemical properties variables (R2 = 0.895), and the combination of both meteorological and the physicochemical properties variables (R2 = 0.921). Accordingly, the developed DFR model was deployed as a fully functional prediction web service into the Azure cloud platform and the Excel add-ins. The validation of the deployed DFR model showed that it was able to predict the DPM count on date palm fruits based on the combination of meteorological and physicochemical properties variables (R2 = 0.918). The deployed DFR model by the web service of Azure Ml studio enhanced the prediction of the DPM count on the date fruits as a fast and easy-to-use approach. These findings demonstrated that the DFR model using Azure Ml Studio integrated into the Azure platform can be a powerful tool in integrated DPM management.

USING INTERPOLATION FOR GENERATING INPUT DATA FOR THE GROSS DOMESTIC PRODUCT MONTE CARLO SIMULATION

Input modelling is a complex task within the Monte Carlo simulation, especially when the systems and processes under investigation reveal the non-linear behavior of several interdependent variables. Commonly used approaches for Monte Carlo simulation input modelling include selecting probability distributions and fitting them to existing data; resampling random variates from historical data; and using real-world data as an input model, which in the age of big data becomes more feasible. Each of the approaches comes with its own set of drawbacks. This note aims to describe a new method of input modelling for GDP Monte Carlo simulation based on interpolation of the GDP historical records. Also, this method has been implemented as a publicly available online tool using the Microsoft Azure Machine Learning Studio. A similar approach can be applied to other macroeconomic indicators, e.g., consumer price index (inflation) or current employment statistics. This note is intended for economists, data scientists, and operations research analysts interested in the GDP Monte Carlo simulation. It can also be used by academics, researchers, and practitioners in a broad subject area for generating input data for Monte Carlo simulation. Specifically, it can be of interest for Ph.D. candidates (VAC specialty 5.2.6) performing development of theory and methods of decision-making in economic and social systems, and application of artificial intelligence and big data methods in management.

Using machine learning to predict venous thromboembolism in patients with uterine cancer (081)

<b>Objectives:</b> Gynecologic cancers are traditionally associated with a high risk of venous thromboembolism (VTE). Recent ASCO clinical practice guidelines have expanded recommendations for pharmacological thromboprophylaxis in high-risk patients; however, high risk is not well defined for the endometrial cancer population. Both the Caprini and the Khorana scores are clinically validated to assess the risk of VTE - Caprini developed for surgical patients and Khorana for cancer patients undergoing chemotherapy. We sought to develop a better tool to assess endometrial cancer patients at high risk for VTE. <b>Methods:</b> A single-institution retrospective cohort study was performed from January 2016 through January 2020. All patients with uterine cancer were screened, and the patients with complete clinical information were included. VTE was evaluated from diagnosis through 90 days post-diagnosis. Both the Caprini and the Khorana scores were calculated for each patient. High-risk Caprini score was defined as ≥ 9 based on prior studies in cancer patients, whereas high-risk Khorana score was defined as ≥ 3. In addition to variables included in the risk assessment tools, additional variables included race/ethnicity, tumor grade, stage, and histology. Microsoft Azure Machine Learning Studio was utilized to develop a machine learning (ML) algorithm to predict VTE. Five different ML models were assessed, including random forest, extreme random trees (ERT), logistic regression, XGBoost classifier, and support vector machine. Data were split into 70% for ML training and 30% for validation. Models were analyzed based on classification accuracy, AUC, and sensitivity and specificity. <b>Results:</b> A total of 1000 patients were included. The majority of patients were White (93.5%), obese (mean BMI of 37.8 kg/m2) with an average age of 61.2 years. Most tumors were grade 1 (62.7%), endometrioid histology (79.3%), and stage I (76.4%). Most patients underwent a hysterectomy (95.4%). A minority received neoadjuvant (4.0%) and adjuvant chemotherapy (13.1%), whereas 269 patients received radiation (26.9%). Only 18 patients (1.8%) were diagnosed with a VTE within 90 days of diagnosis. The Khorana score had a sensitivity of 38.9% and a specificity of 80.1% for VTE. The Caprini score had a sensitivity of 66.7% and a specificity of 61.0%. Combining the features of the Khorana and Caprini scores, the best-performing ML algorithm had an accuracy of 67.3%, AUC of 83.8%, and the sensitivity improved to 100%, retaining specificity at 66.8% (ERT). To further improve the model, histology and grade were included, which performed better as a random forest model with an accuracy of 75.7%, AUC of 84.8%, sensitivity 100%, and specificity of 75.3%. Lastly, we added stage and race/ethnicity to the model (random forest model), which improved the accuracy to 83.7%, AUC 89.4%, sensitivity 100%, and specificity to 83.4%. In the final model, tumor grade and stage were the most weighted for importance, followed by age, platelet count, histology, and hemoglobin. <b>Conclusions:</b> The use of ML to enhance VTE prediction in endometrial cancer is promising. Using the same parameters as clinically validated tools, we were able to greatly improve prediction, and by adding clinically relevant variables (histology, grade, stage, race/ethnicity), we were able to further improve the model with 100% sensitivity and 83.4% specificity for VTE within 90 days of diagnosis. Using ML, VTE thromboprophylaxis could be more individualized for patients at the highest risk for VTE and minimize the potential risks associated with unnecessary anticoagulation.

A Machine Learning Model to Predict Citation Counts of Scientific Papers in Otology Field.

One of the most widely used measures of scientific impact is the number of citations. However, due to its heavy-tailed distribution, citations are fundamentally difficult to predict but can be improved. This study was aimed at investigating the factors and parts influencing the citation number of a scientific paper in the otology field. Therefore, this work proposes a new solution that utilizes machine learning and natural language processing to process English text and provides a paper citation as the predicted results. Different algorithms are implemented in this solution, such as linear regression, boosted decision tree, decision forest, and neural networks. The application of neural network regression revealed that papers' abstracts have more influence on the citation numbers of otological articles. This new solution has been developed in visual programming using Microsoft Azure machine learning at the back end and Programming Without Coding Technology at the front end. We recommend using machine learning models to improve the abstracts of research articles to get more citations.

Data mining of the healthcare system based on the machine learning model developed in the Microsoft azure machine learning studio

This article presents data mining, which is based on the methods of mathematical statistics and machine learning, describes the features of applying regression analysis methods in the machine learning systems. The developed machine learning model includes the regression analysis modules based on the Bayesian linear, artificial neural network, decision tree, decision forest, and linear regressions. In the process of applying this machine learning model, using the mentioned algorithms, the corresponding regression models were constructed and their comparative analysis was performed, the results were analyzed. The results obtained indicate the feasibility of using data mining in the medical research using machine learning systems. The presented methods can serve as a basis for strategic development of a new directions of the medical data processing and decision-making in this field. We have identified the prospects for further research aimed at applying data mining methods to the healthcare system, namely, clustering, classification, anomaly detection.

Developing a Predictive Health Care System for Diabetes Diagnosis as a Machine Learning-Based Web Service

Background:&#x0D; Diabetes is one of the dangerous and silent illnesses that cause sudden death. It can occur at any time and may cause great injury to the organs of the body or damage them completely. So, we must investigate this disease at the beginning of its appearance and before it gets hard to treat. With the fast advancement of Machine Learning (ML), these approaches enhanced the efficiency of decision processes in a wide range of applications, including medical diagnostics.&#x0D; Materials and Methods:&#x0D; In this paper, we chose a medical application field and used supervised machine learning algorithms to construct a high-accuracy prediction model for diabetes in humans at an early stage, before it progresses to the point of morbidity or fatality. The suggested model can extract hidden knowledge from diabetes-related data gathered from the Kaggle machine learning repository. We utilize Microsoft Azure ML Studio to model these ML algorithms.&#x0D; Results:&#x0D; This study will benefit the health industry by offering users an online tool (i.e., web service) that allows them to input data and receive results that predict whether or not the person has diabetes. As a result, prior knowledge and ongoing monitoring of their diabetic health state will lower the risk of complications, morbidity, and death caused by this illness. After running numerous experiments with the classifier models to evaluate the proposed system, several performance indicators, including Recall, Precision, Accuracy, and f1-score, are measured for comparison. Based on the classification output, it was determined that Decision Forest is a better strategy and produces better results than the other ML approaches.&#x0D; Conclusion:&#x0D; The suggested system's key contribution is to improve healthcare quality, minimize hospitalizations, and lower the high expenditures of healthcare and drugs.

Developing a mental health index using a machine learning approach: Assessing the impact of mobility and lockdown during the COVID-19 pandemic

Governments worldwide have implemented stringent restrictions to curtail the spread of the COVID-19 pandemic. Although beneficial to physical health, these preventive measures could have a profound detrimental effect on the mental health of the population. This study focuses on the impact of lockdowns and mobility restrictions on mental health during the COVID-19 pandemic. We first develop a novel mental health index based on the analysis of data from over three million global tweets using the Microsoft Azure machine learning approach. The computed mental health index scores are then regressed with the lockdown strictness index and Google mobility index using fixed-effects ordinary least squares (OLS) regression. The results reveal that the reduction in workplace mobility, reduction in retail and recreational mobility, and increase in residential mobility (confinement to the residence) have harmed mental health. However, restrictions on mobility to parks, grocery stores, and pharmacy outlets were found to have no significant impact. The proposed mental health index provides a path for theoretical and empirical mental health studies using social media.

Establishment of a prediction tool for ocular trauma patients with machine learning algorithm.

To predict final visual acuity and analyze significant factors influencing open globe injury prognosis. Prediction models were built using a supervised classification algorithm from Microsoft Azure Machine Learning Studio. The best algorithm was selected to analyze the predicted final visual acuity. We retrospectively reviewed the data of 171 patients with open globe injury who visited the Pusan National University Hospital between January 2010 and July 2020. We then applied cross-validation, the permutation feature importance method, and the synthetic minority over-sampling technique to enhance tool performance. The two-class boosted decision tree model showed the best predictive performance. The accuracy, precision, recall, F1 score, and area under the receiver operating characteristic curve were 0.925, 0.962, 0.833, 0.893, and 0.971, respectively. To increase the efficiency and efficacy of the prognostic tool, the top 14 features were finally selected using the permutation feature importance method: (listed in the order of importance) retinal detachment, location of laceration, initial visual acuity, iris damage, surgeon, past history, size of the scleral laceration, vitreous hemorrhage, trauma characteristics, age, corneal injury, primary diagnosis, wound location, and lid laceration. Here we devise a highly accurate model to predict the final visual acuity of patients with open globe injury. This tool is useful and easily accessible to doctors and patients, reducing the socioeconomic burden. With further multicenter verification using larger datasets and external validation, we expect this model to become useful worldwide.

Do negative events really have deteriorating effects on stock performance? A comparative study on Tesla (US) and Nio (China)

PurposeBoth investors and the stock markets are believed to behave in a perfectly rational manner, where investors focus on utility maximization and are not subjected to cognitive biases or any information processing errors. However, it has been discovered that the sentiment of the social mood has a significant impact on the stock market. This study aims to analyze how did the protest event of Tesla happened in April 2021 have a significant effect on the company's stock performance as well as its competitors, Nio, under the competitive effect.Design/methodology/approachThe research is based on time series data collected from Tesla and Nio by employing 10 days, 15 days and 20 days anticipation and adjustment period for the event study. This study employed a text sentiment analysis to identify the polarity of the sentiment of the protest event using the Microsoft Azure machine learning tool which utilizes MPQA subjective lexicon.FindingsThe findings provide further evidence to show that a company-specific negative event has deteriorating effects on its stock performance, while having an opposite effect on its competitors.Research limitations/implicationsThe paper argues that negative sentiments through social media word of mouth (SWOM) affect the stock market not just in the short run but potentially in the longer run. Such negative sentiments might create a snowball effect which causes the market to further scrutinize a company's operations and possibly lose confidence in the company.Originality/valueThis study explores how the Tesla's protest event at Shanghai Auto Show 2021 has a significant impact on Tesla's stock performance and prolonged negative impact although Tesla implemented immediate remedial actions. The remedial actions were not accepted positively and induced a wave of negative news which had a more persistent effect.