Introduction: Dementia is a condition (a collection of related signs and symptoms) that causes a continuing deterioration in cognitive function, and millions of people are impacted by dementia every year as the world population continues to rise. Conventional approaches for determining dementia rely primarily on clinical examinations, analyzing medical records, and administering cognitive and neuropsychological testing. However, these methods are time-consuming and costly in terms of treatment. Therefore, this study aims to present a noninvasive method for the early prediction of dementia so that preventive steps should be taken to avoid dementia.

Methods: We developed a hybrid diagnostic system based on statistical and machine learning (ML) methods that used patient electronic health records to predict dementia. The dataset used for this study was obtained from the Swedish National Study on Aging and Care (SNAC), with a sample size of 43040 and 75 features. The newly constructed diagnostic extracts a subset of useful features from the dataset through a statistical method (F-score). For the classification, we developed an ensemble voting classifier based on five different ML models: decision tree (DT), naive Bayes (NB), logistic regression (LR), support vector machines (SVM), and random forest (RF). To address the problem of ML model overfitting, we used a cross-validation approach to evaluate the performance of the proposed diagnostic system. Various assessment measures, such as accuracy, sensitivity, specificity, receiver operating characteristic (ROC) curve, and Matthew’s correlation coefficient (MCC), were used to thoroughly validate the devised diagnostic system’s efficiency.

Results: According to the experimental results, the proposed diagnostic method achieved the best accuracy of 98.25%, as well as sensitivity of 97.44%, specificity of 95.744%, and MCC of 0.7535.

Discussion: The effectiveness of the proposed diagnostic approach is compared to various cutting-edge feature selection techniques and baseline ML models. From experimental results, it is evident that the proposed diagnostic system outperformed the prior feature selection strategies and baseline ML models regarding accuracy.

 Doctoral supervision is a challenging process, and supervisors use multiple supervision styles to ensure the success of their doctoral students in their research and education. To provide valuable insights into the effectiveness of these supervision practices and help identify areas for improvement, it is important to understand how doctoral students perceive different aspects of supervision. This study aims to explore the perspectives of doctoral students regarding the supervision practices of their supervisors at the Blekinge Institute of Technology (BTH). The study employs a deductive approach with a cross-sectional study design. A survey questionnaire was sent to the doctoral students at BTH. The survey included questions on the five facets of supervision based on the theoretical framework by Halse and Malfroy [1]: learning alliance, habits of mind, scholarly expertise, technè, and contextual expertise. The data collected was quantitative, and descriptive analysis was performed. The total number of respondents (N) was 51 (53.12%) out of 96 invited participants. Results indicated that while most students reported effective learning alliance, significant challenges were noted in fostering teamwork, providing constructive feedback, and offering technical guidance. Specific departments, such as the Department of Computer Science (DIDA) and the Department of Health (TIHA), reported more substantial challenges across multiple facets of supervision. In conclusion, while the overall perception of supervision at BTH is positive, there is a clear need for targeted interventions to address the identified weaknesses and enhance the quality of doctoral education. This information can help strengthen the quality of the doctoral programs at BTH.

Background: The most common degenerative condition in older adults is dementia, which can be predicted using a number of indicators and whose progression can be slowed down. One of the indicators of an increased risk of dementia is sleep disturbances. This study aims to examine if machine learning can predict dementia and which sleep disturbance factors impact dementia.

Methods: This study uses five machine learning algorithms (gradient boosting, logistic regression, gaussian naive Bayes, random forest and support vector machine) and data on the older population (60+) in Sweden from the Swedish National Study on Ageing and Care — Blekinge (). Each algorithm uses 10-fold stratified cross-validation to obtain the results, which consist of the Brier score for checking accuracy and the feature importance for examining the factors which impact dementia. The algorithms use 16 features which are on personal and sleep disturbance factors.

Results: Logistic regression found an association between dementia and sleep disturbances. However, it is slight for the features in the study. Gradient boosting was the most accurate algorithm with 92.9% accuracy, 0.926 f1-score, 0.974 ROC AUC and 0.056 Brier score. The significant factors were different in each machine learning algorithm. If the person sleeps more than two hours during the day, their sex, education level, age, waking up during the night and if the person snores are the variables that most consistently have the highest feature importance in all algorithms.

Conclusion: There is an association between sleep disturbances and dementia, which machine learning algorithms can predict. Furthermore, the risk factors for dementia are different across the algorithms, but sleep disturbances can predict dementia.

Context

The Google Play Store is widely recognized as one of the largest platforms for downloading applications, both free and paid1. On a daily basis, millions of users avail themselves of this marketplace, sharing their thoughts through various means such as star ratings, user comments, suggestions, and feedback. These insights, in the form of comments and feedback, constitute a valuable resource for organizations, competitors, and emerging companies seeking to expand their market presence. These comments provide insights into app deficiencies, suggestions for new features, identified issues, and potential enhancements. Unlocking the potential of this repository of suggestions holds significant value.

Objective

This study sought to gather and analyze user reviews from the Google Play store for leading game apps. The primary aim was to construct a dataset for subsequent analysis utilizing requirements engineering, machine learning, and competitive assessment.

Methodology

The authors employed a Python-based web scraping method to extract a comprehensive set of over 429,000+ reviews from the Google Play pages of selected apps. The scraped data encompassed reviewer names (removed due to privacy), ratings, and the textual content of the reviews.

Results

The outcome was a dataset comprising the extracted user reviews, ratings, and associated metadata. A total of 429,000+ reviews were acquired through the scraping process for popular apps like Subway Surfers, Candy Crush Saga, PUBG Mobile, among others. This dataset not only serves as a valuable educational resource for instructors, aiding in the training of students in data analysis, but also offers practitioners the opportunity for in-depth examination and insights (in the past data of top apps).

District Heating (DH) systems are essential for energy-efficient urban heating. However, despite the advancements in automated fault detection and diagnosis (FDD), DH still faces challenges in operational faults that impact efficiency. This study introduces the Shared Nearest Neighbor Enhanced District Heating Anomaly Detection (SHEDAD) approach, designed to approximate the DH network topology and allow for local anomaly detection without disclosing sensitive information, such as substation locations. The approach leverages a multi-Adaptive k-Nearest Neighbor (k-NN) graph to improve the initial neighborhood creation. Moreover, it introduces a merging technique that reduces noise and eliminates trivial edges. We use the Median Absolute Deviation (MAD) and modified z-scores to flag anomalous substations. The results reveal that SHEDAD outperforms traditional clustering methods, achieving significantly lower intra-cluster variance and distance. Additionally, SHEDAD effectively isolates and identifies two distinct categories of anomalies: supply temperatures and substation performance. We identified 30 anomalous substations and reached a sensitivity of approximately 65% and specificity of approximately 97%. By focusing on this subset of poor-performing substations in the network, SHEDAD enables more targeted and effective maintenance interventions, which can reduce energy usage while optimizing network performance. © 2024 IEEE.

Globally, cancer is the second-leading cause of death after cardiovascular disease. To improve survival rates, risk factors and cancer predictors must be identified early. From the literature, researchers have developed several kinds of machine learning-based diagnostic systems for early cancer prediction. This study presented a diagnostic system that can identify the risk factors linked to the onset of cancer in order to anticipate cancer early. The newly constructed diagnostic system consists of two modules: the first module relies on a statistical F-score method to rank the variables in the dataset, and the second module deploys the random forest (RF) model for classification. Using a genetic algorithm, the hyperparameters of the RF model were optimized for improved accuracy. A dataset including 10 765 samples with 74 variables per sample was gathered from the Swedish National Study on Aging and Care (SNAC). The acquired dataset has a bias issue due to the extreme imbalance between the classes. In order to address this issue and prevent bias in the newly constructed model, we balanced the classes using a random undersampling strategy. The model's components are integrated into a single unit called F-RUS-RF. With a sensitivity of 92.25% and a specificity of 85.14%, the F-RUS-RF model achieved the highest accuracy of 86.15%, utilizing only six highly ranked variables according to the statistical F-score approach. We can lower the incidence of cancer in the aging population by addressing the risk factors for cancer that the F-RUS-RF model found. © 2024 The Author(s). International Journal of Imaging Systems and Technology published by Wiley Periodicals LLC.

Teaching research methodologies to graduate students is important to help them succeed in their degree programs. At Blekinge Institute of Technology (BTH), research methodologies in software engineering and computer science is a critical course, as it provides the students with the skills and knowledge necessary to conduct high-quality research for their master's thesis. More importantly, this course is mandatory in most degree programs at BTH, and the students are required to successfully complete this course as a pre-requisite for their master's thesis.  Over the years, this course has been delivered using traditional teaching methods, such as lectures, assignments, readings, and feedback sessions. During the past years, as the course responsible, I observed that there had been a decline in the success rate for students in this course and a lack of student engagement. We identified that traditional teaching methods may not be the most effective way to engage students in this course. To this end, I introduced the concept of flipped classrooms and active learning classrooms (ALCs). During the ALCs, the students were asked to complete different tasks related to the concepts acquired through pre-recorded lectures and mandatory reading assignments. These tasks are done in student groups, and teachers promote open discussions between the students by problematizing the concepts. 

During exploratory testing sessions the tester simultaneously learns, designs and executes tests. The activity is iterative and utilizes the skills of the tester and provides flexibility and creativity. Test charters are used as a vehicle to support the testers during the testing. The aim of this study is to support practitioners in the design of test charters through checklists. We aimed to identify factors allowing practitioners to critically reflect on their designs and contents of test charters to support practitioners in making informed decisions of what to include in test charters. The factors and contents have been elicited through interviews. Overall, 30 factors and 35 content elements have been elicited.

Context: Exploratory testing (ET) is an approach to test software with a strong focus on personal skills and freedom of the tester. ET emphasises the simultaneous design and execution of tests with minimal test documentation. Test practitioners often claim that their choice to use ET as an important alternative to scripted testing is based on several benefits ET exhibits over the scripted testing. However, these claims lack empirical evidence as there is little research done in this area. Moreover, ET is usually considered an ad-hoc way of doing testing as everyone does it differently. There have been some attempts in past to provide structure to ET. Session based test management (SBTM) is an approach that attempts to provide some structure to ET and gives some basic guidelines to structuring the test sessions. However, these guidelines are still very abstract and are very open to individuals' interpretation.

Objective: The main objective of this doctoral thesis is to support practitioners in their decisions about choosing exploratory versus scripted testing. Furthermore, it is also aimed to investigate the empirical evidence in support of ET and find ways to structure ET and classify different levels of exploration that drive the choices made by exploratory testers. Another objective of this thesis is to provide a decision support system to select levels of exploration in overall test process.

Method: The findings presented in this thesis are obtained through a controlled experiment with participants from industry and academia, exploratory surveys, interviews and focus groups conducted at different companies including Ericsson AB, Sony Mobile Communications, Axis Communications AB and Softhouse Consulting Baltic AB.

Results: Using the exploratory survey, we found three test techniques to be most relevant in context of testing software systems and in particular heterogeneous systems. The most frequently used technique mentioned by the practitioners is ET which is not a much researched topic. We also found many interesting claims about ET in grey literature produced by practitioners in the form of informal presentations and blogs but these claims lacked any empirical evidence. Therefore, a controlled experiment was conducted with students and industry practitioners to compare ET with scripted testing. The experiment results show that ET detects significantly more critical defects compared to scripted testing and is more time efficient. However, ET has its own limitations and there is not a single way to use it for testing. In order to provide structure to ET, we conducted a study where we propose checklists to support test charter design in ET. Furthermore, two more industrial focus group studies at four companies were conducted that resulted in a taxonomy of exploration levels in ET and a decision support method for selecting exploration levels in ET. Lastly, we investigated different problems that researchers face when conducting surveys in software engineering and have presented mitigation strategies for these problems.

Conclusion: The taxonomy for levels of exploration in ET, proposed in this thesis, provided test practitioners at the companies a better understanding of the underlying concepts of ET and a way to structure their test charters. A number of influence factors elicited as part of this thesis also help them prioritise which level of exploration suits more to their testing in the context of their products. Furthermore, the decision support method provided the practitioners to reconsider their current test focus to test their products in a more effective way.

The Research Methodology in Software Engineering and Computer Science (RM) is a compulsory course that must be studied by graduate students at Blekinge Institute of Technology (BTH) prior to undertaking their theses work. The course is focused on teaching research methods and techniques for data collection and analysis in the fields of Computer Science and Software Engineering. It is intended that the course should help students in practically applying appropriate research methods in different courses (in addition to the RM course) including their Master’s theses. However, it is believed that there exist deficiencies in the course due to which the course implementation (learning and assessment activities) as well as the performance of different participants (students, teachers, and evaluators) are affected negatively. In this article our aim is to investigate potential deficiencies in the RM course at BTH in order to provide a concrete evidence on the deficiencies faced by students, evaluators, and teachers in the course. Additionally, we suggest recommendations for resolving the identified deficiencies. Our findings gathered through semi-structured interviews with students, teachers, and evaluators in the course are presented in this article. By identifying a total of twenty-one deficiencies from different perspectives, we found that there exist critical deficiencies at different levels within the course. Furthermore, in order to overcome the identified deficiencies, we suggest seven recommendations that may be implemented at different levels within the course and the study program. Our suggested recommendations, if implemented, will help in resolving deficiencies in the course, which may lead to achieving an improved teaching and learning in the RM course at BTH.