Search

Orphan diseases, exemplified by T-cell prolymphocytic leukemia, present inherent challenges due to limited data availability and complexities in effective care. This study delves into harnessing the potential of machine learning to enhance care strategies for orphan diseases, specifically focusing on allogeneic hematopoietic cell transplantation (allo-HCT) in T-cell prolymphocytic leukemia. The investigation evaluates how varying numbers of variables impact model performance, considering the rarity of the disease. Utilizing data from the Center for International Blood and Marrow Transplant Research, the study scrutinizes outcomes following allo-HCT for T-cell prolymphocytic leukemia. Diverse machine learning models were developed to forecast acute graft-versus-host disease (aGvHD) occurrence and its distinct grades post-allo-HCT. Assessment of model performance relied on balanced accuracy, F1 score, and ROC AUC metrics. The findings highlight the Linear Discriminant Analysis (LDA) classifier achieving the highest testing balanced accuracy of 0.58 in predicting aGvHD. However, challenges arose in its performance during multi-class classification tasks. While affirming the potential of machine learning in enhancing care for orphan diseases, the study underscores the impact of limited data and disease rarity on model performance.

Purpose: Connecting autonomous drones to ground operations and services is a prerequisite for the adoption of scalable and sustainable drone services in the built environment. Despite the rapid advance in the field of autonomous drones, the development of ground infrastructure has received less attention. Contemporary airport design offers potential solutions for the infrastructure serving autonomous drone services. To that end, this paper aims to construct a framework for connecting air and ground operations for autonomous drone services. Furthermore, the paper defines the minimum facilities needed to support unmanned aerial vehicles for autonomous logistics and the collection of aerial data. Design/methodology/approach: The paper reviews the state-of-the-art in airport design literature as the basis for analysing the guidelines of manned aviation applicable to the development of ground infrastructure for autonomous drone services. Socio-technical system analysis was used for identifying the service needs of drones. Findings: The key findings are functional modularity based on the principles of airport design applies to micro-airports and modular service functions can be connected efficiently with an autonomous ground handling system in a sustainable manner addressing the concerns on maintenance, reliability and lifecycle. Research limitations/implications: As the study was limited to the airport design literature findings, the evolution of solutions may provide features supporting deviating approaches. The role of autonomy and cloud-based service processes are quintessentially different from the conventional airport design and are likely to impact real-life solutions as the area of future research. Practical implications: The findings of this study provided a framework for establishing the connection between the airside and the landside for the operations of autonomous aerial services. The lack of such framework and ground infrastructure has hindered the

The Internet of Things (IoT) is the ecosystem of networked devices encountered in both work and home. IoT security is a great concern and vulnerabilities are reported daily. IoT is mixed into other digital infrastructure both in terms of sharing the same networks and using the same software components. In this paper, we analyze Common Vulnerabilities and Exposures (CVE) entries, including known exploited vulnerabilities, to describe the vulnerabilities in the IoT context. The results indicate that 88% of reported vulnerabilities are relevant to IoT systems. Half of the vulnerabilities are in the backend or frontend systems while 10‐20% concern the IoT devices. HTTP servers are the vulnerability hotspots wherever they are located. Software components are used in all IoT subsystems and tracking and updating them is essential for system security. The results can be used to understand where and what kind of vulnerabilities are in IoT systems.

Purpose: To gain an understanding of the heterogeneous group of type 2 diabetes (T2D) patients, we aimed to identify patients with the homogenous long-term HbA1c trajectories and to predict the trajectory membership for each patient using explainable machine learning methods and different clinical-, treatment-, and socio-economic-related predictors. Patients and Methods: Electronic health records data covering primary and specialized healthcare on 9631 patients having T2D diagnosis were extracted from the North Karelia region, Finland. Six-year HbA1c trajectories were examined with growth mixture models. Linear discriminant analysis and neural networks were applied to predict the trajectory membership individually. Results: Three HbA1c trajectories were distinguished over six years: “stable, adequate” (86.5%), “improving, but inadequate” (7.3%), and “fluctuating, inadequate” (6.2%) glycemic control. Prior glucose levels, duration of T2D, use of insulin only, use of insulin together with some oral antidiabetic medications, and use of only metformin were the most important predictors for the long-term treatment balance. The prediction model had a balanced accuracy of 85% and a receiving operating characteristic area under the curve of 91%, indicating high performance. Moreover, the results based on SHAP (Shapley additive explanations) values show that it is possible to explain the outcomes of machine learning methods at the population and individual levels. Conclusion: Heterogeneity in long-term glycemic control can be predicted with confidence by utilizing information from previous HbA1c levels, fasting plasma glucose, duration of T2D, and use of antidiabetic medications. In future, the expected development of HbA1c could be predicted based on the patient’s unique risk factors offering a practical tool for clinicians to support treatment planning.

With respect to physical fitness, activity and wellbeing, digital games are often considered as “screen-time”, which reduces healthy physical activities, often causing the player to sit down and eat unhealthily while playing. However, when recognising and identifying the persuasive properties in games, their “magic circle” can be used to activate and move the players instead of passivating them. Many Augmented Reality Games (ARGs), such as Ingress Prime and Pokémon GO, can increase the physical activity of the players. This multidisciplinary dissertation examines, identifies and categorises persuasive properties of several ARGs, studied in the arctic context. The findings of this study are based on qualitative material, such as participant observation, interviews, discussions and comments of the players as well as quantitative data. This study is an applied digital ethnography, comprising of both offline and online interactions, material and methods. The qualitative material was analysed applying content analysis and the statistical material was analysed using statistical software analysis. The findings were reflected in the light of several cultural and motivational theories, such as the flow and self-determination theories, as well as theories considering digital persuasion. Digital, as well as traditional, games are inherently persuasive; they lure the player to continue playing. Based on this study, the “digital magic circle” comprises multiple cultural and psychological factors, as well as several factors related specifically to digital media. Essentially, games offer the players purposeful activity with clear and achievable goals: the more successful the game design, the stronger the persuasive power. Playing such games may cause the optimal experience (flow) in player, often resulting in the player to play even more. Such persuasiveness could be implemented in “moving games” like ARGs, increasing the player’s physical activity, affecting health and w, Tiivistelmä Fyysisen kunnon, aktiivisuuden ja hyvinvoinnin kannalta digitaalisten pelien pelaaminen käsitetään usein ”ruutuajaksi”, joka istuttaa pelaajan sohvalle, ja usein myös syömään epäterveellisesti pelatessaan. Samalla pelaaminen vähentää terveyden kannalta tärkeän fyysisen aktiivisuuden määrää. Pelien suostuttelevien ominaisuuksien tiedostamisella ja tunnistamisella niiden ”taikapiiriä” voidaan kuitenkin hyödyntää pelaajien aktivoimisessa ja liikuttamisessa passivoinnin sijaan. Monien Lisätyn Todellisuuden Pelien (ARGs, Augmented Reality Games), kuten Ingress Prime ja Pokémon GO, ajatellaan lisäävän pelaajien fyysistä aktiivisuutta. Tässä monitieteisessä väitöstyössä tarkastellaan, tunnistetaan ja luokitellaan lisätyn todellisuuden pelien suostuttelevia ominaisuuksia arktisessa viitekehyksessä. Tutkimuksen tulokset perustuvat osallistuvaan havainnointiin, pelaajien haastatteluihin, keskusteluihin ja kommentteihin sekä tilastolliseen aineistoon. Kyseessä on soveltava etnografinen tutkimus, joka rakentuu niin verkossa kuin sen ulkopuolella tapahtuvaan toimintaan, tutkimusaineistoon ja menetelmiin. Laadullinen aineisto analysoitiin sisällönanalyysin menetelmin, ja määrällinen aineisto tilastoanalyysin menetelmin. Tuloksia reflektoitiin erinäisiin kulttuuri- ja motivaatioteorioihin, kuten Flow- ja itseohjautuvuus teorioihin, sekä digitaalista suostuttelevuutta koskeviin teorioihin. Digitaaliset, siinä missä perinteisetkin pelit, ovat itsessään suostuttelevia; ne houkuttelevat pelaajaa jatkamaan pelaamista. Tämän tutkimuksen perusteella ”Digitaalinen taikapiiri” muodostuu erinäisistä kulttuurisista ja psykologisista tekijöistä, sekä digitaalisiin medioihin liittyvistä erityispiirteistä. Pelit tarjoavat pelaajilleen tarkoituksenmukaista toimintaa, johon kuuluvat selkeät ja saavutettavissa olevat päämäärät. Mitä paremmin peli on suunniteltu, sitä tehokkaammin se myös suostuttelee. Tällaisten pelien pelaaminen voi aiheuttaa pelaajassa optimaalisen kokemuksen (Flow)

Cell-secreted extracellular vesicles (EVs), carrying components such as RNA, DNA, proteins, and metabolites, serve as candidates for developing non-invasive solutions for monitoring health and disease, owing to their capacity to cross various biological barriers and to become integrated into human sweat. However, the evidence for sweat-associated EVs providing clinically relevant information to use in disease diagnostics has not been reported. Developing cost-effective, easy, and reliable methodologies to investigate EVs’ molecular load and composition in the sweat may help to validate their relevance in clinical diagnosis. We used clinical-grade dressing patches, with the aim being to accumulate, purify and characterize sweat EVs from healthy participants exposed to transient heat. The skin patch-based protocol described in this paper enables the enrichment of sweat EVs that express EV markers, such as CD63. A targeted metabolomics study of the sweat EVs identified 24 components. These are associated with amino acids, glutamate, glutathione, fatty acids, TCA, and glycolysis pathways. Furthermore, as a proof-of-concept, when comparing the metabolites’ levels in sweat EVs isolated from healthy individuals with those of participants with Type 2 diabetes following heat exposure, our findings revealed that the metabolic patterns of sweat EVs may be linked with metabolic changes. Moreover, the concentration of these metabolites may reflect correlations with blood glucose and BMI. Together our data revealed that sweat EVs can be purified using routinely used clinical patches, setting the foundations for larger-scale clinical cohort work. Furthermore, the metabolites identified in sweat EVs also offer a realistic means to identify relevant disease biomarkers. This study thus provides a proof-of-concept towards a novel methodology that will focus on the use of the sweat EVs and their metabolites as a non-invasive approach, in order to monitor wellbeing and changes

Artificial intelligence (AI) is increasingly being used to improve patient care and management. In this paper, we propose explainable AI (XAI) models for predicting severe hypoglycemia (SH) and diabetic ketoacidosis (DKA) episodes in adults with type 1 diabetes (T1D) and relapses in adults with relapsing-remitting multiple sclerosis (RRMS). We follow a three-step process in this study: (1) develop baseline machine learning (ML) models, (2) improve the models using ReliefF feature selection technique, and develop sex-stratified models, (3) explain the models and their results using SHapley Additive exPlanations (SHAP). We built six ML models (XGBoost, LightGBM, CatBoost, AdaBoost, random forest, and linear regression) for all scenarios. Applying the ReliefF feature selection led to improved model performance in predicting all outcomes compared to the baseline models. Additionally, sex-stratified models further improved the prediction of SH episodes and relapses. The F1 scores for predicting SH episodes in male and female patients were 84.07% and 84.95%, respectively, and the DKA prediction model achieved an F1 score of 78.67%. The proposed relapse prediction models outperformed existing models with F1 scores of 84.55% (males) and 76.11% (females), and ROCs of 70.26% (males) and 69.05% (females). Our results highlight the importance of considering sex differences, socioeconomic factors, and physical and mental health in medical outcome prediction. Boosting ML algorithms were found to be effective in detecting SH and DKA in T1D patients and relapses in RRMS patients compared to conventional tree-based ML and statistical models.

Clinical data analysis could lead to breakthroughs. However, clinical data contain sensitive information about participants that could be utilized for unethical activities, such as blackmailing, identity theft, mass surveillance, or social engineering. Data anonymization is a standard step during data collection, before sharing, to overcome the risk of disclosure. However, conventional data anonymization techniques are not foolproof and also hinder the opportunity for personalized evaluations. Much research has been done for synthetic data generation using generative adversarial networks and many other machine learning methods; however, these methods are either not free to use or are limited in capacity. This study evaluates the performance of an emerging tool named synthpop, an R package producing synthetic data as an alternative approach for data anonymization. This paper establishes data standards derived from the original data set based on the utilities and quality of information and measures variations in the synthetic data set to evaluate the performance of the data synthesis process. The methods to assess the utility of the synthetic data set can be broadly divided into two approaches: general utility and specific utility. General utility assesses whether synthetic data have overall similarities in the statistical properties and multivariate relationships with the original data set. Simultaneously, the specific utility assesses the similarity of a fitted model’s performance on the synthetic data to its performance on the original data. The quality of information is assessed by comparing variations in entropy bits and mutual information to response variables within the original and synthetic data sets. The study reveals that synthetic data succeeded at all utility tests with a statistically non-significant difference and not only preserved the utilities but also preserved the complexity of the original data set according to the data standard establish

It is well-known that security issues in medical devices, services and applications have potentially catastrophic consequences. To avoid compromising patient data or information systems, it is essential that healthcare services and products meet the relevant information security and data protection requirements. For these reasons, the Digi-HTA assessment includes information security and data protection assessment domains. The outcome of the Digi-HTA process is a recommendation that decision-makers can use during the procurement process. We present results and experiences from the first assessments made in the Digi-HTA process. We have assessed six products so far and multiple assessments are in progress. The results indicate that healthcare product manufacturers have found the process useful, and usually, the manufacturers have had to improve the security of their product during the Digi-HTA process to get a favourable recommendation for their product. The assessment processes have taken longer than expected due to shortcomings and ambiguities in the provided self-assessment forms, and due to feedback cycles and meetings prompted by assessment findings. Of the six assessed products, four received a green light in information security and data protection, whereas two have received a yellow light due to issues that were not fixed during the process. In addition to shortcomings in adhering to best practices, we have also found exploitable security issues.

Industry is increasingly moving towards data-driven business. With an abundance of data, the problem is no longer how to get access to data but how to extract the most value from it. The extracted knowledge helps to control industrial processes efficiently and automatically. The quality of products can be improved by finding root causes behind poor quality, hence improving the yield and competitiveness of the whole plant. This thesis helps to understand the benefit of using statistical machine learning methods to improve manufacturing processes. With the methods presented, the future can be predicted based on historical data and data-driven decision support for the processes can be offered. The application area of this work is the steel industry. This work gives step-by-step advice for successfully implementing AI applications in the industry. In addition, methods for finding root causes behind poor quality are presented to improve the process and the quality of the products. Since data collected under challenging measurement conditions in the industry is never flawless, a quality model which can utilize incomplete data is also developed. Machine learning methods are used in this work to process data and to develop data-driven quality models, which help to predict the desired quality characteristics and hence support the decision-making of the workers managing the process. Both supervised and semi-supervised machine learning methods are used. In addition, explainable machine learning is used to increase the transparency of the models. With the help of these methods, predictive quality models are developed for four different datasets consisting of measurements from steel manufacturing processes. The developed models have been implemented as a smart manufacturing tool, which enables real-time support for process workers during the manufacturing process. Using the tool, a worker can recognize the factors that can cause quality problems in the process in advanc, Tiivistelmä Teollisuus on siirtynyt yhä enenemissä määrin kohti dataohjautuvaa liiketoimintaa. Dataa on tarjolla niin paljon, että ongelmana ei ole sen saatavuus vaan se mitä siitä saadaan irti. Datasta louhittu tieto auttaa kontrolloimaan teollisuusprosesseja tehokkaammiksi ja automaattisemmiksi sekä parantamaan tuotteen laatua löytämällä syyt huonon laadun takaa ja näin parantamaan koko tehtaan tuottoa ja kilpailukykyä. Tämä väitöskirja auttaa ymmärtämään tilastollisten koneoppimismenetelmien tärkeyden teollisuusprosesseja kehitettäessä. Näiden menetelmien avulla pystymme ennakoimaan tulevaisuutta historian avulla ja kasvattamaan dataohjautuvan päätöksenteon tukea prosesseissa. Tämän väitöskirjan sovellusalueena on terästeollisuus. Työ antaa työkaluja siihen miten AI sovellus voidaan toteuttaa onnistuneesti teollisuudessa askel askeleelta. Lisäksi työssä esitellään menetelmiä joiden avulla voidaan löytää juurisyitä huonolle laadulle ja näin kehittää prosessia tuottamaan laadukkaampia tuotteita. Koska teollisuuden haastavista olosuhteista kerätty data ei koskaan ole täydellistä, on tässä työssä kehitetty myös laatumalli, joka pystyy hyödyntämään puuttuvia tietoja. Tässä työssä käytetään koneoppimismenetelmiä, joiden tarkoituksena on prosessoida dataa ja kehittää datamukautuvia laatumalleja, joiden avulla voidaan tehdä ennustuksia haluttavasta laatuominaisuudesta, ja näin tukea prosessissa työskentelevien ihmisten päätöksentekoa. Työssä on käytetty sekä ohjattuja-, että osittain-ohjattuja koneoppimismenetelmiä sekä mallien läpinäkyyvyyttä lisääviä selittäviä koneoppimismenetelmiä. Näiden menetelmien avulla on kehitetty ennustavia laatumalleja neljän erilaisen teräsprosessia kuvaavaan datasetin avulla. Kehitetyt laatumallit on implementoitu teollisuutta varten kehitettyyn älykkään päätöksenteon työkaluun, joka mahdollistaa reaaliaikaisen tuen teollisuusprosessissa työskenteleville henkilöille. Työkalun avulla työntekijä voi jo etukäteen tunnistaa tekijät, jotka voiva

In this study, chaos game representation (CGR) is introduced for investigating the pattern of genome sequences. It is an image representation of the genome for the overall visualization of the sequence. The CGR representation is a mapping technique that assigns each sequence base into the respective position in the two-dimension plane to portray the DNA sequence. Importantly, CGR provides one to one mapping to nucleotides as well as sequence. A coordinate of the CGR plane can tell the corresponding base and its location in the original genome. Therefore, the whole nucleotide sequence (until the current nucleotide) can be restored from the one point of the CGR. In this study, CGR coupled with artificial neural network (ANN) is introduced as a new way to represent the genome and to classify intra-coronavirus sequences. A hierarchy clustering study is done to validate the approach and found to be more than 90% accurate while comparing the result with the phylogenetic tree of the corresponding genomes. Interestingly, the method makes the genome sequence significantly shorter (more than 99% compressed) saving the data space while preserving the genome features.

The cybersecurity of the Internet of Things (IoT) is an increasing concern and product vendors are advised to follow security standards, best practices, and guidelines. From the many requirement sources, a vendor is likely to choose only a few. How does this selection impact the security requirements of an IoT product? To answer the question, we collect requirements from 16 sources and divide them into categories for comparison. Common categories are identified, with all sources covering Security design, Interface security, Authentication, Data protection, and System updates. The agreement on the high-level categories does not hold in the subcategories and the selection of the sources have a big impact to the requirement details. Consolidation of the IoT security requirements would be desirable and possible.

Sensor devices are becoming omnipresent, supplying data to a wide range of applications. In the building sector, sensors along with other information sources provide the basis for smart building functionalities. Predicting energy loads and inferring occupancy status of spaces are important tasks that promote energy efficiency and user comfort in buildings. For them, as for many other smart building applications, machine learning modelling utilizing sensor data is commonly applied. This article builds understanding of the environment where this kind of machine learning models have to operate by bringing up properties and quality aspects of the public building data provided by indoor sensor devices. This is done by performing a thorough case study on two real life data sets from university campus buildings located in different climates and applying very different sensor network settings. Outcomes include information about heterogeneity, correlations and temporal patterns present in sensor data, and show the need of the building field for better acknowledging the quality deficiencies that sensor data have. Our results aid in assessing and improving the quality of sensor-based indoor data utilized in machine learning modelling, in evaluating whether a data set is representative enough to build a model that is robust under changing conditions in the building, and in choosing an appropriate number of sensors per space when building an indoor wireless sensor network.

Unmanned aerial vehicles (UAV) can help in improving safety and decreasing costs in the inspection of industrial plants such as oil and gas pipings. UAV-based inspection brings endurance requirements and safety constraints to the inspection system as a whole. This article focuses on the ground support segment of the HYFLIERS inspection systems and more particularly it looks at effectively and efficiently supporting the inspection data flow and at competent and safe power supply.

Safety risks and high costs of human inspection of oil and gas plants drive towards the adoption of robotic inspection. The challenging cluttered inspection environment and the constraints dictated by legislation on potentially explosive atmospheres implying energy-efficient solutions suggest the use of an inspection-tool-equipped hybrid rolling-flying unmanned vehicle and of a mobile ground platform supporting the connected inspection robot. These two design choices together with their development are described in this article.

This article demonstrates the use of data mining methods for evidence-based smart decision support in quality control. The data were collected in a measurement campaign which provided a new and potential quality measurement approach for manufacturing process planning and control. In this study, the machine learning prediction models and Explainable AI methods (XAI) serve as a base for the decision support system for smart manufacturing. The discovered information about the root causes behind the predicted failure can be used to improve the quality, and it also enables the definition of suitable security boundaries for better settings of the production parameters. The user’s need defines the given type of information. The developed method is applied to the monitoring of the surface roughness of the stainless steel strip, but the framework is not application dependent. The modeling analysis reveals that the parameters of the annealing and pickling line (RAP) have the best potential for real-time roughness improvement.

This paper presents Kinetic Energy Difference (KED) as a metric for collision proximity. The calculation of KED for differentially driven robots is explained, along with an example obstacle avoidance algorithm that utilizes it. This example algorithm is computationally efficient and simulations show that it is capable of guiding robots with slow dynamics through narrow corridors.

Lightweight, affordable spherical cameras can be utilized in mobile robotics to build a 3D map of the working environment of a robot. This paper demonstrates how to use the optical flow between two spherical images to quickly construct a semantically meaningful 3D model spanning all directions. The main contribution of the paper is the use of semantic segmentation to increase the robustness of the reconstruction method in both indoor and outdoor applications.

We examined the popularity of open-source tools used for information security analysis (infosec tools). This information would be useful, e.g. in security research, but it was not available. In our study, we created first a corpus of 423 tools from various sources. Then we collected source popularity metrics by Google search, tweets, GitHub stars, SecTools.org ranking, and cross-references between tools. We found a strong correlation between the metrics. We created an aggregate popularity metric from Google search, GitHub stars, and tool cross-reference source metrics using principal component analysis. The aggregate metric explains 70% of the variance in the source metrics. The three most popular tools are Metasploit, Nmap, and Wireshark. We estimated the impact of source metric errors and concluded that the aggregate metric gives an estimate of tool popularity, rather than an exact popularity rank. Furthermore, we divide the tools into overlapping categories by tool scope and type of activity. In the top 100, 51 tools are in the network scope, 27 in the host scope, 15 in the storage scope, 13 in the passwords scope, and 4 in the other tools scope.

This study introduces an ensemble-based personalized human activity recognition method relying on incremental learning, which is a method for continuous learning, that can not only learn from streaming data but also adapt to different contexts and changes in context. This adaptation is based on a novel weighting approach which gives bigger weight to those base models of the ensemble which are the most suitable to the current context. In this article, contexts are different body positions for inertial sensors. The experiments are performed in two scenarios: (S1) adapting model to a known context, and (S2) adapting model to a previously unknown context. In both scenarios, the models had to also adapt to the data of previously unknown person, as the initial user-independent dataset did not include any data from the studied user. In the experiments, the proposed ensemble-based approach is compared to non-weighted personalization method relying on ensemble-based classifier and to static user-independent model. Both ensemble models are experimented using three different base classifiers (linear discriminant analysis, quadratic discriminant analysis, and classification and regression tree). The results show that the proposed ensemble method performs much better than non-weighted ensemble model for personalization in both scenarios no matter which base classifier is used. Moreover, the proposed method outperforms user-independent models. In scenario 1, the error rate of balanced accuracy using user-independent model was 13.3%, using non-weighted personalization method 13.8%, and using the proposed method 6.4%. The difference is even bigger in scenario 2, where the error rate using user-independent model is 36.6%, using non-weighted personalization method 36.9%, and using the proposed method 14.1%. In addition, F1 scores also show that the proposed method performs much better in both scenarios that the rival methods. Moreover, as a side result, it was noted that the

The generation of glass wool waste and mine tailings has raised increasing concerns. This paper deals with the reuse of glass wool waste and lithium mine tailings from spodumene ore (quartz feldspar sand; QFS) in the development of building ceramic materials. The effect of glass wool particle size and sintering temperatures (750, 850 and 950 °C) were investigated. Phase composition and sintering reactions were studied using several techniques including X-ray diffraction with Rietveld refinement, differential scanning calorimetry, scanning electron microscopy, density, water absorption and mechanical tests. The results showed that glass wool acted as fluxing agent, with melting reactions observed from about 700 °C. Grinding glass wool improved its reactivity, enhancing densification and strength development at lower temperatures. The properties of the prepared building ceramics satisfied the requirement of building materials according to ASTM C62, achieving high performance values of 90 MPa and 25 MPa for compressive and flexural strength respectively. These results are of interest for the reuse of glass wool waste, QFS and similar waste streams in building ceramics.

Inter Beat intervals (IBI) in heart rate variability (HRV) are balanced by the actions of the sympathetic (SNS), parasympathetic neural systems (PSNS), and central cardiovascular plasticity. The neural systems are easy to reach by Photoplethymographic (PPG) means. PPG is also much used for fitness testing purposes. For the PPG pulse series analysis we used merged Poincaré plots parameters which have been much extended to evaluate pulse rate variability (PRV). The preliminary results demonstrated that the combination of merged Poincaré plots could show promise as a method for distinguishing between healthy individuals and different vascular disease groups. In addition, the hypothesis of this paper is that PPG pulse wave signals measured before and after the sauna exposure can lead to improve the understanding and estimation of the human body’s healthy status.

In a scenario where fire accidents takes place the priority is always human safety and acting swiftly to contain the fire from further spreading. The modern autonomous systems can promise both human safety and can perform actions rapidly. One such scenario which is motivated by urban firefighting was designed in challenge 3 of MBZIRC 2020 competition. In this challenge the UAV’s and UGV collaborate autonomously to detect the fire and quench the flames with water. So, in this project we have developed Robot Operating System (ROS)-based autonomous system to solve the challenge for UGV criteria by detecting targeted objects in real-time, in our case its simulated fire and red colored softballs. Then finally localize those targets as markers in the map and navigate autonomously to all those targets. This work has two sections, in the first section mapping and localizing the fire and softballs in highly cluttered environment and then reaching those targets autonomously. Robustly mapping the area with adequate sensors and detection of targets with optimally trained CNN based network is the key to localizing of the targeted objects in a highly cluttered environments.

Photoplethysmography (PPG) has proved to be valuable signal carrier information from many biomedical sources. Especially, sauna studies have shown that heart pulse inter beat intervals, heart rate variability (HRV), shape and morphology of PPG can provide much interesting information about arterial elasticity during and after a sauna exposure. These parameters are pulse amplitude, pulse width, and pulse to pulse beat intervals changes caused by the sauna exposure. In our preliminary study there is searched on heart pulse variability information and other hemodynamic parameters. However, there is not yet standard detailed calculation algorithms for these parameters. Merged Poincaré plots, or so called return maps, could give valuable information which is here at the first time noticed.

This paper deals with the synthesis and thermal stability of one-part metakaolin geopolymer composites containing high volume of spodumene tailings (Quartz Feldspar Sand; QFS) and glass wool (GW). One of the objectives of the study was to prepare materials encompassing a maximum amount of waste streams with some potential thermal stability. Several compositions were prepared with sodium metasilicate anhydrous (Na2SiO3) wt.% of 0.5, 2.5, 5, 10 and 12,5. The one-part metakaolin geopolymer composites were cured at 60 °C for 24 h and the mechanical properties were assessed at 7 days and after post-heat treatment at 500, 750, 1000, 1100 or 1200 °C. X-ray diffraction, dilatometry, scanning electron microscopy and thermogravimetry analyses were used to study the stability of the prepared geopolymer composites until 1100–1200 °C. The results showed that more than 20 MPa compressive strength could be achieved with metakaolin geopolymer composites containing only 20 wt% of metakaolin. Metakaolin-GW geopolymer composites were stable up to 500 °C. Meanwhile, their counterparts containing QFS were stable up to 1100–1200 °C; samples prepared with higher dosage of sodium (Na2SiO3 > 5 wt%) retained more than 50% of their initial strength after thermal treatment at 1100 °C. Interestingly, for dosages of Na2SiO3 ≤ 5 wt%, more than 300% increase of strength was observed after thermal treatment at 1100–1200 °C. The use of QFS limited the thermal shrinkage at mild temperatures (<1000 °C), but favoured densification and strength development at 1100–1200 °C.

The main aim of the work is to study the effect of different fibres (steel, glass and basalt) on resistance of blast furnace slag-based alkali-activated mortar in acidic environment. The alkaliactivated slag mortars were exposed to 5% sulfuric and acetic acid solutions for 30 days. Mass change, compressive strength and microstructural changes were evaluated. In plain mortar, it was observed that 70% of the strength was retained in acetic acid environment whereas only 20% of residual strength remains in sulphuric acid environment. FTIR spectroscopy shows the degradation of the matrix, which implies the alkali-activated mortar was more vulnerable in sulphuric acid environment due to its aggressive nature compared to acetic acid. Decalcification and formation of calcium acetate also hinders the further progress of damage in acetic acid attack. Fibres helped in improving the performance of the mortar by holding the matrix together when the degradation occurred in acidic environment. Compared to plain mortar, incorporation of steel fibres exhibited a maximum strength retention of 19% in acetic acid and 7% in sulphuric acid, followed by glass and basalt fibres. SEM images clearly show the debonding of fibres and disintegration of matrix in acidic environment, which resulted in strength loss.

In this article, regression and classification models are compared for stress detection. Both personal and user-independent models are experimented. The article is based on publicly open dataset called AffectiveROAD, which contains data gathered using Empatica E4 sensor and unlike most of the other stress detection datasets, it contains continuous target variables. The used classification model is Random Forest and the regression model is Bagged tree based ensemble. Based on experiments, regression models outperform classification models, when classifying observations as stressed or not-stressed. The best user-independent results are obtained using a combination of blood volume pulse and skin temperature features, and using these the average balanced accuracy was 74.1% with classification model and 82.3% using regression model. In addition, regression models can be used to estimate the level of the stress. Moreover, the results based on models trained using personal data are not encouraging showing that biosignals have a lot of variation not only between the study subjects but also between the session gathered from the same person. On the other hand, it is shown that with subject-wise feature selection for user-independent model, it is possible to improve recognition models more than by using personal training data to build personal models. In fact, it is shown that with subject-wise feature selection, the average detection rate can be improved as much as 4%-units, and it is especially useful to reduce the variance in the recognition rates between the study subjects.

Non-metallic inclusions, especially large or clustered inclusions, in steel are usually harmful. Thus, the microscopic analysis of test specimens is an important part of the quality control. This steel purity analysis produces a large amount of individual inclusion information for each test specimen. The interpretation of the results is laborious and the comparison of larger product groups practically impossible. The purpose of this study was to develop an easy-to-use tool for automatic interpretation of the SEM analysis to differentiate clustered and large inclusions information from the manifold individual inclusion information. Because of the large variety of the potential users, the tool needs to be applicable for any steel grade and application, both for liquid and final product specimen, to analyse automatically steel specimen inclusions, especially inclusion clusters, based on the INCA Feature program produced data from SEM/EDS. The developed tool can be used to improve the controlling of the steel purity or for automatic production of new inclusion cluster features that can be utilised further in quality prediction models, for example.

It is common knowledge that even simple machine learning algorithms can improve in performance with large, good quality data sets. However, limited data sets, be it because of limited size or incomplete instances, are surprisingly common in many real-world modeling problems. In addition to the overall classification accuracy of a model, it is often of interest to know the uncertainty of each individual prediction made by the model. Quantifying this uncertainty of classification models is discussed in this thesis from the perspective of limited data. When some feature values are missing, uncertainty regarding the classification result is increased, but this is not captured in the metrics that quantify uncertainty using traditional methods. To tackle this shortcoming, a method is presented that, in addition to making incomplete data sets usable for any classifier, makes it possible to quantify the uncertainty stemming from missing feature values. In addition, in the case of complete but limited sized data sets, the ability of several commonly used classifiers to produce reliable uncertainty, i.e. probability, estimates, is studied. Two algorithms are presented that can potentially improve probability estimate calibration when data set size is limited. It is shown that the traditional approach to calibration often fails on these limited sized data sets, but using these algorithms still allows improvement in classifier probability estimates with calibration. To support the usefulness of the proposed methods and to answer the proposed research questions, main results from the original publications are presented in this compiling part of the thesis. Implications of the findings are discussed and conclusions drawn., Tiivistelmä Yleisesti tiedetään, että yksinkertaistenkin koneoppimismenetelmien tuloksia saadaan parannettua, jos käytettävissä on paljon hyvälaatuista aineistoa. Vaillinaiset aineistot, joiden puutteet johtuvat aineiston vähäisestä määrästä tai puuttuvista arvoista, ovat kuitenkin varsin yleisiä. Pelkän luokittelutarkkuuden lisäksi mallin yksittäisten ennusteiden epävarmuus on usein hyödyllistä tietoa. Tässä väitöskirjassa tarkastellaan luokittimien epävarmuuden määrittämistä silloin, kun saatavilla oleva aineisto on vaillinainen. Kun aineistosta puuttuu arvoja joistakin piirteistä, luokittelutulosten epävarmuus lisääntyy, mutta tämä lisääntynyt epävarmuus jää huomioimatta perinteisillä puuttuvien arvojen käsittelymenetelmillä. Asian korjaamiseksi tässä väitöskirjassa esitetään menetelmä, jolla puuttuvista arvoista johtuva epävarmuuden lisääntyminen voidaan huomioida. Lisäksi tämä menetelmä mahdollistaa minkä tahansa luokittimen käytön, vaikka luokitin ei muutoin tukisi puuttuvia arvoja sisältävien aineistojen käsittelyä. Tämän lisäksi väitöskirjassa käsitellään useiden yleisesti käytettyjen luokittimien kykyä tuottaa hyviä arvioita ennusteiden luotettavuudesta, eli todennäköisyysarvioita, kun käytettävissä oleva aineisto on pieni. Tässä väitöskirjassa esitetään kaksi algoritmia, joiden avulla voi olla mahdollista parantaa näiden todennäköisyysarvioiden kalibraatiota, vaikka käytettävissä oleva aineisto on pieni. Esitetyistä tuloksista ilmenee, että perinteinen tapa kalibrointiin ei pienillä aineistoilla onnistu, mutta esitettyjen algoritmien avulla kalibrointi tulee mahdolliseksi. Alkuperäisten artikkeleiden tuloksia esitetään tässä kokoomaväitöskirjassa tukemaan esitettyjä väittämiä ja vastaamaan asetettuihin tutkimuskysymyksiin. Lopuksi pohditaan esitettyjen tulosten merkitystä ja vedetään johtopäätökset.

Classifier calibration does not always go hand in hand with the classifier’s ability to separate the classes. There are applications where good classifier calibration, i.e., the ability to produce accurate probability estimates, is more important than class separation. When the amount of data for training is limited, the traditional approach to improve calibration starts to crumble. In this article, we show how generating more data for calibration is able to improve calibration algorithm performance in many cases where a classifier is not naturally producing well-calibrated outputs and the traditional approach fails. The proposed approach adds computational cost but considering that the main use case is with small datasets this extra computational cost stays insignificant and is comparable to other methods in prediction time. From the tested classifiers, the largest improvement was detected with the random forest and naive Bayes classifiers. Therefore, the proposed approach can be recommended at least for those classifiers when the amount of data available for training is limited and good calibration is essential.

The recycling of industrial side streams is of interest for a sustainable use of resources and from an environmental perspective. This paper deals with the reuse of copper slag and spodumene tailings in the development of ceramic materials for potential application in construction. Copper slag included fayalite and magnetite as crystalline phases while spodumene tailings (quartz-felspar sand, QFS) mainly consisted of albite, quartz and microcline. Kaolin (10 wt%) was added as green strength increasing agent in some compositions and the prepared formulations were sintered between 950 and 1150 °C at 50 °C intervals. The phase composition was studied by scanning electron microscopy and X-ray diffraction using Rietveld refinement. Tests such as compressive and flexural strength, water absorption, apparent density and freeze thaw cycles were used to assess the material performances. The results showed that densification and strength development were mainly due to partial melting of QFS particles at 1050–1100 °C, leading to water absorption in the range 0.5–7 %. At 1100 °C, a high-performance compressive strength of 140 MPa was achieved with the formulation made fully of recycled materials (50 wt% copper slag and 50 wt% QFS) suggesting its suitability for high strength demanding materials such as load bearing bricks; values of ultrasonic pulse velocity and compressive strength of these ceramics remained stable after 200 freeze thaw cycles, demonstrating their potential suitability as construction materials in severe weathering environments. Additionally, the leaching test proved a good encapsulation of heavy metals in these ceramics. These results are of interest for waste management and efficient use of resources.

This article presents a statistical prediction model-based intelligent decision support tool for center line deviation monitoring. Data mining methods enable the data driven manufacturing. They also help to understand the manufacturing process and to test different hypotheses. In this study, the original assumption was that the shape of the strip during the hot rolling has a strong effect on the behaviour of the steel strip in Rolling, Annealing and Pickling line (RAP). Our goal is to provide information that enables to react well in advance to strips with challenging shape. In this article, we show that the most critical shape errors arising in hot rolling process will be transferred to critical errors in RAP-line process as well. In addition, our results reveal that the most critical feature characterizes the deviation better than the currently used criterion for rework. The developed model enables the user to understand better the quality of the products, how the process works, and how the quality model predicts and performs.

Heterogeneous mobile devices are constrained by their limited battery capacity and available processing power. Yet, they are able to run the same or similar applications compared to a personal computer. Extensible Internet based services allow the originators to change the content and functionalities at any given time. The content of such a service varies by its size, complexity, datatype and availability. This in its turn exposes a problem for heterogeneous types of mobile devices to manage such ever changing content efficiently. The problem is emphasized even further when the content is related to complex 3D graphics as utilized, for example, by the most recent extensible 3D virtual environment services. This study is about 3D graphics rendering process of a mobile device and supporting content delivery architecture when visualizing and delivering 3D content of such extensible services. A number of content simplification methods are discussed, and a proposal is given as for in which part of the delivery architecture the methods should be used. The analysis is done from the viewpoint of the following attributes: (1) resulting mobile device energy consumption, (2) visual appearance of the 3D rendering, and (3) perceived user experience. The results are presented via two constructs: (1) the effect of the proposed simplification methods on the mobile device energy consumption and (2) the structure of 3D content delivery chain and the proposed network architecture for it. The constructs are evaluated from both practical and theoretical viewpoints. The main contribution is a proposal of 3D content delivery chain in detail and evidence how a properly implemented delivery chain allows utilization of content simplification mechanisms achieving up to 50% energy saving during the rendering process among the group of selected mobile devices. Finally, it shows that regardless of possible quality degradation, the users are ready to accept the visual artefacts if the co, Tiivistelmä Mobiililaitteet ovat rajoitettuja niiden akkukeston ja prosessointitehon suhteen. Silti ne ovat tarpeeksi kyvykkäitä ajamaan samoja tai samankaltaisia sovelluksia kuin tietokoneetkin. Laajennettavan Internet-pohjaisen palvelun käyttö mahdollistaa palvelun sisällön tai ominaisuuksien laajentamisen milloin tahansa. Tällaisen palvelun sisältö varioituu sen koon, kompleksisuuden, tietotyyppien ja saatavuuden mukaan. Tämä taas on ongelmallista rajoitetuille mobiililaitteille niiden yrittäessä hallita alati muuttuvaa sisältöä tehokkaasti. Ongelma korostuu entisestään, silloin kun sisältö liittyy 3D grafiikkaan, jota käytetään uusimmissa laajennettavissa 3D virtuaalitila palveluissa. Tämä työ tutkii mobiililaitteen 3D grafiikan visualisointiprosessia ja sitä tukevaa sisällön toimitusketjua, kun se liittyy laajennettavien 3D virtuaalitilojen visualisointiin. Työssä käydään läpi useita sisällön yksinkertaistamismenetelmiä ja, työn tuloksena ehdotetaan, missä kohden toimitusketjua menetelmiä tulisi käyttää. Menetelmien tarkastelua tehdään seuraavista näkökulmista: (1) yksinkertaistamista seuraava mobiililaitteen energiankulutus, (2) 3D visualisoinnin laadullinen lopputulos, ja (3) koettu loppukäyttäjäkokemus. Tulokset esitellään kahden konstruktion kautta: (1) ehdotettujen yksinkertaistamismenetelmien vaikutus mobiililaitteen energiankulutukseen ja (2) 3D sisällön jakeluketjun rakenne ja sen ehdotettu arkkitehtuuri. Konstruktiot evaluoidaan sekä käytännön että teorian näkökulmista. Työn pääkontribuutio on ehdotus 3D sisällön jakeluketjun rakenteelle ja näyttöä siitä, kuinka useilla eri mobiililaitteilla todennettuna työssä esitetyillä menetelmillä voidaan säästää jopa noin 50 % kokonaisenergiankulutuksesta riippuen valittujen menetelmien ja laitteiden yhdistelmästä. Työ osoittaa myös sen, että jos optimointi eri tietotyypeille tehdään hallitusti, voidaan lopputuloksen laatua kontrolloida siten, että se saavuttaa hyväksyttävän tason myös loppukäyttäjien arvioimana.