Narrow your search

Library

KU Leuven (47)


Resource type

dissertation (47)


Language

English (46)

Dutch (1)


Year
From To Submit

2024 (8)

2023 (6)

2022 (8)

2021 (8)

2020 (5)

More...
Listing 1 - 10 of 47 << page
of 5
>>
Sort by

Dissertation
Heat thermal manipulation during incubation of broiler embryos: a meta-analysis

Loading...
Export citation

Choose an application

Bookmark

Abstract

Heat stress is one of the most challenging environmental stresses for broilers. Nowadays, it is becoming more prominent since the temperature around the world is increasing, due to climate change. This results in more and longer heat waves with higher temperature maxima. Farmers must find solutions for these severe heat exposures to avoid economic losses and to maintain the welfare of the broilers. Increased heat tolerance seems to be a promising solution. This systematic review investigates the effect of heat thermal manipulation (TM) during incubation on the post-hatch life of TM broilers compared to non-treated broilers, in standard conditions and under chronic/acute heat stress. Therefore, keywords regarding oxidative stress, thyroid gland, and production factors are searched to form a Boolean expression. This expression is used in the advanced search from Web of Science (WoS) to collect relevant papers. Then, some filtering techniques in WoS are applied together with exclusion criteria in the program StArt, to eliminate the irrelevant papers. From the remaining papers, the data is extracted according to a premade template in Excel. Excel is also used to visualise and analyse the results. Finally, the program R is used for a statistical analysis via a chi-squared test and the calculation of the standardised residuals. The results showed that the thermal tolerance of broilers can be increased via TM during incubation. This increased thermal tolerance is reached via a decrease in T3 levels, which leads to a lower body temperature and results in less stress/corticosterone levels, when exposed to heat stress. Eventually this results in lower mortality and an increased body weight of heat stressed TM broilers, compared to non-treated broilers. All of this can be achieved without negatively affecting these parameters under standard conditions. This means that in regions where broilers experience heat stress, TM during incubation can increase the profits of the farmer and increase the welfare of the broilers. However, it is important to apply the correct TM parameters during incubation. These are estimated at 38.5-39.0°C around embryotic days 7-18 for a duration between 9-18 hours/day.

Keywords


Dissertation
Depression and Burnout Diagnosis by combining Data-Based Mechanistic Modelling and Machine Learning Methods

Loading...
Export citation

Choose an application

Bookmark

Abstract

In this thesis, we have developed a new diagnostic tool for depression and burnout by using a two step approach: first, a data-based mechanistic modelling whose results are fed in the second step into machine learning models, to classify a healthy or non healthy patient. We used ECG signals from 27 participants biking in a controlled environment as well as their power exerted during the biking as input for data-based mechanistic modelling. As a result, we could extract personalised and biologically based model features: YIC, R2, order of the A- and B- polynomials, the time delay, the slope, stressPRE, stressMIST, stressPOST, the steady state gain (SSG). These features were the outcome of the model identification phase. Furthermore, these model features were fed into a classifier by using various machine learning models, including logistic regression, k-NN, decision tree, SVM and LDA. The best accuracy for was given by decision tree classifier which is a higly interpretable model, showing how the decision making process leading to the result. Hereby, the most important model features were YIC and R2. Furthermore, its accuracy was found as 74.07% which is a promising result for a new diagnostic tool as 75% is the current state of the art. In addition, thanks to biologically-based and personalised model features, the diagnoses are believed to be more accurate and reliable.

Keywords


Dissertation
Monitoren van cardiovasculaire responsies tijdens sportvissen

Loading...
Export citation

Choose an application

Bookmark

Abstract

Om allerlei chronische niet-overdraagbare aandoeningen zoals cardiovasculaire ziektes, diabetes, overgewicht en kanker te voorkomen heeft de mens nood aan een bepaalde hoeveelheid fysieke activiteit. Al sinds het ontstaan van de mensheid werd voldoende fysieke activiteit gehaald uit dagelijkse bezigheden, zij het in de oudheid door jagen en eten verzamelen, of in modernere tijden door fysieke inspanningen op het werk of door het uitvoeren van huishoudelijke taken. De fysieke inspanning tijdens vele van deze dagelijkse bezigheden is in de hedendaagse samenleving echter sterk afgenomen door de automatisering van zowel de werk- als de privéomgeving. Ondanks de vele voordelen van deze automatisering ondervindt de globale bevolking dan ook één groot nadeel: het gebrek aan fysieke activiteit en de daarmee verbonden risico’s voor de menselijke gezondheid. Dit eindwerk kadert dan ook in het doel naar het vinden voor een alternatief om voldoende fysiek actief te zijn. Concreter worden de volgende twee onderzoeksvragen opgesteld: Ten eerste wat is de hartslagrespons tijdens het sportvissen op activiteiten die typisch zijn voor het vissen met een vaste hengel en een werphengel onder gecontroleerde omstandigheden? Ten tweede wat is de hartslagrespons tijdens het sportvissen op activiteiten die typisch zijn voor het vissen met een vaste hengel in de praktijk? In een eerste experiment werd de fysieke activiteit op basis van hartslag en 3D acceleratie geschat tijdens het simuleren van hengelen met een vaste hengel en het simuleren van actief hengelen met een werphengel en kunstaas op roofvis. Ter vergelijking werden hartslag en 3D acceleratie ook gemeten tijdens rustperiodes en wanneer er gewandeld werd. Uit de resultaten bleek dat voor het simuleren van hengelen met de vaste hengel de hartslag wel hoger lag ten opzichte van de rustperiodes, maar dat deze gemiddeld gezien 20.00% (75 BPM ten opzichte van 90 BPM) onder de hartslag tijdens het wandelen bleef. Enkel tijdens het simuleren van hengelen met de werphengel steeg de hartslag significant in vergelijking met deze tijdens het wandelen (respectievelijk 90 BPM en 94 BPM, of dus een stijging van 4.44%). Het actief hengelen met een werphengel lijkt dan ook een inspanning te zijn met voldoende fysieke activiteit. In een tweede experiment werd er effectief gevist langs de waterkant. Er werd geopteerd om enkel met de vaste hengel te vissen en dit op stromend water aangezien dat het meest voorkomende type hengelen is in Vlaanderen. Opnieuw werden hartslag en 3D acceleratie gemeten tijdens het experiment om nadien te kunnen vergelijken met rust. Opvallend was dat dit maal de hartslag gemiddeld gezien lager lag tijdens het hengelen dan tijdens de rustperiodes, namelijk 2.94% in vergelijking met de hartslag tijdens het zitten en 14.71% met die tijdens het staan. Hier kon dus uit afgeleid worden dat het hengelen met een vaste hengel eerder een relaxerende en dus anti-stresserende werking heeft, maar dus geen oplossing biedt voor het probleem van de fysieke inactiviteit. De fysieke activiteit tijdens het recreatief sportvissen is dus sterk afhankelijk van het type hengelen waarvoor geopteerd wordt. Actievere vormen zoals het hengelen met de werphengel met kunstaas op roofvis bieden een goed alternatief om het probleem van de fysieke inactiviteit aan te pakken, rustigere vormen zoals het hengelen met de vaste hengel echter niet. Belangrijk is wel om in het achterhoofd te houden dat het niet zo is

Keywords


Dissertation
Residential Environment, Urban Heat, and Distress: a Longitudinal Panel Study

Loading...
Export citation

Choose an application

Bookmark

Abstract

In the last few decades, heat waves have become regular phenomena that led to significant mortality worldwide. In the future, climate change can further amplify the prevalence, magnitude and duration of such events. This is especially true for cities, generally experiencing higher temperatures than their surroundings, also referred to as the urban heat island (UHI) effect. With cities expanding due to an increasing urban population, the UHI effect will only intensify and affect more people, leading to negative health effects. In order to effectively implement mitigation strategies, city managers need guidance from researchers highlighting heat stress and heat vulnerability hotspots. This thesis identified the underlying heat stress drivers in Leuven by combining local climate zone (LCZ), temperature and heat stress, and heat vulnerability data into three piecewise structural equation models (SEM). First, we made use of an automated classification tool to produce a LCZ map, which divided the city into different zones that share similar characteristics in terms of building height, building spacing and amount of vegetation. Additionally, we used urban temperature data and heat stress indicators to test the performance of the LCZ map to differentiate between the various LCZ classes. Next, we extracted heat vulnerability data from a questionnaire that was distributed in the summer of 2021. Based on heat stress data from the same questionnaire, stress components were calculated, including a mental stress component (PSS) that served as the final outcome variable in the models. The first model was based on data from a period of relatively normal temperatures (control event) and showed that socio-economic variables determined basal levels of distress in the participants. Secondly, a model was built based on a period of relatively high temperatures (heat event) and indicated that socio-economic variables still explained most of the distress. Finally, we subtracted the control event data from the heat event data to derive a normalized model. Here, the temperature and LCZ data explained the additional distress due to heat, while socio-economic variables were insignificant. These results suggest that we need to tackle additional heat from heat waves by focusing on infrastructure in densely built LCZs. This research encourages city planners to prioritize adapting city infrastructure in order to mitigate for future heat waves and prevent an escalation of heat stress and human health issues.

Keywords


Dissertation
Development of Light-weighted pig pose estimation algorithm in diverse conditions

Loading...
Export citation

Choose an application

Bookmark

Abstract

The demand for pork continues to increase worldwide, making it crucial to ensure that pig production is sustainable while maintaining high standards of animal welfare and health. Our motivation behind this task was to optimize the pose estimation model for practical deployment in farm settings. This dissertation aims to develop a lightweight and accurate key point detection model for pigs on a farm, based on HRnet, a high-resolution network that preserves spatial accuracy throughout the process. To achieve the objective, we first review the latest methods for pig pose estimation and compare two paradigms for key point detection: top-down and bottom-up. We then describe the dataset of pig pictures that covers various environments and poses, and the annotation protocol that follows COCO annotator. We also introduce several metrics to evaluate the performance of key point detection algorithms, such as PCK, AUC, and EPE. Next, we report the experiments and results of applying HRnet to our dataset, and analyse the effects of different hyperparameters, such as batch sizes and input image size. We find that HRnet can achieve fast and accurate detection of 10 key points for pigs even with low-resolution images of [64,32] pixels, which enables efficient and ethical livestock farming. Finally, we conclude with some future work suggestions and contributions to the field of animal behaviour analysis.

Keywords


Dissertation
Monitoring and Controlling of Human Thermal Comfort using Wearable Technology

Loading...
Export citation

Choose an application

Bookmark

Abstract

Thermal comfort is a subjective feeling and it affects human well-being and productivity. So, there has been lots of attempts to develop thermal sensation models and to design controllers for HVAC system. However, the conventional control approaches are based on a static approach, in which humans are considered as passive recipients. Besides, thermal sensation models involve invasive/inconvenient measurements such as measurement of core body temperature and metabolic rate. Moreover, the models are not completely personalized. This dissertation investigates a completely personalized, non-intrusive thermal sensation modelling and controlling system that can be implemented with the help of wearable technology. The data collected from 25 subjects at 3 activity levels (low, medium and high activity levels), under 3 ambient temperatures (5 °C, 24 °C and 37 °C) were used for this purpose. Thermal sensation prediction models (reduced ordered, multi-input single output transfer function models) were developed for each person at each ambient temperature, with 16 different input combinations. The best results were obtained when heart rate (HR), skin temperature (T_skin) and ambient temperature (T_amb) were used as inputs, with an average R^2 of 0.956, 0.955 and 0.954 (and YIC values of -6.63, -6.75 and -6.68) at 5 °C, 24 °C and 37 °C respectively. With these best inputs (HR, T_skin and T_amb), the thermal sensation prediction models were also developed by taking all the three T_amb data together (instead of one model for one T_amb), with an average R^2 and YIC values of 0.942 and -7.64 respectively. Additionally, model Predictive Controllers were designed with each developed prediction model to maintain thermal sensation of each individual within the comfort level by using T_amb as the manipulated variable and HR and T_skin as measured disturbances. It is possible to maintain the output (thermal sensation) within the limits, however, the expected change in T_amb with changes in HR and T_skin (which is an indication of activity level/metabolic rate) is depicted only when the prediction models were developed with all the three T_amb data, rather than with one T_amb data. The controlled T_amb can be fed into the HVAC as the set point. This work gives promising results for the development of non-invasive thermal sensation monitoring and thermal environment controlling system with the help of wearable technology.

Keywords


Dissertation
Time and frequency domain analysis of eggshell heat flux and temperature during chicken embryonic development

Loading...
Export citation

Choose an application

Bookmark

Abstract

In recent years, the genetic selection of the meat-type broiler chicken resulted in a fast-growing animal. Apart from the fast body and muscle growth, an undesirable alteration to the chicken' thermoregulatory system have made these animals vulnerable for heat stress, especially in warm regions. Pre-hatch thermal conditioning or manipulation has become a major topic to prevent heat stress by altering the thermotolerance and thermosensation of the developing embryo. More knowledge is needed on how to assess these modifications due to the changes in thermal regime, in a non-invasive way. In this thesis, both frequency- and time-domain analysis were used to characterise the measured thermal respons of the chicken using sensors, expressed in both eggshell temperature and heat flux. Using the Short-Time Fourier Transformation (STFT), the spectral content of the whole incubation period was explored in a time-frequency distribution (TFD), i.e. the spectrogram. This frequency representation of the signal yielded new insights about some rhythmic components linked with embryonic development. However, the fixed time-frequency resolution of the STFT lacked performance in and around the transient parts of the signal, e.g. the temperature step inputs of the thermal treatment. Therefore, three different methods were proposed to visualise and quantify these step inputs in both time and frequency. Continuous Wavelet Transformation (CWT) in the form of a scalogram was proposed as good alternative for the spectrograms. These multi-resolution TFDs reveiled a new frequency component in between the temperature step which increased in frequency and magnitude each consecutive day of treatment. This component seemed to associate with the embryo' respons during the step input in temperature. In order to justify and quantify this assumption, wavelet decomposition was proposed as a quantitative tool to discriminate these treatments days. Apart from a clear difference between the first and last days in terms of energy, the visualisation of the detailed coefficient values suggest a more coordinated respons during the temperature steps in the last three days of incubation. Next, SISO Continuous-Time Transfer Function modelling was performed for both temperature and heatflux. The developing embryo' transient response is represented by a (mechanical) characteristic equation of the estimated transfer function, of which the eggshell temperature and heatflux were proven to be the step- and impulse respons of the respective system. Finally, the natural frequency and damping coefficient were linked to the frequency results, indicating a shift in the thermoregulatory system from proximate non-adaptive to a proximate adaptive mechanism in the last phase of incubation. It was concluded that simply placing a surrounding temperature sensor is enough to capture the rhythmic expressions from the developing embryo.

Keywords


Dissertation
Dynamic analysis of movement protocol during simulated Trojan war combat

Loading...
Export citation

Choose an application

Bookmark

Abstract

It is of great interest to investigate the soldier in ancient war combat from a physiological point of view. The Ancient war re-enactment with wearable technology allows us to analyse this scenario. This analysis allows us to gain more knowledge about the severity of the ancient war, the physiological state change of the soldier, and the appropriate level of ancient weapons and armour. It can also help us gain insight into the physiological reaction of the human body under high-intensity exercise regimes. In this research, we use the data generated from a Trojan war combat re-enactment in the summer of 2018. There are 13 volunteered participants wearing a Dendra panoply (oldest armour) perform the tasks of the experiment protocol, and their primary physiological data like heart rate, temperature, and acceleration were continuously recorded. Besides, their other physiological states like fatigue, perceived exertion, and reaction to the visual and acoustic stimulus are tested during the small break. For a better understanding of the dynamic process of the human body in this kind of scenario, dynamic models representing the physiological process of the human body are constructed. After that, some features (SSG of the model) are extracted from the model and interpreted physiologically. When analysing the extracted features as time goes through the experiment protocol, the SSG_(HR_to_T_(Core-Skin) ) and SSG_(HR_to_T_(Skin-Air) ) are consistent increase and decrease, respectively. The value level of SSG for different combat activities are similar, thus not recommended as a feature for exercise pattern recognition. The fatigue, perceived exertion, and thermal index for the average results of the group have a significant linear relationship (absolute correlation coefficients range from 0.85 to 0.97) with time during the experiment. Besides, the extracted features have a very weak linear relationship with those physiological state with all absolute correlation coefficient smaller than 0.5. To conclude, the data collected by wearable devices from the experiment can be modelled well, and the results together with physiological states can be interpreted physiologically. However, for the use of recognizing activities and relating the extracted features with the physiological state, it is not promising.

Keywords


Dissertation
Innovating the fabrication of microfluidic chips for ultrasensitive detection of biomarkers

Loading...
Export citation

Choose an application

Bookmark

Abstract

Previous studies have shown microwell arrays can be used in conducting digital bioassays. In the digital ELISAs (dELISAs) explored in this thesis, single molecules are captured on magnetic beads, labelled with an enzyme, and confined in individual microwells with a fluorogenic substrate. As the resulting fluorescent signal is confined to a small volume of solution, local concentrations are high enough to produce a detectable signal. An efficient way to confine the contents of each well is by using hydrophilic in hydrophobic (HIH) microwell arrays. This thesis builds on previous work in order to innovate the fabrication of microfluidic chips with HIH microwell arrays for ultrasensitive detection of biomarkers using a reaction injection molding (RIM) process on the mOSTE polymer. The RIM process was chosen for its scalability, for many other microwell fabrication techniques thus far had relied on technologies more difficult to scale to commercial environments. This thesis further optimized the RIM procedure for mOSTE into a process that was more durable, robust, and produced chips with fewer quality defects. A mOSTE casting procedure was developed as one that was more amenable to laboratory environments, and this was compared with RIM in terms of ease-of-use and the quality of the chips produced. Surface energy characterization experiments of the chips led to increased mold incubation time before curing and the development of a more precise predictive model to describe the effect of incubation. This served to better facilitate droplet movement, a key factor in executing a dELISA. Finally, using chips generated with the improved fabrication procedures, the potential of performing digital bioassays in the microwells was assessed by examining the ability to seal aqueous reagents and seed magnetic beads, key steps of the dELISA. A seeding efficiency of 97% was obtained, indicating a high potential for this process. The model bioassay was then conducted as a proof-of-concept that the new microwell arrays were capable of enabling protein detection. Although further work is needed in order to have a more repeatable bioassay, this study demonstrates feasibility and shows the potential for RIM and mOSTE to produce HIH microwell arrays for bioassays.

Keywords


Dissertation
Monitoring truck drivers' mental energy via wearables

Loading...
Export citation

Choose an application

Bookmark

Abstract

Due to the pressure of shift working and long driving times, truck drivers have a very stressful life. The combination of time pressure, an unhealthy lifestyle and general fatigue among drivers cause them to have a very high mental load, which consequently can impact their performance. Therefore, this thesis aims to reduce the mental load of drivers and investigate the impact of their mental state on their driving performance. For this purpose, 19 truck drivers were monitored using wearables. Using physiological data and information on their logged activities made it possible to determine which of their performed activities uses and recovers the most mental energy. Resting appeared to be the only refuelling activity among all drivers and had an average mental balance score (MBS) of -11.24%. On the other hand, Unhitching was the activity that used the most mental and had the highest average MBS (22.35%). Although there were general trends between the drivers, the energy drainer and refuelling activities proved to be individually different (average IQR=16.17%). Therefore, with concrete examples, it was shown that both the individual and group level could be used as insights for making recommendations in mentoring drivers or planning activities. Furthermore, the stepwise regression method was used to find the linear models that could best describe driver performance by means of the physiological data. The individual linear models showed an average model fit of 49.1% (R2min=15.4%, R2max=76.4%) and 40.0% (R2min=9.4%, R2max=97.8%), for hard braking and fuel efficiency, respectively. Although the average correlation was rather limited, it was shown that a part of the variability in performance could be described by changes in the physical and mental status of the driver. The impact of a change in physiological data on the drivers’ performance was investigated via odds ratios. These odds ratios indicate, for hard braking, that a unit increase of the physical heart rate component and the mental energy use will result, on average, respectively, in a 1.46 and 1.30 times higher probability of more hard braking events. For fuel efficiency, a unit increase of the physical heart rate component, on average, resulted in a 1.14 times higher probability of higher fuel consumption. Truck companies can use the methodology used in this thesis to improve the working conditions for truckers, which will lead to improved mental health for their truckers and more efficient performance for both the truckers and the company.

Keywords

Listing 1 - 10 of 47 << page
of 5
>>
Sort by