Road accidents are often related to drivers’ psychological state and are frequently overlooked and dismissed. This includes the drivers’ mental health, which can be negatively affected by conditions such as stress, fatigue, and sadness. Emotional disturbances can impair driving abilities, posing significant risks to drivers and road users. Public health initiatives must focus on integrating innovative solutions to minimize fatalities and injuries. The Internet of Medical Things (IoMT) is a field of research that focuses on developing cost-effective nonintrusive new methods for assessing vital signs in non-clinical settings, such as homes and vehicles. The increasing use of these applications underscores the potential to address healthrelated road risks. Given the pressing concerns of road safety, this dissertation proposes the creation of an IoMT system that can revolutionize the driving experience. By introducing real-time monitoring of driver health, this system aims to address these challenges and significantly enhance road safety, a crucial need in our society. A systematic review, including the choice of thirtytwo relevant scientific publications on wearable devices for healthcare monitoring, was conducted to create a reliable system. The review utilized Natural Language Processing and the PRISMA methodology to analyze papers from various databases and considered population, methods, sensors, features, and communication protocols. The studies highlighted various hardware and software technologies used to enhance healthcare monitoring applications and the benefits and challenges associated with these applications, providing an overview of how to build an efficient system. Based on the results of the systematic review, the Driver Health System was proposed, integrating multiple layers with distinct roles to ensure efficiency and high performance. This dissertation proposes an innovative device for measuring the driver’s health data, integrating a comprehensive set of sensors and power management components to ensure reliable functionality. The device’s printed body encapsulates the PCB and battery, optimizing functionality and user comfort. The firmware developed for the device presented in this dissertation showcases the sensor drivers for photoplethysmography (PPG), accelerometer, barometric pressure, and fuel gauge sensors. The dissertation proposes a deep learning model designed to estimate the user’s heart rate by leveraging data from the PPG and accelerometer sensors. The model development involves multiple processing steps. Leaveone-session-out cross-validation and hyperparameter tuning techniques were employed for the model training and evaluation. The model achieved an outstanding Mean Absolute Error (MAE) of 3.450 ± 1.324 bpm and a Mean Squared Error (MSE) of 69.50 ± 93.57 bpm2 . The model was deployed in a custom WEB application for testing purposes. The dissertation describes the development of a custom mobile application for the Driver Health System, which offers crucial features such as intuitive real-time access to health status, device compatibility, power management, and integration of the heart rate estimation model to provide users with deeper insights into their health condition. This dissertation successfully enables a robust, innovative, real-time driver health monitoring solution. The Driver Health System represents a significant advancement at the intersection of healthcare industry and automotive sector. It aims to enhance road safety and establish a connected network that empowers to monitor and manage the drivers’ health effectively.

The main goal of this paper consists on the adaption and validation of a method for the measurement of the energy expenditure during physical activities. Sensors available in a mobile device, e.g., a smartphone, a smartwatch, or others, allow the capture of several signals, which may be used to the estimation of the energy expenditure. The adaption consists in the comparison between the units of the data acquired by a tri-axial accelerometer and a mobile device accelerometer. The tests were performed by healthy people with ages between 12 and 50 years old that performed several activities, such as standing, gym (walking), climbing stairs, walking, jumping, running, playing tennis, and squatting, with a mobile device on the waist. The validation of the method showed that the energy expenditure is underestimated and super estimated in some cases, but with reliable results. The creation of a validated method for the measurement of energy expenditure during physical activities capable for the implementation in a mobile application is an important issue for increase the acceptance of the mobile applications in the market. As verified the results obtained are around 124.6 kcal/h, for walking activity, and 149.7 kcal/h, for running activity.

Purpose – In this work, the paper aims to demonstrate the feasibility of plastic optical fiber (POF) based accelerometers for the structural health monitoring (SHM) of civil engineering structures based on measurements of their dynamic response, namely to estimate natural frequencies. These sensors use POFs, combining the advantages of the optical technology with the robustness of this particular kind of fiber. The POF sensor output is directly compared with the signal from an electrical sensor, demonstrating the potential use of such sensors in structural monitoring applications. Design/methodology/approach – Within this work, the paper demonstrates the feasibility of using a low-cost acceleration system based on a POF accelerometer on the dynamic monitoring of a civil engineering structure, aiming its natural frequency evaluation, which is a primary parameter to be used in SHM methods and numerical models calibration. Findings – A low-cost POF-based accelerometer was used in the characterization of a civil engineering structural component, located in a building at the University of Aveiro Campus, being used to estimate its natural frequency with a relative error of 0.36 percent, comparatively to the value estimated recurring to a calibrated electronic sensor. Originality/value – Optical fiber sensors take advantage of the fibers properties, such as immunity to electromagnetic interference and electrical isolation. They are very attractive for use in hostile environments, like submerse environments or flammable atmospheres where electrical currents might pose a hazard. The advantages of POF itself should also be considered, like resistance to hash environments, robustness, flexibility, low-cost interrogation units and high numeric aperture (lower cost components). The paper demonstrates the feasibility of using a low-cost acceleration system based on a POF accelerometer on the dynamic monitoring of a civil engineering structure, aiming its natural frequency evaluation.

Active vibration control systems can greatly reduce unwanted vibrations to improve the performance of machines or structures. Bosch Home Comfort Group (Wernau) intends to test the applicability of active vibration control systems in its heat pumps to reduce operating noise. Thus, it is necessary to monitor its mechanical vibrations. Piezoelectric transducers are generally used to measure accelerations on component surfaces, therefore monitoring their vibration levels. However, these types of sensors have a very high cost, which means that active vibration control systems are generally not economically viable. Micro Electric Mechanical Sensor (MEMS) accelerometers have been rapidly evolving, getting close to the performance of typical piezoelectric for certain frequency ranges. Given that their cost is much lower, they emerge as an alternative that could completely change the economic viability of active control systems, which in turn would allow their large-scale application in heat pumps. During the internship, a throughout analysis of several MEMS accelerometers was conducted, under three distinct frequency spectrums, by comparing their performance to reference piezoelectric accelerometer. Initially, a graphical user interface was developed to be used with a shaker testing setup and, on a latter stage, tests were conducted by mounting the accelerometers on Bosch’s heat pump compressor. Results have revealed that MEMS transducers can have a very comparable response to the piezoelectric accelerometers. The BMI323, IIS2DH, and IIS3DWB accelerometers are suitable for integration onto an active vibration control system operating within the 500Hz frequency range. Both the IIS2DH and IIS3DWB have also shown very satisfactory response signals comparatively to the reference transducer in the wider 1kHz range. Notably, the IIS3DWB revealed extended capabilities, demonstrating a reliable performance up to the 5kHz range. Utilising MEMS accelerometers onto an active vibration control system is essential to give this system the necessary economic viability for its largescale implementation in Bosch’s heat pumps. This would greatly improve heat pumps, making them less noise pollutant, and ultimately enhancing the safety and well-being of individuals.

Accelerometers are widely used in monitoring the structural integrity of many constructions, such as buildings and bridges. In this application, the sensor must have a high sensitivity and resonant frequency, both inversely proportional. Its working principle is based on the energy conversion of a stimulated mass (usually a cantilever) in a magnitude that allows signal acquisition. Typically, a sensor based on fibre Bragg gratings (FBG) technology translates the mechanical signals into wavelength shifts. The FBG-based accelerometers are highly efficient and have numerous advantages, such as easy multiplexing, minimised cabling and immunity to electromagnetic interference. This internship report aims to describe the developing process of an FBG-based accelerometer with high sensitivity (1000 pm/g) and resonant frequency (200 Hz) from the conceptual stage to prototype development and testing. The internship was carried out at HBK FiberSensing, in Porto, Portugal. The accelerometer developed consists of two masses connected by a main frame through flexures, configuring a symmetrical push-pull design. These flexures make the cantilever-based structure flexible. The use of two suspended gratings doubled the sensitivity and guaranteed temperature compensation. When comparing the simulated and experimental values, the resonant frequency computed an error of 3.65% and sensitivity was tested at different frequencies and accelerations, with the highest error value being 13.13%. However, the sensitivity and resonant frequency achieved with the prototype were respectively 24.10% and 74.10% higher than the requirements. The experimental tests have shown that the application of damping will be necessary for future improvement. Regarding the cross-sensitivity to temperature variation, the tests showed a value of 0.01 nm/°C, which is very low compared to the accelerometer’s sensitivity. The fatigue test allowed us to conclude that, considering the 100 million cycles applied, the accelerometer can have an infinite life. The proof of concept was validated with only one prototype modelling/manufacturing iteration. Some optimisations are still necessary to reach a minimum viable product (MVP), although the experimental results highlight the great potential of this solution.

ActiGraph® is a valid, frequently used, accelerometer to quantify moderate to vigorous physical activities (MVPA) in people with COPD. The impact of ActiGraph processing techniques on this population is unknown. This study aimed to explore the effect of data reduction techniques on MVPA in people with COPD. MVPA/day, through ActiGraph GT3X+, was estimated using: Troiano, Freedson 98 and FreedsonVM3 cutoffs, 15-s and 60-s epochs, and normal and low-frequency extension (LFE) filters. Cutoff, epoch, and filter effects were explored with Aligned Rank Transform-ANOVA. Lin’s concordance correlation coefficients and Bland–Altman plots were used to evaluate agreement and bias between different techniques. The analysis included 136 people with COPD (79% male; 68 ± 8 years; FEV1 51 ± 17% predicted). MVPA/day differed according to cutoff, filter, and epoch selection (p-value < 0.001). FreedsonVM3 cutoff, 15-s epochs, and LFE yielded the highest MVPA (45 min/day, 68% of physically active participants). Troiano cutoff, 60-s epochs, and normal filter yielded the lowest MVPA (8 min/day, 20% of physically active participants). Only comparisons between Troiano and Freedson98 cutoffs presented an almost perfect agreement. ActiGraph data reduction techniques affected MVPA/day estimates and their interpretation at the individual and group level. Studies using different processing criteria should not be compared in people with COPD. Future studies with a gold standard are required to ascertain which processing technique produces the most accurate MVPA estimates in COPD. Meanwhile, future trials employing the ActiGraph GT3X+ may consider estimating MVPA based on Freedson VM3 cutofffs, 60-s epochs, and normal filter.

Introduction: Self-report physical activity (PA) measures are often used to assess changes before/after chronic obstructive pulmonary disease (COPD) rehabilitation, as they are easy to employ and more feasible than objective measures. However, the ability of self-report measures to detect intervention-related changes in PA should be determined.

Animal behaviour monitoring is a topic of increasing importance in the past few years, due to the needs to optimize the multiple fields of animal production to answer worldwide population growth. Considering that animal behaviour is an indicator of animal welfare, by automating and scaling this process, it is possible to constantly monitor entire flocks, allowing for the early detection of diseases which might otherwise incur into huge costs and losses. Simultaneously, the cost of microprocessors and small sensors is decreasing. This allows for the construction of low-powered devices capable of both retrieving and classifying animal behaviour, resorting to Machine Learning. Here, the study revolves around a collar equipped with 3 sensors: accelerometer, gyroscope, and thermometer, resulting in an initial set of 12 features. From the wide variety of algorithms, Decision Trees are particularly suitable for this task due to their many advantages, namely simplicity and readability. The latter allows for a conversion of the developed models into raw code, passive of being directly applied on the devices. Therefore, the goal of this dissertation is to evaluate the ability of Decision Trees to classify six main behaviours (Lying, Standing, Inclined, Eating, Walking and Ruminating), while simultaneously assessing which sensors and features are more valuable to optimize the production costs of said devices. Overall, it is possible to classify these behaviours reliably (with a balanced accuracy around 90%) and remove one of the sensors, without impacting model’s performance significantly. Notwithstanding, Temperature surprisingly arises as a potential important feature to predict postures. This thesis was realized with the support of the Instituto de Telecomunicações.

This study assessed the test-retest reliability/agreement and construct validity of the International Physical Activity Questionnaire short-form (IPAQ-sf) in patients with chronic obstructive pulmonary disease (COPD). It also explored differences in its validity according to age, sex and GOLD airflow obstruction levels. 62 participants (68 ± 8 years, 53 males, FEV1 51 ± 23%pred) completed the Portuguese IPAQ-sf, wore an accelerometer for 7 days and completed a second IPAQ-sf. Test-retest reliability/agreement was assessed with Intraclass Correlation Coefficient (ICC2,1), 95% Limits of Agreement (LoA), standard error of measurement (SEM) and minimal detectable change (MDC95) for continuous variables, and percentage of agreement (%agreement) for categories (“active”/“inactive”). Validity was assessed with 95% LoA and Spearman's correlations (ρ) between IPAQ-sf 2 (METs-min/week, time in vigorous [VPA], moderate PA [MPA] and walking) and accelerometry (time in MVPA, VPA, MPA and step counts) for continuous variables; %agreement, Cohen's kappa, and sensitivity specificity and±predictive values for categories. Correlations were also performed for age, sex and GOLD airflow obstruction grades. Reliability was good (ICC2,1 = 0.707) with wide LoA (-6446—6409 METs-min/week). SEM and MDC95 were 1840 and 4971 METs-min/week, respectively. %agreement between the two IPAQ-sf was 84% (kappa = 0.660). Positive, moderate and significant correlations were found between IPAQ-sf and accelerometry (0.396 ≤ ρ ≤ 0.527, p < 0.001), except for VPA (p > 0.05). The strongest correlations were found in age (<65 years) and male (0.466 ≤ ρ ≤ 0.653, p < 0.05). %agreement between tools was 65% (kappa = 0.313), with high sensitivity (0.830) but low specificity (0.500). The IPAQ-sf seems valid to be used in COPD but caution on its widespread use is recommended as its accuracy may be limited.

This dissertation addresses the challenge of automatically detecting lambing events in real-time by leveraging sensor data gathered from collars worn by ewes. The primary motivation stems from the labor-intensive nature of manual observation and the need for prompt interventions around lambing, which can significantly affect animal well-being and farm efficiency. Fifty-four pregnant ewes from a norwegian farm were equipped with collars that recorded accelerometer X, Y, and Z axes and temperature data around the clock. Key preprocessing steps included thresholding and Interquartile Range (IQR) filtering to remove invalid or extreme values, Median aggregation to downsample the raw data (reducing size by over 95%) while preserving critical trends, scaling through min-max normalization to ensure consistency across features and files, and sampling strategies like undersampling and oversampling to mitigate severe class imbalance between the relatively brief lambing window and the much larger null periods. Using this refined dataset, multiple machine learning approaches were explored. Ensemble-based classifiers, specifically Random Forest and Extra Trees, consistently outperformed alternative algorithms, achieving high accuracy and Mathews Correlation Coefficient (MCC). Incorporating a short-term rolling window, like 60 seconds of past data, further boosted performance by capturing temporal patterns in sensor signals leading up to lambing. The top model achieved 0.83 MCC and 0.85 accuracy in distinguishing lambing states with sub-second inference speeds. Additionally, lightweight model variants were tested to measure suitability for on-collar microcontrollers. Although restricting model depth and estimator counts did reduce computational load, it also lowered overall accuracy, achieving MCC 0.58, suggesting further optimization or specialized hardware might be needed for fully embedded, real-time detection. Overall, the integrated pipeline and spanning data collection, rigorous preprocessing, and ensemble-based modeling shows promising potential for collar-based lambing detection. Nonetheless, false positives remain an area of concern, and further refinements are required to achieve the level of real-time, high-precision performance needed in practical farm settings. By reducing reliance on labor-intensive observation, this framework holds promise for advancing animal welfare and management efficiency across modern sheep farming operations.