L'utilizzo dei casi di studio per favorire l'apprendimento di metodologie e strumenti per lo sviluppo organizzativo o di percorsi di cura nelle aziende sanitarie

La formazione, in ambito sanitario, è considerata una grande leva di orientamento dei comportamenti, ma la metodologia tradizionale di formazione frontale non è la più efficace, in particolare nella formazione continua o “long-life education”. L’obiettivo primario della tesi è verificare se l’utilizzo della metodologia dello “studio di caso”, di norma utilizzata nella ricerca empirica, può favorire, nel personale sanitario, l’apprendimento di metodi e strumenti di tipo organizzativo-gestionale, partendo dalla descrizione di processi, decisioni, risultati conseguiti in contesti reali. Sono stati progettati e realizzati 4 studi di caso con metodologia descrittiva, tre nell’Azienda USL di Piacenza e uno nell’Azienda USL di Bologna, con oggetti di studio differenti: la continuità di cura in una coorte di pazienti con stroke e l’utilizzo di strumenti di monitoraggio delle condizioni di autonomia; l’adozione di un approccio “patient-centred” nella presa in carico domiciliare di una persona con BPCO e il suo caregiver; la percezione che caregiver e Medici di Medicina Generale o altri professionisti hanno della rete aziendale Demenze e Alzheimer; la ricaduta della formazione di Pediatri di Libera Scelta sull’attività clinica. I casi di studio sono stati corredati da note di indirizzo per i docenti e sono stati sottoposti a quattro referee per la valutazione dei contenuti e della metodologia. Il secondo caso è stato somministrato a 130 professionisti sanitari all’interno di percorso di valutazione delle competenze e dei potenziali realizzato nell’AUSL di Bologna. I referee hanno commentato i casi e gli strumenti di lettura organizzativa, sottolineando la fruibilità, approvando la metodologia utilizzata, la coniugazione tra ambiti clinico-assistenziali e organizzativi, e le teaching note. Alla fine di ogni caso è presente la valutazione di ogni referee.

Confronto fra periodizzazione Tradizionale e a Blocchi per l’allenamento della forza in atleti maschi e partecipanti di genere femminile.

La prima parte dello studio riguarda la descrizione dell’origine e delle caratteristiche che differenziano la periodizzazione tradizionale e quella a blocchi per l’allenamento della forza. L’obiettivo della seconda parte del lavoro è stato quello di confrontare gli adattamento ormonali e prestativi ad un programma di allenamento della forza periodizzato secondo il modello tradizionale o secondo quello a blocchi in un campione di atleti di forza. Venticinque atleti maschi sono stati assegnati con procedura randomizzata al gruppo con programmazione tradizionale (TP) o a quello a blocchi (BP). Prelievi di saliva sono stati effettuati prima e dopo 6 diverse sedute di allenamento durante il programma al fine di rilevare i livelli di testosterone (T) e cortisolo (C). Le valutazioni dei parametri antropometrici e prestativi sono state effettuate prima e dopo le 15 settimane di allenamento previste. In nessuno dei due gruppi vi sono state variazioni significative nei livelli ormonali. I risultati indicano che il gruppo BP ha ottenuto incrementi superiori a quello TP riguardo alla forza massima (p = 0,040) ed alla potenza (p = 0,035) espressa alla panca piana. Nella terza parte dello studio, la periodizzazione tradizionale e quella a blocchi sono state confrontate riguardo agli effetti sulla forza massima e sull’ipertrofia in donne allenate di livello amatoriale. Diciassette donne hanno partecipato all’esperimento allenandosi 3 volte a settimana per 10 settimane. I risultati dimostrano che entrambe le periodizzazioni hanno portato a miglioramenti significativi di forza e potenza; il gruppo TP ha tuttavia ottenuto incrementi superiori di forza massima (p = 0,039) e ipertrofia degli arti inferiori (p = 0,004). La periodizzazione tradizionale quindi si è dimostrata più efficace per aumentare la forza massima e la sezione muscolare della coscia in partecipanti di genere femminile. I risultati contrastanti nei due generi potrebbero essere legati a rapporti diversi fra processi anabolici e catabolici.

Effect of training systems and pruning methods on fruit quality in apple

The introduction of dwarfed rootstocks in apple crop has led to a new concept of intensive planting systems with the aim of producing early high yield and with returns of the initial high investment. Although yield is an important aspect to the grower, the consumer has become demanding regards fruit quality and is generally attracted by appearance. To fulfil the consumer’s expectations the grower may need to choose a proper training system along with an ideal pruning technique, which ensure a good light distribution in different parts of the canopy and a marketable fruit quality in terms of size and skin colour. Although these aspects are important, these fruits might not reach the proper ripening stage within the canopy because they are often heterogeneous. To describe the variability present in a tree, a software (PlantToon®), was used to recreate the tree architecture in 3D in the two training systems. The ripening stage of each of the fruits was determined using a non-destructive device (DA-Meter), thus allowing to estimate the fruit ripening variability. This study deals with some of the main parameters that can influence fruit quality and ripening stage within the canopy and orchard management techniques that can ameliorate a ripening fruit homogeneity. Significant differences in fruit quality were found within the canopies due to their position, flowering time and bud wood age. Bi-axis appeared to be suitable for high density planting, even though the fruit quality traits resulted often similar to those obtained with a Slender Spindle, suggesting similar fruit light availability within the canopies. Crop load confirmed to be an important factor that influenced fruit quality as much as the interesting innovative pruning method “Click”, in intensive planting systems.

The Influence of Baseline Metabolic Rate and Exercise Training on Cardio-Respiratory Dynamics

Speeding the VO2 kinetics results in a reduction of the O2 deficit. Two factors might determine VO2 kinetics: oxygen delivery to muscle (Tschakovsky and Hughson 1999) and a muscle 'metabolic inertia' (Grassi et al. 1996). Therefore, in study 1 we investigated VO2 kinetics and cardiovascular system adaptations during step exercise transitions in different regions of the moderate domain. In study 2 we investigated muscle oxygenation and cardio-pulmonary adaptations during step exercise tests before, after and over a period of training. Study 1 methods: Seven subjects (26 ± 8 yr; 176 ± 5 cm; 69 ± 6 kg) performed 4 types of step transition from rest (0-50W; 0-100W) or elevate baseline (25-75W; 25-125W). GET and VO2max were assessed before testing. O2 uptake and were measured during testing. Study 2 methods: 10 subjects (25 ± 4 yr; 175 ± 9 cm; 71 ± 12 kg) performed a step transition test (0 to 100 W) before, after and during 4 weeks of endurance training (ET). VO2max and GET were assessed before and after of ET (40 minutes, 3 times a week, 60% O2max). VO2 uptake, Q and deoxyheamoglobin were measured during testing. Study 1 results: VO2 τ and the functional gain were slower in the upper regions of the moderate domain. Q increased more abruptly during rest to work condition. Q τ was faster than VO2 τ for each exercise step. Study 2 results: VO2 τ became faster after ET (25%) and particularly after 1 training session (4%). Q kinetics changed after 4 training sessions nevertheless it was always faster than VO2 τ. An attenuation in ∆[HHb] /∆VO2 was detectible. Conclusion: these investigations suggest that muscle fibres recruitment exerts a influence on the VO2 response within the moderate domain either during different forms of step transition or following ET.

Learning Methods and Algorithms for Semantic Text Classification across Multiple Domains

Information is nowadays a key resource: machine learning and data mining techniques have been developed to extract high-level information from great amounts of data. As most data comes in form of unstructured text in natural languages, research on text mining is currently very active and dealing with practical problems. Among these, text categorization deals with the automatic organization of large quantities of documents in priorly defined taxonomies of topic categories, possibly arranged in large hierarchies. In commonly proposed machine learning approaches, classifiers are automatically trained from pre-labeled documents: they can perform very accurate classification, but often require a consistent training set and notable computational effort. Methods for cross-domain text categorization have been proposed, allowing to leverage a set of labeled documents of one domain to classify those of another one. Most methods use advanced statistical techniques, usually involving tuning of parameters. A first contribution presented here is a method based on nearest centroid classification, where profiles of categories are generated from the known domain and then iteratively adapted to the unknown one. Despite being conceptually simple and having easily tuned parameters, this method achieves state-of-the-art accuracy in most benchmark datasets with fast running times. A second, deeper contribution involves the design of a domain-independent model to distinguish the degree and type of relatedness between arbitrary documents and topics, inferred from the different types of semantic relationships between respective representative words, identified by specific search algorithms. The application of this model is tested on both flat and hierarchical text categorization, where it potentially allows the efficient addition of new categories during classification. Results show that classification accuracy still requires improvements, but models generated from one domain are shown to be effectively able to be reused in a different one.

Robotic Sensing and Manipulation of Deformable Linear Objects with Learning-based methods

Nowadays robotic applications are widespread and most of the manipulation tasks are efficiently solved. However, Deformable-Objects (DOs) still represent a huge limitation for robots. The main difficulty in DOs manipulation is dealing with the shape and dynamics uncertainties, which prevents the use of model-based approaches (since they are excessively computationally complex) and makes sensory data difficult to interpret. This thesis reports the research activities aimed to address some applications in robotic manipulation and sensing of Deformable-Linear-Objects (DLOs), with particular focus to electric wires. In all the works, a significant effort was made in the study of an effective strategy for analyzing sensory signals with various machine learning algorithms. In the former part of the document, the main focus concerns the wire terminals, i.e. detection, grasping, and insertion. First, a pipeline that integrates vision and tactile sensing is developed, then further improvements are proposed for each module. A novel procedure is proposed to gather and label massive amounts of training images for object detection with minimal human intervention. Together with this strategy, we extend a generic object detector based on Convolutional-Neural-Networks for orientation prediction. The insertion task is also extended by developing a closed-loop control capable to guide the insertion of a longer and curved segment of wire through a hole, where the contact forces are estimated by means of a Recurrent-Neural-Network. In the latter part of the thesis, the interest shifts to the DLO shape. Robotic reshaping of a DLO is addressed by means of a sequence of pick-and-place primitives, while a decision making process driven by visual data learns the optimal grasping locations exploiting Deep Q-learning and finds the best releasing point. The success of the solution leverages on a reliable interpretation of the DLO shape. For this reason, further developments are made on the visual segmentation.

Concepts and methods for efficient decentralized learning in federated settings

Deep Neural Networks (DNNs) have revolutionized a wide range of applications beyond traditional machine learning and artificial intelligence fields, e.g., computer vision, healthcare, natural language processing and others. At the same time, edge devices have become central in our society, generating an unprecedented amount of data which could be used to train data-hungry models such as DNNs. However, the potentially sensitive or confidential nature of gathered data poses privacy concerns when storing and processing them in centralized locations. To this purpose, decentralized learning decouples model training from the need of directly accessing raw data, by alternating on-device training and periodic communications. The ability of distilling knowledge from decentralized data, however, comes at the cost of facing more challenging learning settings, such as coping with heterogeneous hardware and network connectivity, statistical diversity of data, and ensuring verifiable privacy guarantees. This Thesis proposes an extensive overview of decentralized learning literature, including a novel taxonomy and a detailed description of the most relevant system-level contributions in the related literature for privacy, communication efficiency, data and system heterogeneity, and poisoning defense. Next, this Thesis presents the design of an original solution to tackle communication efficiency and system heterogeneity, and empirically evaluates it on federated settings. For communication efficiency, an original method, specifically designed for Convolutional Neural Networks, is also described and evaluated against the state-of-the-art. Furthermore, this Thesis provides an in-depth review of recently proposed methods to tackle the performance degradation introduced by data heterogeneity, followed by empirical evaluations on challenging data distributions, highlighting strengths and possible weaknesses of the considered solutions. Finally, this Thesis presents a novel perspective on the usage of Knowledge Distillation as a mean for optimizing decentralized learning systems in settings characterized by data heterogeneity or system heterogeneity. Our vision on relevant future research directions close the manuscript.

Directly training spiking neural networks for cyber-physical systems: from supervised to reinforcement learning

Spiking Neural Networks (SNNs) are bio-inspired Artificial Neural Networks (ANNs) utilizing discrete spiking signals, akin to neuron communication in the brain, making them ideal for real-time and energy-efficient Cyber-Physical Systems (CPSs). This thesis explores their potential in Structural Health Monitoring (SHM), leveraging low-cost MEMS accelerometers for early damage detection in motorway bridges. The study focuses on Long Short-Term SNNs (LSNNs), although their complex learning processes pose challenges. Comparing LSNNs with other ANN models and training algorithms for SHM, findings indicate LSNNs' effectiveness in damage identification, comparable to ANNs trained using traditional methods. Additionally, an optimized embedded LSNN implementation demonstrates a 54% reduction in execution time, but with longer pre-processing due to spike-based encoding. Furthermore, SNNs are applied in UAV obstacle avoidance, trained directly using a Reinforcement Learning (RL) algorithm with event-based input from a Dynamic Vision Sensor (DVS). Performance evaluation against Convolutional Neural Networks (CNNs) highlights SNNs' superior energy efficiency, showing a 6x decrease in energy consumption. The study also investigates embedded SNN implementations' latency and throughput in real-world deployments, emphasizing their potential for energy-efficient monitoring systems. This research contributes to advancing SHM and UAV obstacle avoidance through SNNs' efficient information processing and decision-making capabilities within CPS domains.