The decision whether or not to discharge an outpatient is vital in determining outpatient clinic attendance numbers, directly affecting overall patient care efficiency. A review of the factors influencing discharge decisions revealed that there was limited evidence of these factors and a lack of understanding how clinicians take discharge decisions. This project’s objectives were to describe the influential factors on discharge decisions from the clinicians’ and patients’ perspectives, to demonstrate the development and clinical evaluation of a novel "Traffic-light” design dermatology outpatient discharge information checklist to improve appropriateness and consistency in discharge decision-making. Semi-structured interviews were carried out with 40 consultant dermatologists across England. 148 influences were generated and thematically analysed manually and using NVivo10 software. A wide array of nonclinical factors, clinician-based, patient-based, practice-based and policy-based, influence discharge decision-making. Observations of 64 consultations and 56 semi-structured interviews with dermatology outpatients were carried out to understand their experience concerning the decision for their discharge. Twelve of 31 patients (39%) who were discharged considered their discharge inappropriate. A three-round Delphi exercise with 17 dermatology consultants (100% response) was carried out to reach agreement on what a high quality discharge information checklist should contain. There was strong inter-rater reliability (ICC=0.958) and fair inter-rater agreement (Fleiss Kappa=0.269). Thirteen items were identified that formed the "Trafficlight" design checklist. Twelve (67%) dermatology clinicians who evaluated the checklist found it useful. This study has demonstrated the importance of approaching discharge decision taking in an informed, structured manner. The checklist provide the basis for making discharge decisions more systematic, auditable and transparent, improving patient safety and optimising healthcare costs. These methods are potentially useful in other clinical disciplines.

No description available.

In 5G, Ultra-reliable and low-Latency communications (URLLC) service is envisioned to enable use cases with strict reliability and latency requirements on wireless communication. For the upcoming 6G network, machine learning (ML) also stands an important role that introduces intelligence and further enhances the system performance. This thesis explores the deployment of reinforcement learning (RL), a popular sub-field of ML, to optimize the application-layer availability and reliability of URLLC service in factory automation scenarios. In conventional RL methods, the decision variables are typically optimized in the same control loop. However, wireless systems’ parameters can be optimized either on a cell level or globally, depending on the inter-cell dynamics’ impact on their optimal value. Although global optimizations can provide a better performance, such optimizations introduce major practical limitations on the control loop’s delay. Besides, global optimization of all decision variables leads to excessive signalings, and thus, it is costly in terms of communication overhead. In this thesis, we propose a more flexible hierarchical reinforcement learning (HRL) framework that enables the implementation of multiple agents and multi-level policies with different time scales for each optimization. Therefore, we selected a use case from the prior art, optimizing the maximum number of retransmissions and transmission power to industrial devices, and solved it with our HRL framework. Our simulation results on factory automation scenario shows that HRL framework achieves similar performance as the ideal RL method, which highly improves the availability and reliability compared to the baseline solutions. Besides, the new HRL framework allows a more flexible allocation of agents. By allocating the low-level agents close to the base stations, our framework also significantly decreases the overhead of signal transmissions compared to the one-agent RL method. / Inom 5G är tjänster kallade “Ultra-reliable and low-latency communication” URLLC tänkta att möjliggöra trådlös kommunikation i användningsfall med strikta krav på tillförlitlighet och latens. För framtidens 6G nätverk har även maskininlärning ML en viktig roll som introducerar intelligens och ytterligare förbättrar systemens prestanda. Den här avhandlingen utforskar implementeringen av förstärkande inlärning (reinforcement learning eller RL), ett populärt underområde av ML, för att optimera tillgängligheten och tillförlitligheten av URLLC-tjänster i automatiserade fabriker. I traditionella RL-metoder optimeras beslutsvariablerna vanligtvis i samma kontrollslinga. Parametrarna för trådlösa system kan dock optimeras antingen på cellnivå eller globalt, beroende på inverkan av dynamiken mellan cellerna på deras optimala värde. Även om globala optimeringar kan ge bättre prestanda introducerar sådana optimeringar stora praktiska begränsningar på kontrollslingans latens. Dessutom leder global optimering av beslutsvariablerna till ökad signalering och är därför kostsamt. I denna avhandling föreslår vi ett mer flexibelt ramverk med hierarkisk förstärkande inlärning HRL som möjliggör implementering av flera agenter och flernivå-policys med olika tidsskalor för varje optimering. Därför valde vi ett tidigare känt användningsfall, optimeringen av det maximala antalet återsändningar samt överföringseffekten till industriella enheter, och löste det med vårt HRL ramverk. Resultaten från våra simuleringar på fabriksscenariot visar att HRL-ramverket uppnår liknande prestanda som den ideala RL-metoden, vilket i hög grad förbättrar tillgängligheten och tillförlitligheten jämfört med standardlösningarna. Dessutom tillåter det nya HRL ramverket en mer flexibel fördelning av agenter. Genom att allokera lågnivåagenterna nära basstationerna minskar vårt ramverk också avsevärt kostnaden för signalöverföringar jämfört med RL-metoden med endast en agent.

Background: Metastases are the major cause of death from melanoma, a skin cancer which has the fastest rising incidence of any malignancy in the Western world. Molecular pathways that drive melanoblast migration in development are believed to underpin the movement and ultimately the metastasis of melanoma. Aims: In this thesis we use genetically modified mice models to characterise two novel anticancer targets, P-Rex1 and focal adhesion kinase (FAK). Embryonic melanoblast migration is compared with cancer outcomes for each genetic modification. Results: Mice lacking P-Rex1, a Rac-specific Rho GTPase guanine nucleotide exchange factor (GEF), have a melanoblast migration defect during development evidenced by a white belly. These P-Rex1-/- mice are resistant to metastasis when crossed to a murine model of melanoma, an effect specifically channeled through loss of P-Rex1 GEF activity. FAK disruption compromises melanoblast cell numbers and migration in development, but has no long-term effect on melanocyte homeostasis. FAK-deleted mice have a divergent role in melanomagenesis, delaying primary melanoma onset whilst promoting metastasis following disease onset. Conclusions: We conclude that P-Rex1 and FAK play important roles in melanoblast embryology and melanoma development and progression. Both P-Rex1 and FAK represent interesting therapeutic targets for the treatment of cancer.

Melanoma is notoriously resistant to immuno- and even targeted chemotherapeutic strategies despite recent advances in drug development. The overall mortality of melanoma correlates with its ability to metastasise. Breslow thickness or the depth of invasion remains the most useful prognostic indicator, thereby linking the ability of the cells to invade with their propensity to metastasise. Invasion occurs early during tumour development, but the factors driving this process remain poorly understood. There is a growing appreciation that chemotaxis plays an important role in driving the migration and invasion of melanoma cells, but the key stimuli are not known. Through the generation and validation of an improved chamber for cancer cell chemotaxis, melanoma cells are shown to create chemotactic gradients that drive or disperse themselves outwards with remarkable efficiency. This process is driven by strikingly positive chemotaxis and depends on the melanoma reaching a critical density to generate the gradient. The principal attractant is the inflammatory signal lysophosphatidic acid (LPA). Unexpectedly, it is active across all stages of melanoma evolution and LPA is both necessary and sufficient for chemotaxis in 2D & 3D assays. Growth Factors were previously considered to play essential roles in driving directed migration, but instead facilitate LPA chemotaxis. Sampling across the margins of melanomas in vivo, gradients of LPA are reliably identified, which are capable of driving accurate chemotaxis. This not only confirms the physiological importance of the results, but also is the first time a chemoattractant gradient has been measured in vivo. The corollary of these findings is that, provided with an external homogenous source of LPA, a large enough melanoma will degrade the local LPA to generate an outward gradient. Therefore it is the ability to degrade the gradient that acts as the signal to drive chemotaxis and invasion rather than the presence of LPA per se. In the chambers, cells are observed dispersing in a wave and in addition to driving efficient melanoma invasion, this may be responsible for the patterning of melanocytes across the skin during embryogenesis. Ultimately, identifying key aspects of and targeting the LPA-axis may prove a novel prognostic indicator and therapeutic target for invasion and metastasis.

Human cancer stem cells are proposed to play a critical role in tumour initiation and maintenance by their exclusive ability to regenerate the tumour. Thus cancer stem cells share many of the properties of normal stem cell including self-renewal and ability to give rise to progeny which undergo tissue-specific differentiation. Thus we hypothesised that by determining the normal patterns of tissue differentiation within cancer we could identify tumour type specific factors that promote differentiation, for therapeutic development. Therefore the aim of this study is to define patterns of human hair follicle differentiation in human basal cell carcinoma (BCC) in order to elucidate potential drug-able targets that can promote tumour specific differentiation. To test this hypothesis we analysed 20 different hair follicle specific differentiation markers, which define distinct layers within the normal adult hair, in six different human BCC samples using RT-PCR with normal hair follicle tissue as control. For the 12 specific keratin genes expressed in the BCC, we analysed expression by immunofluorescence on 20 different BCC samples, using hair follicle samples as positive controls. Our findings suggest that human BCC demonstrates both inward and upward differentiation patterns similar to the hair follicle, with expression of: outer root sheath (K5,14,16,and k17), companion layer (K75), inner root sheath (K26,27,28,71,72,and k74), and cuticle (K32,35,82,and k85); but not hair shaft (K31) markers. Consistent with these findings we observed the mutually exclusive relationship between expression of the early differentiation marker K19 and cell proliferation in the hair follicle and BCC. Similarly, expression of the outer root sheath keratins coincided with nuclear translocation of both GLI1 and NFIL-6, suggesting that BCC also share normal hair follicle tissue regulatory pathways. To further test the hypothesis that normal tissue factors observed in the hair follicle regulate BCC differentiation we have developed an in vitro BCC assay. Using this tissue culture model we hypothesised that BCC’s are stuck in the telogen part of the hair follicle cycle, resulting from autocrine expression of bone morphogenic proteins 2 and 4. Inhibition of BMP signalling by addition of noggin as well as addition of TGF-β to BCC colonies in tissue culture led to further induction of inner root sheath, cuticle and medulla keratins. In summary we have shown that BCC exhibit hair follicle differentiation, which is similarly regulated, but is stuck in telogen arrest and can be rescued by addition of noggin and TGF- β2.

This thesis considers three complications that arise from applying reinforcement learning to a real-world application. In the process of using reinforcement learning to build an adaptive electronic market-maker, we find the sparsity of data, the partial observability of the domain, and the multiple objectives of the agent to cause serious problems for existing reinforcement learning algorithms. We employ importance sampling (likelihood ratios) to achieve good performance in partially observable Markov decision processes with few data. Our importance sampling estimator requires no knowledge about the environment and places few restrictions on the method of collecting data. It can be used efficiently with reactive controllers, finite-state controllers, or policies with function approximation. We present theoretical analyses of the estimator and incorporate it into a reinforcement learning algorithm. Additionally, this method provides a complete return surface which can be used to balance multiple objectives dynamically. We demonstrate the need for multiple goals in a variety of applications and natural solutions based on our sampling method. The thesis concludes with example results from employing our algorithm to the domain of automated electronic market-making.

Introduction: Malignant Melanoma is the most lethal of the skin cancers and the UK incidence is rising faster than that of any other cancer. Breslow thickness remains the best predictor of metastasis and Sentinel Lymph Node Biopsy is the only method of detecting nodal spread in clinically node negative patients. Surgery is the only effective therapy. Angiogenesis – the growth of new vessels from pre-existing vasculature - is an absolute requirement for tumour survival and progression beyond a few hundred microns in diameter. Anti- angiogenic isoforms of VEGF have been demonstrated previously to be protective with regard to metastasis. The aims of this thesis were to determine whether VEGF expression within the tumour may allow prediction of the nodal status. Furthermore another aim was to determine whether via the “Seed and Soil” theory, by examination of angiogenic and lymphangiogenic profiles of the tumour and node can we determine that the tumour may control the microenvironment around the Sentinel Node? Finally, as a cohort of false negative patients emerged with a higher mortality rate than their true negative and true positive patient cohort counterparts, could any further patterns be established by performing the same experiments on these patients? Methods: Archived human tumour and corresponding Sentinel Node samples were used and immunohistochemistry was used to investigate the role of pro and anti angiogenic isoforms of VEGF, VEGF-C, LYVE-1 and CD31 within these patients. Results: VEGF-C expression was significantly increased in the intranodal component of positive Sentinel Lymph Nodes (p < 0.01 Bonferroni). This increased expression appeared to be independent of tumoural influences and no strong evidence for the “Seed and Soil” theory was proved. A significantly higher number of lymphatic vessel counts were identified within node negative patients (p < 0.05 ANOVA). No further significant findings were defined on examination of the false negative cohort of patients. Conclusions: This study has shown that positive Sentinel Lymph Nodes exhibit high levels of intranodal VEGF-C. This expression does not appear to be related to tumoural influences. It would therefore appear that VEGF-C expression within Sentinel Nodes warrants further investigation and may aid diagnosis of spread or represent a target to slow or even prevent the onset of metastasis.

Leishmania parasites are digenetic protozoans which infect human hosts and are causative agents of a series of diseases known under the name of leishmaniasis. Macrophages represent the main host. Hence the interaction between Leishmania and macrophages is a fundamental step in the development of the disease. Many studies have been undertaken to understand early stages of the parasite interaction with macrophages; however, few have investigated the later stages of infection. This study was undertaken to develop an experimental model to examine the fate of the parasites when they leave the safe environment represented by their host macrophage. Primarily, the study investigated how Leishmania spread to neighbouring cells without being recognized and killed by the immune system defences. Three Old World species of Leishmania parasites: L. aethiopica, L. major and L. tropica, all responsible for the cutaneous form of the disease, were used. A model of infection was described using two cell lines well known for supporting infection: THP-1 and U937. Axenic amastogotes for L. aethiopica parasites were obtained and used to identify drugs active against the infection. On the basis of the information available in the literature, a model was suggested involving interaction of intracellular parasites with the host cells’ apoptotic machinery. Specifically it was suggested that Leishmania parasites were able to induce incomplete activation of apoptosis in the host cells. This hypothesis was confirmed by the findings that during infection an increased number of host cells showed two features associated with early apoptosis but not the one associated with the later stage. Results were validated in peripheral blood derived human macrophages. The information obtained from comparative proteomics analysis of the infection confirmed that Leishmania regulates apoptotic processes. On the basis of the results obtained a model was presented to explain how induction of apoptosis allows intracellular amastigotes to spread unrecognised to uninfected macrophages without inducing an inflammatory response or losing the host cell’s protection.

Methicillin-resistant Staphylococcus aureus (MRSA) and Pseudomonas aeruginosa are the main multi-drug resistant pathogens associated with deep wound infections which then may cause septicaemia. Treatment is problematic and re-infection is quite common. Free Living Amoebae (FLA), such as Acanthamoeba, are widely distributed in the environment and may also contaminate wounds. It is well known that Acanthamoeba feed on and protect bacteria. The role of Acanthamoeba in wound infections is not very well understood. It is possible that the presence of Acanthamoeba in wounds is one of the key factors for such re-infections. In this study, 140 wound swabs were collected to check for the presence of Acanthamoeba spp. Only one sample was positive for Acanthamoeba spp. Sequencing of the highly variable DF3 region of 18S rRNA gene for the sample showed that this isolate belongs to genotype T4. In addition, clinical isolates of MRSA and Pseudomonas from wound infections were used in this study. The results showed that MRSA and Pseudomonas were able to bind with, invade, survive and multiply inside Acanthamoeba species. One of the essential compounds for microorganisms to grow is iron. The role of iron chelators, including deferiprone and selected novel compounds based on hydroxyl pyridine moiety, was studied. Findings revealed that all novel iron chelators have an antimicrobial activity against both bacteria. In addition, all novel iron chelators were able to kill Acanthamoeba. Cytotoxic effects of MRSA, P. aeruginosa and Acanthamoeba were investigated using the KB epithelial cell line and mesenchymal stem cells (MSC) using a general caspase inhibitor. The results revealed that the ability of live bacteria to induce cell death was higher compared with heat-killed bacteria, bacteria conditioned medium (BCM) and Acanthamoeba conditioned media (CM). The exact trigger for the cell death in this study was not investigated but the relative contributions of apoptosis and necrosis were investigated using fluorescent technique, caspase inhibition and LDH assay. In conclusion, presence of Acanthamoeba in wounds could be the reason of prolong treatment and reinfection in wounds.