This thesis is concerned with understanding how Electronic Patient Record systems (EPRs) are being problematised and enacted in NHS acute hospital Trusts following the dismantling of the National Programme for IT (NPfIT). The term EPR is widely used but has no uniform definition and has been applied to a wide range of systems. In policy, EPRs have often been envisaged as integrated, large-scale systems capable of transforming how healthcare is delivered. However, in practice such systems have proved difficult to achieve and many of the EPRs to be found in hospitals are much smaller standalone specialist departmental systems. Working with Actor-Network Theory (ANT), this thesis takes a sociotechnical approach in which differences in EPRs are understood as matters of ontology (variations in their enactment by multiple heterogeneous actors producing multiple, sometimes conflicting, realities) rather than matters of epistemology (differences in the interpretation of a single underlying reality).

There is now increasing commitment internationally to using electronic healthcare records collected during routine care delivery to conduct clinical research. This must be rigorously controlled by an extensive set of information governance requirements defining the legal, ethical and practical guidelines to respect the privacy rights of the people about whom the records are kept, uphold the clinical profession’s duty of confidentiality and protect the interests of participants, practitioners and researchers. The development of information security policies is a highly regarded method of meeting these requirements. This is hampered by the need to interpret a complex framework of legislation and guidelines, lack of clear advice and inconsistency in authoring, interpretation and understanding amongst the people whose behaviour they are expected to guide. By using the results of several UK and European research and information platform development projects in which the author has participated and by gathering requirements from stakeholders in the clinical and research communities, this thesis defines a knowledge management representation to specify policy requirements in a computable form. The work provides the first set of knowledge requirements for governing research uses of electronic healthcare records, and a knowledge model that describes information security policies and generates a web application tool. The tool allows policy control authoring that provides a consistent, clear and unambiguous view of governance requirements to researchers and service providers. The model and tool have been evaluated in a laboratory setting to explore their effects on behaviour and understanding of invited participants when authoring policy about handling healthcare records in research and making decisions about sharing information. The work has resulted in a validation of the model and demonstrated the potential positive effects of this new approach on practice. It makes recommendations about how it should be used in working practice and for educating people about information governance when performing clinical research to improve care provision.

A systematic approach to the extraction of data from disparate data sources is proposed. The World Wide Web is a most diverse dataset; identifying ways in which this large database provides means for data quality verification with concepts such as data lineage and provenance allows to follow the same approach as a means to aid decision-making in sensitive domains such as healthcare. Through lessons learned from research in the UK and internationally, we conclude that emphasis on interoperable and model-based support of the data syndication can enhance data quality, an issue still current (American Hospital Association, 2015) and with data barriers in healthcare due to governance concerns. To improve on the above, we start by proposing a system for solution-orientated reporting of errors associated with the extraction of routinely collected clinical data. We then explore key concepts to assess the readiness of data for research and define an ontology-driven approach to create data dictionaries for quality improvement in healthcare. Finally, we apply this research to facilitate the enablement of consistent data recording across a health system to allow for service quality comparisons. Work deriving from this research and built by the author commissioned and aided by the UK NHS, University of Surrey, Green Cross Medical, particularly in creating and testing software systems in real-world scenarios, has facilitated: quality improvement in healthcare data extraction from GP practices in the UK, a state-of-art system for Web-enabling Hospital Episode Statistics (HES) data for dermatology and, finally, an online system designed to enable cancer Multi-Disciplinary Teams (MDTs) to self-assess and receive feedback on how their team performs against the standards set out in ‘The Characteristics of an Effective MDT’ provided by NHS IQ, formerly part of National Cancer Action Team (NCAT), which in 2016 won the Quality in Care Programme’s “Digital Innovation in the Treatment of Cancer” award. Further experimentation shows there is potential for the methods proposed to be applicable in other sectors such as the investment sector (initial investigation has happened through the early stages of this research) but it is suggested that this potential be explored further.

The purpose of this study is to investigate how emotionally enabled virtual agents (VAs) in healthcare provision training simulations allow for a more effective level of understanding on how an emotionally enhanced scenario can affect different aspects of learning. This is achieved by developing virtual agents that respond to the user’s emotions and personality. The developed system also provides visual and auditory representations of the virtual agents’ state of mind. To enable the fulfilment of this purpose an experimental framework for incorporating emotional enhancements (concentrating on negative emotions such as stress, fear, and anxiety) into virtual agents in virtual training applications for healthcare provision is designed and implemented. The framework for incorporating emotional enhancements is designed based on previous research, on psychological theories (with input by experienced psychologists) and from input of experts in the area of healthcare provision. For testing the framework and answering the research question of this thesis the researcher conducted nine case studies. The participants were nursing students in the area of healthcare provision, and more specifically in the area of mental health, specialising in caring for patients with dementia. The results of the study showed that the framework and its implementation succeeded in providing a realistic learning experience, stimulated a better set of responses from the user, improved their level of understanding on how an emotionally enhanced scenario can affect the learning experience and helped them become more empathetic towards the person they cared for.

Background and aim: Hospitals in the developed world are increasingly adopting digital systems such as electronic health records (EHRs) for all kinds of documentation. This move means that traditional paper case notes and nursing records are often documented in EHRs. Documentation of vital signs is important for monitoring a patient's physiological condition and how vital signs are presented in a clinical record can have a profound impact on the ability of clinicians to recognise changes, such as deterioration in a patient's condition. Vital signs have received minimal attention with regard to how they are documented in EHRs which suggests that there is an urgent need for this to be examined. Design, methodology and approach: A mixed methods study was conducted in a 372-bed county hospital in two phases. Phase one was a quantitative study, and was followed by a qualitative study in phase two. The aim of the quantitative study was to examine the vital signs documented in the electronic health records of patients who had previously suffered a cardiac arrest. The aim of the qualitative study was to investigate how medical and nursing staff measured, reported and retrieved information on vital signs. Observations were made and interviews were conducted in four clinical areas. Findings: The quantitative study found that documentation of vital signs was incomplete in relation to current universal standards for monitoring vital signs, and that vital signs were dispersed inconsistently throughout the EHR. The qualitative study provided a detailed understanding of the routines and practices for monitoring vital signs and demonstrated variation in routines and in methods of documentation in the four clinical areas. Documenting and retrieving vital signs in the EHR was problematic because of usability issues and led to workflow problems. Workflow problems were solved at ward level by the creation of paper workarounds. Contribution to knowledge: This thesis has shown that poor facilities for the documentation of vital signs in EHRs could have a negative impact on patient safety because it reduces the possibility of good record keeping. This leads to limited availability of easily accessible, up-to-date information, essential for identifying clinical deterioration and, thus, is a challenge to patient safety. Related to this, the thesis has identified possible solutions to usability problems in the EHR. Inconsistent routines and practices were also identified and suggestions were made for how this problem might be approached.

With improvements in living standards and innovations in medical care, life expectancy has increased. However, although people are living longer, particularly in developed countries, they are not necessarily healthier during the additional years of life, with a rising number of people with long-term physical and mental health conditions that require supported living, for example, within a care home or hospital environment. In response to the rising economic costs of managing long term conditions, successive governments have developed policies to reduce the use of institutional environments (e.g., care homes) and of unplanned hospital admissions, and are encouraging the development of systems which aim to monitor, support and manage people’s health in their own home. These developments have lead to increased research on using remotely monitored, sensor-based technologies to provide relatives, carers and health care professionals with timely data about the well-being of older people living independently, and so provide timely and appropriate support effectively, thus helping them remain in their own homes, especially when they have long-term health problems. The aim of the research described in this thesis was to investigate the use of an electricity monitor to recognise and monitor changes in resident’s daily activities. This was achieved using two phases; the first conducted a survey to gather information about which activities and features that carers and relatives would like to have access to, so as to be reassured about their relative’s health and well being. The second phase collected and analysed electricity consumption data from four households for a one-week period, to develop models to identify when specific activities had been undertaken, e.g., using the shower, using a kettle. This research concluded that the monitoring of general and some specific activities is important to the relatives and carers, although the best form of reassurance about their relative’s situation was felt to be human contact. Following the analysis of the electricity consumption data, it was concluded that while it is possible to recognise appliance usage from whole house electricity consumption data, the variability and lack of transferability between houses and appliances would mean that the large-scale use of this type of monitoring would require considerable further development.

No description available.

The development of biomedical research and fast point-of-care diagnostics require large-scale sensing equipment and smart portable healthcare devices. Therefore, integrating chemical sensors into solid state platforms becomes the most popular solution in modern chemical sensing applications. As a result, the continuous trend of scaling the transistor in semiconductor engineering and sensors' feature size in biomedical or biochemical areas, converge into the concept of Lab-On-Chip (LOC). By combining LOC and the well developed fabrication process, Complementary Metal Oxide Semiconductor transistor (CMOS), a high level integration incorporating sensors and processing circuitry can be realized with minimal fabrication costs and convenient data processing ability. This work focuses on the engineering chemical sensing systems based on the CMOS ISFET, which provides high scalability and integration ability. An extended model for CMOS ISFETs is proposed to create an accurate model for robust sensors design. The origins of threshold variation and transconductance reduction are explained in detail by using this model. A complete study on the electrolyte-insulator interface across the sensing membrane is provided to qualitatively explain the non-ideal effects such as drift and noise. Based on the study of both electronic and chemical sides, a design strategy is presented and indicates that large sensors are better for accurate measurements and small sensors are suitable for large-scal parallel sensing. Using this knowledge, we investigate the interface circuit with capabilities to reduce the non-linear effects of ISFETs. To reduce the trapped charge effect in the device, an auto-offset-removal approach is presented and demonstrated in complimentary sensing pairs based on autozeroing techniques. The trapped charge effects and the transistor low frequency noise are attenuated to provide a larger dynamic range. Moreover, by using pH-to-Time sensors and Time-to-Digital converters, a highly compact readout scheme is designed with a minimized analogue biasing and processing. This scheme shows a great potential in large scale sensing platforms, which require low power consumption and high sensors density. Finally, a novel system, which feedbacks the electrical signal into the chemical environment, is developed to reduce the non-ideal characteristics. By incorporating the principle of a Sigma- Delta modulator, we develop the chemical sigma-delta modulator. By using the ISFETs as the quantizer, a titrator as the feedback delta modulator, and the chemical diffusion characteristics as the Sigma modulator, a chemical noise shaping is realized to minimize the drift and low frequency noise.

Background: Adverse drug reactions are responsible for a significant proportion of hospitalisations. This PhD aimed to develop and optimise methods for detecting serious adverse drug reactions in databases of electronic health records for use in pharmacoepidemiology and genetic epidemiology, with a focus on cholestatic liver injury. Methods: A systematic review was performed before developing a multiple database source (“multisource”) algorithm for identifying cholestatic liver injury. Multisource algorithm case status was used to guide the development of another algorithm using data from a standard UK Clinical Practice Research Datalink (CPRD) record only (the CPRD algorithm). Testing of the CPRD algorithm was performed within a cohort analysis of an established cause of the injury (flucloxacillin), before carrying out a casecontrol study investigating a number of putative associations (drug exposures carbamazepine, celecoxib, duloxetine, ramipril and risperidone). Results: The majority of reviewed studies lacked a reproducible case definition, and case assignment generally required information external to database records. Secondary care (HES) data provided little additional information than that found in primary care (CPRD), meaning that the CPRD algorithm had a very good ability to discriminate between multisource algorithm cases statuses (ROC area under the curve 0.95). The flucloxacillin 45-day risk estimate obtained from the cohort study using the highest specificity CPRD algorithm (6.15 per 100 000 users, 95% CI 4.61 – 8.04) was very similar to previous studies. Celecoxib and risperidone were associated with cholestatic liver injury (celecoxib multivariablemultivariable RR recent vs. current users low specificity CPRD algorithm 1.89, 95% CI 1.11 – 3.22, risperidone multivariablemultivariable RR high specificity CPRD algorithm 2.59, 95% CI 1.41 – 4.75). Conclusions: The CPRD algorithm detected similar flucloxacillin effects as (1) the multisource algorithm and (2) previous studies. Associations with risperidone and celecoxib were also detected. Algorithm characteristics that could facilitate (1) pharmacovigilance and (2) recruitment to genetic association studies include the ability to (a) detect cases without using information external to the EHR and (b) apply varying levels of specificity and sensitivity.

The extensive use of the Internet has been accompanied by the augmentation of e-services, such as e-health. Particularly, the improvement in e-health has put a massive load of sensitive information in the hands of service providers and other parties, where privacy risks might exist when accessing sensitive data stored in the form of electronic patient records (EPRs). EPRs support efficient access to patient data by multiple healthcare providers and third party users, which will consequently improve patient care. However, the sensitive nature of this data requires access restrictions to only those 'who need to know'. How to achieve this without compromising patient privacy remains an open issue that needs further consideration. This thesis, therefore, addresses privacy problems with distributed EPRs and how to allow authorised users to access them with multiple levels of identity privacy preservations. The thesis investigates existing security solutions for achieving privacy preserving distributed data access and analyses their strengths and weaknesses. It then proposes a novel method to support secure access to distributed EPRs with three levels of patient identity privacy preservations, i.e., the 3LI2P version 1 (3LI2Pv1) method. The idea of the method is to integrate a number of significant features, which have not been considered in the related work, and these features are: (1) supporting three levels of controlled distributed EPR accesses by different legitimate user groups while preserving patient identity privacy; (2) making use of different digital credentials to support the three levels of access; (3) simplifying key management distribution; (4) optimising performance; and(5) supporting separation of duties among trusted third parties, ensuring accountability. The 3LI2Pv1 method makes use of three layers of pseudonyms to achieve these properties, i.e., each patient has multiple pseudonyms layered at three levels. The method relies on a combined cryptographic primitives, symmetric cryptosystem, asymmetric cryptosystem and a hash function, to generate these pseudonyms. The security properties and the performance of the 3LI2Pv1 method are analysed, evaluated and compared with related work. The results from the comparison show that our 3LI2Pv1 method is better in terms of supporting the requirements necessary to preserve a patient's identity privacy in a distributed setting at no significant additional costs. The thesis also proposes an enhanced version of the above method called the 3LI2P version 2 (3LI2Pv2) method. This latter method enhances the 3LI2Pv1 method in terms of reducing key management burden on central trusted third party, enforcing the least access privilege principle, not only among users and central trusted third party, but also among health service providers who manage the patients' data, further improving performance, ensuring the integrity of patient pseudonyms, providing pseudonyms uniqueness and finally, facilitating a more ne-grained access control by introducing an additional linkable anonymousa ccess sub-level. The 3LI2Pv2 method has been analysed in terms of security and performance. Based on the 3LI2Pv2 method, the thesis introduces a novel 3LI2Pv2 protocol. The protocol is designed specifically for the 3LI2Pv2 method to facilitate different types of accesses, linkable access, Level-2 inter-HSP linkable anonymous access, Level-2 intra-HSP linkable anonymous access and anonymous access, and to allow different user groups to securely access distributed EPRs according to their privileges, without compromising the patient's privacy. The security properties of the 3LI2Pv2 protocol are formally verified using the Casper/FDR2 verification tool. To evaluate its performance, a prototype of the 3LI2Pv2 protocol has been implemented using Java under two different settings, a single machine and distributed machines settings. Using these implementation settings, performance evaluations of the protocol were conducted. The results from the evaluations (under both settings) confirmed that we have successfully balanced between security and performance without compromising the patient's privacy.