Foundations of Health Information Engineering and Systems

This paper presents an overview of smart medication dispensing and administration devices and software tools designed to minimize dispensing and administration errors. Some of them are for users who take medications on a long term basis without close professional supervision; others are for pharmacy and nursing staffs in hospitals, long term care, and assisted living facilities. These tools should be configurable, customizable, easy to use and effective for diverse users and care-providing institutions. The paper describes approaches taken to meet these objectives.

The western world enjoys a far greater degree of sophistication in computer technology than the developing world; challenges include the fact that infrastructure and resources in the developing world lag far behind; education to use and maintain the technology is often lacking; and cultural and societal issues prohibit its use. Questions include: What is the role of the first world in promoting more widespread use of technology for health informatics? What are the advantages to the first world in promoting more widespread use? What is the role of scientists in the first world in developing technology suitable for less technologically savvy regions? Should we interfere in cultures that do not use technology as we do? What can we practically expect to positively influence by engaging with developing world? How do we engage with the developing world in a sustainable way? Finally, modern vaccines have made huge differences in health the world over; can we expect similar positive transformation from increased computer/information technology in the developing world?

Connected Health involves the use of ICT to improve healthcare quality and outcomes. In a connected heath environment, stakeholders can struggle to make best use of this information coming from a variety of sources. Given this, we are investigating the challenge of how to use available information to make informed decisions about the care pathway which the patient should follow to ensure that prevention and treatment services are efficient and effective. In this paper, we outline our research into care pathway and information modelling in a Connected Health setting. The research is currently underway, and follows a series of stages using sources in industry and academia. In this paper, we present an overview of the project work packages including an explanation on how the different stages of the research form a continuum in which the developed models will be continually adjusted. We describe how empirical evidence will be used in the development of the models through following an evolutionary multi-method research approach.

Attempts to extend process management to support pathways in the health domain have not been as successful as workflow management for routine business processes. In part this is due to the dynamic nature of knowledge-intensive work such as care pathways: the actions performed change continuously in response to the knowledge developed by those actions. Also, care pathways involve significant informal communications between those involved in caring for the patient and between these carers and the patient / patient family which are difficult to capture. We propose using an approach to supporting care pathways that embraces these difficulties. Rather than attempting to capture every nuance of individual activities, we seek to facilitate communication and coordination among knowledge workers to disseminate knowledge and pathway expertise throughout the organization.

The efficient use of health care ressources requires the use of Information and Communication Technology (ICT). During a treatment process, patients have often been tested and partially treated with different diagnoses in mind before the precise diagnosis is identified. To use ressources well it becomes necessary to adapt the prescribed treatments to make use of the tests and partial treatments already performed, rather than always starting from square one. We propose to facilitate this through the design of declarative process models accounting for the involvement of distributed groups of medical specialists and the adaptation of treatments, and through the evaluation of the trustworthiness of models taking account of test results and actual treatments compared to the clinical guidelines.

This position paper presents research work involving the development of a publicly available Realistic Synthetic Electronic Healthcare Record (RS-EHR). The paper presents PADARSER, a novel approach in which the real Electronic Healthcare Record (EHR) and neither authorization nor anonymisation are required in generating the synthetic EHR data sets. The GRiSER method is presented for use in PADARSER to allow the RS-EHR to be synthesized for statistically significant localised synthetic patients with statistically prevalent medical conditions based upon information found from publicly available data sources. In treating the synthetic patient within the GRiSER method, clinical workflow or careflows (Cfs) are derived from Clinical Practice Guidelines (CPGs) and the standard local practices of clinicians. The Cfs generated are used together with health statistics, CPGs, medical coding and terminology systems to generate coded synthetic RS-EHR entries from statistically significant observations, treatments, tests, and procedures. The RS-EHR is thus populated with a complete medical history describing the resulting events from treating the medical conditions. The strength of the PADARSER approach is its use of publicly available information. The strengths of the GRiSER method are that (1) it does not require the use of the real EHR for generating the coded RS-EHR entries; and (2) the generic components for obtaining careflow from CPGs and for generating coded RS-EHR entries are applicable in other areas such as knowledge transfer and EHR user interfaces respectively.

The insulin pump is a safety-critical embedded medical device used for treatment of type 1 and insulin treated type 2 diabetes. Malfunction of the insulin pump will endanger the user’s life. All countries impose some regulation on the sale and use of medical devices. The purpose of such regulation is to protect the public by imposing standards of safety for medical devices, including insulin pumps. The regulator in the USA, the USA Food and Drug Administration (FDA), actually goes further, and includes efficacy in the regulatory requirement. Until recently, regulatory approval was dependent on process based guidance. However, this has proven to be inadequate in some (most) cases where the device depends on software for its safe and effective operation, and the FDA recently changed its approval process for infusion pumps (including insulin pumps), so that the production of an assurance case that demonstrates that the device is safe and effective is now a strongly suggested regulatory requirement. However the current regulatory guidance does not recommend any particular software development methodology, and does not include definitive guidance on the evaluation component of the certification process. In this paper, we briefly review the related USA regulatory standards for insulin pumps, highlight development and certification challenges, briefly discuss attributes of a safe, secure and dependable insulin pump, and propose an effective certification process for insulin pumps.

eHealth governance and regulation are necessary in low resource African countries to ensure effective and equitable use of health information technology and to realize national eHealth goals such as interoperability, adoption of standards and data integration. eHealth regulatory frameworks are under-developed in low resource settings, which hampers the progression towards coherent and effective national health information systems. Ontologies have the potential to clarify issues around interoperability and the effectiveness of different standards to deal with different aspects of interoperability. Ontologies can facilitate drafting, reusing, implementing and compliance testing of eHealth regulations. In this regard, we have developed an OWL ontology to capture key concepts and relations concerning interoperability and standards. The ontology includes an operational definition for interoperability and is an initial step towards the development of a knowledge representation modeling platform for eHealth regulation and governance.

Multilevel modeling has been proven in software as a viable solution for semantic interoperability, without imposing any specific programming languages or persistence models. The Multilevel Healthcare Information Modeling (MLHIM) specifications have adopted the XML Schema Definition 1.1 as the basis for its reference implementation, since XML technologies are consistent across all platforms and operating systems, with tools available for all mainstream programming languages. In MLHIM, the healthcare knowledge representation is defined by the Domain Model, expressed as Concept Constraint Definitions (CCDs), which provide the semantic interpretation of the objects persisted according to the generic Reference Model classes. This paper reports the implementation of the MLHIM Reference Model in XML Schema Definition language version 1.1 as well as a set of examples of CCDs generated from the National Cancer Institute – Common Data Elements (NCI CDE) repository. The set of CCDs was the base for the simulation of semantically coherent data instances, according to independent XML validators, persisted on an eXistDB database. This paper shows the feasibility of adopting XML technologies for the achievement of semantic interoperability in real healthcare scenarios, by providing application developers with a significant amount of industry experience and a wide array of tools through XML technologies.

Deterioration in patient condition is often preceded by deterioration in the patient’s vital signs. “Track-and-Trigger” systems have been adopted in many hospitals in the UK, where manual observations of the vital signs are scored according to their deviation from “normal” limits. If the score exceeds a threshold, the patient is reviewed. However, such scoring systems are typically heuristic. We propose an automated method for detection of deterioration using manual observations of the vital signs, based om Bayesian model averaging. The proposed method is compared with an existing technique - Parzen windows. The proposed method is shown to generate alerts for 79% of patients who went on to an emergency ICU admission and in 2% of patients who did not have an adverse event, as compared to 86% and 25% by the Parzen windows technique, reflecting that the proposed method has a 23% lower false alert rate than that of the existing technique.

Acute hospital wards in the UK are required to use Early Warning Scoring (EWS) systems to monitor patients’ vital-signs. These are often paper-based, and involve the use of heuristics to score the vital signs which are measured every 2-4 hours by nursing staff. If these scores exceed pre-defined thresholds, the patient is deemed to be at risk of deterioration. In this paper we compare the performance of EWS systems, that use different approaches to score abnormal vital-signs, to identify acutely-ill patients, while attending the Emergency Department (ED). We incorporate the use of data acquired from bed-side monitors into the EWS system, thereby offering the possibility of performing patient observations automatically, between manual observations.

In the medical domain there is a tension between the requested speed of innovation and the time needed to deliver a certifiable system. To ensure the required safety, usually a long test and integration phase is needed. To shorten this phase and to avoid late bug fixing, the aim is to detect faults (if any) much earlier in the development process. This can be achieved by combining a number of model-based techniques such as (1) architecture validation by simulating executable models, (2) development of a Domain-Specific Language (DSL) to combine precision with higher levels of abstraction, and (3) transformations from DSLs to analysis models for performance evaluation and formal verification. We illustrate such techniques using an industrial study project on a new architecture for movement control including collision prevention.

This position paper gives an overview on the OR.NET project which focuses on the dynamic and safe interconnection of medical devices in an operating room. A brief overview of the legal situation for the approval of medical devices is given to highlight today’s limitations of the dynamic interconnection of safety critical devices in hospitals. A collection of methods equipped with a methodology is presented and discussed, which is intended to replace current integration tests at runtime.

The on-demand approach, where systems are assembled from components by lay users, has seen success in the consumer electronics industry. Currently, there is growing demand for on-demand capabilities in medical systems so caregivers can create larger medical systems from smaller medical devices. Unlike consumer electronics, medical systems pose challenges for the on-demand approach due to attributes such as device complexity, device variability and safety requirements. In this paper, we propose a formal specification language for on-demand (medical) systems. Our approach is based on the formalism of Modal I/O Automata, which allows system designers to express complex device requirements and can be used to reason about safety and liveness properties of ondemand medical systems directly from their specifications. We illustrate the applicability of our approach through a case study of a closed-loop patient controlled analgesia system.

Feature-orientation has proven beneficial in the development of software product lines. We investigate formal safety analysis and verification for product lines of software-intensive embedded systems. We show how to uniformly augment a feature-oriented, model-based design approach with the specification of safety requirements, failure models and fault injection. Therefore we analyze system hazards and identify the causes, i.e. failures and inadequate control systematically.

As features are the main concept of functional decomposition in the product line approach, features also direct the safety analysis and the specification of system level safety requirements: Safety (design) constraints are allocated to features. Subsequently, the behavior including possible faults is formally modeled. Then formal verification techniques are employed in order to prove that the safety constraints are satisfied and the system level hazards are prevented. We demonstrate our method using Scade Suite for the model-based product line design of cardiac pacemakers. VIATRA is employed for the model graph transformation generating the individual products. Formal safety analysis is performed by using Scade Design Verifier. The case study shows that our approach leads to a fine-grained safety analysis and is capable of uncovering unwanted feature interactions.

The development of drug resistance is a major factor impeding the efficacy of antiretroviral treatment of South Africa’s HIV infected population. While genotype resistance testing is the standard method to determine resistance, access to these tests is limited in low-resource settings. In this paper we investigate machine learning techniques for drug resistance prediction from routine treatment and laboratory data to help clinicians select patients for confirmatory genotype testing. The techniques, including binary relevance, HOMER, MLkNN, predictive clustering trees (PCT), RAkEL and ensemble of classifier chains were tested on a dataset of 252 medical records of patients enrolled in an HIV treatment failure clinic in rural KwaZulu-Natal in South Africa. The PCT method performed best with a discriminant power of 1.56 for two drugs, above 1.0 for three others and a mean true positive rate of 0.68. These methods show potential for application where access to genotyping is limited.

This paper is part of work aimed at investigating an approach to knowledge incorporation into solution models of the Meal Planning Problem (MPP) for use in mobile web-based HIV/AIDS nutrition therapy management within the context of developing countries, particularly, in Sub-Saharan Africa. This paper presents a characterisation of the incorporation of knowledge into the models for the MPP. The characterisation is important for assessing the extent to which MPP models can be adapted for use in different clinical problems with different nutrition guideline knowledge and in different regions of the world with differently customised versions of the guidelines. The characterisation was applied to thirty one works in the literature on MPP models. The main outcome of the application of the characterisation was the finding that the existing MPP models do not provide for the incorporation of nutrition guideline knowledge as first class concepts with identifiable and manageable structures, which makes almost impossible the transfer of knowledge from health experts to patients and from one region of the world to another.

Recent developments in data de-identification technologies offer sophisticated solutions to protect medical data when, especially the data is to be provided for secondary purposes such as clinical or biomedical research. So as to determine to what degree an approach– along with its tool– is usable and effective, this paper takes into consideration a number of de-identification tools that aim at reducing the re-identification risk for the published medical data, yet preserving its statistical meanings. We therefore evaluate the residual risk of re-identification by conducting an experimental evaluation of the most stable research-based tools, as applied to our Electronic Health Records (EHRs) database, to assess which tool exhibits better performance with different quasi-identifiers. Our evaluation criteria are quantitative as opposed to other descriptive and qualitative assessments. We notice that on comparing individual disclosure risk and information loss of each published data, the μ-Argus tool performs better. Also, the generalization method is considerably better than the suppression method in terms of reducing risk and avoiding information loss. We also find that sdcMicro has the best scalability among its counterparts, as has been observed experimentally on a virtual data consisted of 33 variables and 10,000 records.