Ebook: Knowledge and Decisions in Health Telematics
Medical doctors are used to rely on technical devices for decision support for diagnosis and choice of treatment. Thus, if a phonocardiogram reveals a heart murmur which the physician did not hear in the stethoscope, he is more likely to mistrust his ear rather than the recording.
Nevertheless, the same doctor will hesitate to use artificial intelligence or decision support from an expert system because of lack of confidence in its reliability except in very selected cases. He will claim that patients and diseases are so individual that a computer system is too crude an instrument to rely on, and right or wrong he might feel that a disagreement between his own judgement and that offered by the software might offer a legal problem, he would prefer to be without. Yet, the development of expert systems in medicine remains a real challenge and actually systems are being successfully used.
The study presented in this book is a successful effort to present a sober assessment of status quo in terms of success factors, market potential and first and foremost identification of the development work that needs to be done. It makes stimulating reading and will guide anyone fascinated by the field. Time is now ripe to really test systems in a variety of situations, also in terms of clinical outcome.
For health telematics it is essential that only useful developments be integrated into the networks services foreseen for the future. It is therefore essential to build on the experience compiled in the book and test systems in real user environments. Together with national programmes and projects, the European Union health telematics programme of the future should be designed to provide the needed testbeds.
Let us try to provide the doctor with a reliable hearing aid when he needs it.
Head of Unit
The complexity of medical knowledge has been steadily increasing, the cost of Healthcare services has surged over the years, and growing demands exists for assessing and improving the quality of the Healthcare services. Knowledge based decision support might provide considerable help in solving many of these problems. This report addresses the perspectives of knowledge based decision support applications within the health sector in the next five to ten years, from the complementary perspectives of users, industry and researchers.
The report identifies potential users of decision support in Healthcare, which include not only doctors, but also nurses, Healthcare administrators, as well as patients. It then addresses their requirements and expectations with respect to knowledge based decision support, and discusses the need to integrate such systems and techniques with the Healthcare environment and their routine practices. Some topics are discussed in more depth, namely the change in emphasis from diagnostic support into more general patient management support, the emergence of clinical guidelines developed for this purpose as well as the need to support their development and management, the integration with medical data sources (eg. electronic patient records) and knowledge sources (eg. literature and patient databases), and the need for adequate interfaces with the users and with other, local or remote, information systems.
From the industry perspective, the report discusses the value that knowledge based decision support systems may have for the users, seen as purchasers of these systems. It identifies the industrial sectors whose expertise should be combined to enable the development of commercial knowledge based systems, namely Healthcare information systems suppliers, medical equipment industries, scientific and education publishers, value-added network operators, and pharmaceutical companies. It identifies opportunities for these industries to cooperate in the development of more innovative knowledge based products to extend their traditional product lines. It finally addresses some aspects of engineering of knowledge based applications in Healthcare, to facilitate integration in current products, areas needing research for future products development and competing technologies that must be taken into account.
The report suggests that knowledge based systems research should focus in the problems of integration with the Healthcare environment and discusses priorities for research in the next decade. These include: efforts towards the standardisation of terminologies and ontologies and translation between existing standards; refocussing reasoning methods from pure diagnosis into broader planning, incorporating diagnosis, treatment and monitoring; support to the management and development of patient management guidelines; coordination of these research efforts with the development of structured patient records; research into advanced architectures that allow sharing and reuse of medical knowledge, to ease the problems of knowledge acquisition and validation; integration of knowledge based techniques with telematics and multimedia technology to provide adequate user interfaces to these patient records (with a view on their management) and other Healthcare services; and finally the need to develop metrics and methodologies to assess the value and impact of knowledge based decision support systems.
The main conclusions of these complementary perspectives are summarised in the final chapter which ends with a number of recommendations regarding the planning of future research and development of knowledge based systems and techniques in the Health sector for the next decade.
The lack of demonstrable success of computer based decision support technologies in health care requires a re-examination of the assumptions that support initiatives into these technologies. There are many ways in which one could explain away this lack of success, for example lack of technological maturity, lack of appropriate computer record infrastructures, or indeed professional resistance to novel technologies. However, one can also look at the fundamental assumptions that are taken as given when designing decision support systems. These assumptions about the problems that decision support systems should assist with, and the way that assistance should be provided are open to question. Much of the current work in decision support system research and development is based upon outdated views of the clinical workplace, and it is likely that the failure of the technology is in large part due to this. What is required is a principled re-examination of clinical practice, aimed at identifying the ways in which clinical workers should really be assisted. Until this is done, decision support technology will continue to be mismatched to the needs of health care workers.
The actual use of computer-based decision-support systems in routine practice has been disappointingly low. Although underlying reasons need to be studied, actual use is often a trailing indicator for success. One of the characteristics of our technological age seems to be that developments may take decades to mature followed by a rapid dissemination. In the area of computer-based patient records, for example, decades of research had little impact. In recent years, however, the Dutch general practitioners are rapidly introducing computer-based patient records in their practices. Although actual use may be a trailing indicator, paradigms underlying research in the area of computer-based decision support are changing. Two distinct lines of research seem to be emerging. First, those who develop systems for actual use increasingly focus on the environment and setting in which medical decisions are made. That is, the emphasis will shift from casting medical knowledge in some formalism to enabling physicians and other health-care professionals to deliver care, to communicate and to interact. In a second line of research, the notion of casting an individual's expertise in some formalism has been replaced by a focus on the nature of medical knowledge, and how to model that knowledge.
Medical knowledge systems are computer systems which contain medical knowledge and use it to reason about patients. Although the first was described over three decades ago, few have passed into routine use, compared to the large numbers of laboratory databases, CT scanners or computerised electrocardiogram (ECG) interpreters, which now process over 50 million ECGs per annum. This paper considers the lessons that we can learn from computerised ECG interpreters, and discusses six major issues: establishing closer dialogue between system developers and users, evaluating the impact of systems on users and their problems, integration with patient record systems, assembly of public, validated medical knowledge bases, reducing the emphasis on novel reasoning methods and considering the legal, regulatory and educational implications of this technology. Specific research topics are proposed under each of these headings.
Almost two decades' experience in building expert systems has shown that strong development methodologies are desirable if we are to ensure that systems are effective. Logic programming techniques combined with software engineering methods offer a potent discipline for ensuring that software is sound. However, if we are to achieve trust in any medical technology it is not enough to demonstrate soundness. We must also show, so far as is possible, that it is safe. Alongside the pursuit of rigorous software development techniques, therefore, we need to investigate general design and development methods that minimise the occurrence of hazards. We can adopt some general techniques from existing safety engineering practices, but AI also supplies some important new ideas for improving safety.
The clinical information environment is of a heterogeneous nature. A hypermedia information system is a means for integrating different types of data and knowledge a clinician needs for making his clinical decisions. An integrated multimedia system with various kinds of knowledge processing and decision support components as well as clinical registries is supposed to be the “intelligent” clinical assistant system of the future. Various knowledge based systems have to communicate with other (not necessarily knowledge based) components of a clinical information and communication system in a hypermedia network. A decisive step towards a realization of this vision is the progress in structuring and modelling medical knowledge. The success of introducing knowledge based products in everyday clinical practice depends on their degree of flexibility in use and integration in a clinicians work.
In addition to issues directly related to problem solving (expertise modelling, computational realisation), there are two areas which are critical for successful use of KBS technology in (medical) practice. The first area concerns the analysis of the organisation in which the KBS has to function and of the tasks in this organisation that can profit from KBS or other automated support. The second area concerns the analysis of the required interaction of a KBS with other agents, such as various types of users (expert and non-expert physicians, nurses, other health-care professionals) and other software systems. We argue that there is a need for a R&D effort towards a comprehensive methodology for medical KBS construction that covers all these areas. This ensures that KBS technology is applied for real-life problems and that users actually benefit from it. It also establishes clear routes for integrating KBS in a principled way with other medical information technology.
Knowledge engineering should meet the users, that is, physicians, nurses and other health care providers. They all handle natural language medical texts. Natural Language Understanding is thus a rapidly growing field of direct concern to medical informatics. Its potential for tomorrow's applications is important although it is limited by its ability to ground its components on a solid model of the domain. This paper considers the benefit of a closed connection between a semantic network modelling a medical domain, and a Natural Language Understanding system working in this same domain. It overviews a number of natural language applications foreseen in the future and provides some recommendations about the future use and benefits of natural language techniques.
Artificial Intelligence in Medicine (AIM) made significant progresses over the last years. In this position paper we analyse four aspects and research themes that, in our opinion, deserve particular attention and should be taken into account in the next years. All of them require the analysis and solutions of new problems both on the computer science and the medical points of view and we believe that their achievement can provide important advantages for AIM as a whole.
This paper presents a perspective on the future use of knowledge processing and decision support in the health sector on a number of topics, namely the integration of decision support techniques into clinical information systems, use of model-based reasoning techniques, wide-spread use of information systems, multimedia information systems, and utilization of virtual reality technology.
A great deal of research has been ongoing in the medical informatics community in an attempt to create medical knowledge-based systems (KBSs) to assist the health care provider in both diagnostic and management decisions. In order for these systems to have a significant impact on health care, sharing of knowledge among institutions must take place. There are many challenges to sharing, including differences in vocabularies, local medical practice, local prior probabilities, syntax of the knowledge-base (KB), evaluation, royalties, and liability. The Arden Syntax has recently become an approved standard for representing knowledge with the purpose of allowing multiple users to create, criticise and share knowledge modules. While providing a common syntax for sharing by itself does not solve most of the challenges, it allows institutions to begin sharing, to see which issues are the most important and the most difficult. It has been found that differences in clinical vocabularies and data schemas provide the greatest current challenge to sharing practical KBSs. Thus, in order for KBSs to succeed on a large scale, ways to coordinate vocabularies and schemas across institutions must be developed. As sharing increases, evaluation and maintenance will become greater challenges.
This paper attempts to identify the necessary features that make the difference between a KBS which will remain in its authors laboratory and a system that becomes really used, i.e a product. We illustrate these features with a prescribing optimization decision support system, called OPADE; this software is being developed in the setting of the AIM initiative and definitively aims at becoming a product used in daily clinical practice in several European countries. While developing the system, we purposedly take a pragmatical, product oriented approach, which may hurt by its simplicity. But it is our feeling that the only future of KBS is to prove that, even if simplified, it does work now.
Within the next decade profound changes in the health care delivery systems are likely to occur, including continuous efforts at cost control and an increased emphasis on quality assurance and outcomes by health payers. There will be an increased reliance on evidence-based medicine, and an increased emphasis on decentralized and coordinated care through regional health care networks. There will be a need for guideline-driven, distributed decision support systems integrated with clinical information systems, and added value, regional health information networks, to enable implementation of the future health care programmes relying on wide dissemination and use of clinical practice guidelines. The theoretical and technical foundations for development of these systems already exist or are going to emerge from the ongoing R & D activities within AIM and other CEC programmes. Main issues to be addressed within the fourth Framework Programme include integration of guideline-driven Decision Support Systems with Clinical Information Systems into clinical Workstations, Tools for collaborative development, validation and maintenance of large clinical knowledge bases (KB), KB safety and soundness issues and development of the necessary infrastructure for regional or national health information networks.
This opinion paper discusses the potentials of telemedicine in relation to the user needs as well as the challenges that are foreseen in medical informatics research and development. Given the developments in telecommunications (ie. the ENS) and the fact that Knowledge Based Systems (KBS) become integrated functionalities the perspective emerges of an open health care information system profiting from the access to a net of multiple KBS. Particular user needs in this respect are: I) means to assess the validity of the KBS applications, 2) features for getting access to more than one KBS when seeking advice, ie. a Decision Support System (DSS) functionality, and 3) a generic interfacing system towards knowledge based DSS. Subsequently the challenges for medical informatics are: 1) validity and applicability assessment, 2) transferability of medical data, information and knowledge, and 3) managing a bidding approach for seeking advice.
A concept of telematics based chronic patient management system is presented, which comprises two main elements: a) the ambulatory systems, an intermittently connectable smart device for patient interfacing, data collection as well as some processing capabilities, and b) the therapy advising tools installed at the departmental, hospital or community sites. The concept proposed rests on a main principle: that chronic patient care based on self-care supported by portable equipments cannot be conceived but integrated with the rest of the available resources that the health care system offers to support current modes of patient treatment, whose efficiency we are trying to enhance.
This paper identifies three requirements for a successful commercialization of a medical decision support system. The system should be convenient, i.e. time-saving and integrated into existing procedures of data acquisition. It should be safe to use for the medical staff, i.e. the advice provided by the system should be within the range allowed by “medical judgement”. Finally, clinical testing of the system should document that usage of the system improves patient outcomes.
There is a widespread expectation regarding future increased productivity within Health Care, with more and better services to be provided with better resource management. In some distinct clinical areas, knowledge based systems have proven the ability to improve the quality of health care services. Based on our experience, this paper proposes the collaboration between groups with clinical and technical expertise that will lead to the development of viable medical KBSs, whose main features are discussed and that will be integrated in the medical telematic infrastructure as integral parts of an electronic patient record.
This paper aims to highlight the strategies developed in the field of computerized electrocardiography and to present the future trends of on-going research in this area, with emphasis on the work performed within the framework of european programs of medical research.
In this paper perspectives are described for the possible future use of knowledge processing and decision support technologies in the area of image related diagnosis and therapy. Starting from a general perspective of technical developments and changes in this field during the next ten years, options are presented and discussed how diagnostic and therapeutic procedures may be supported using knowledge-based techniques. Some research and development activities are proposed which appear relevant from the industrial point of view. The need for application driven projects under industrial leadership is emphasized.
This paper examines the legal issues that may arise in the use of electronic support for cognition, knowledge processing and decision making in the medical field which encompasses a wide range of electronically based techniques simulating at least part of the processes so far considered to belong to the human intellect. It is essential that the legal aspects should be examined at the earliest possible stage so as to develop legal recommendations alongside technological progress rather than retroactively. Needless to say, the topic is extremely vast and this paper only tries to outline the fundamental legal and ethical issues, rather than to formulate detailed recommendations and guidelines.