Computer-based consultation. Electrolyte and acid-base disorders

Development of a computer-interpretable clinical guideline model for decision support in the differential diagnosis of hyponatremia

INTRODUCTION

Hyponatremia is the most common type of electrolyte imbalance, occurring when serum sodium is below threshold levels, typically 135mmol/L. Electrolyte balance has been identified as one of the most challenging subjects for medical students, but also as one of the most relevant areas to learn about according to physicians and researchers. We present a computer-interpretable guideline (CIG) model that will be used for medical training to learn how to improve the diagnosis of hyponatremia applying an expert consensus document (ECDs).

METHODS

We used the PROForma set of tools to develop the model, using an iterative process involving two knowledge engineers (a computer science Ph.D. and a preventive medicine specialist) and two expert endocrinologists. We also carried out an initial validation of the model and a qualitative post-analysis from the results of a retrospective study (N=65 patients), comparing the consensus diagnosis of two experts with the output of the tool.

RESULTS

The model includes over two-hundred "for", "against" and "neutral" arguments that are selectively triggered depending on the input value of more than forty patient-state variables. We share the methodology followed for the development process and the initial validation results, that achieved a high ratio of 61/65 agreements with the consensus diagnosis, having a kappa value of K=0.86 for overall agreement and K=0.80 for first-ranked agreement.

CONCLUSION

Hospital care professionals involved in the project showed high expectations of using this tool for training, but the process to follow for a successful diagnosis and application is not trivial, as reported in this manuscript. Secondary benefits of using these tools are associated to improving research knowledge and existing clinical practice guidelines (CPGs) or ECDs. Beyond point-of-care clinical decision support, knowledge-based decision support systems are very attractive as a training tool, to help selected professionals to better understand difficult diseases that are underdiagnosed and/or incorrectly managed.

A fuzzy discrete event system approach to determining optimal HIV/AIDS treatment regimens

Treatment decision-making is complex and involves many factors. A systematic decision-making and optimization technology capable of handling variations and uncertainties of patient characteristics and physician's subjectivity is currently unavailable. We recently developed a novel general-purpose fuzzy discrete event systems theory for optimal decision-making. We now apply it to develop an innovative system for medical treatment, specifically for the first round of highly active antiretroviral therapy of human immunodeficiency virus/acquired immunodeficiency syndrome (HIV/AIDS) patients involving three historically widely used regimens. The objective is to develop such a system whose regimen choice for any given patient will exactly match expert AIDS physician's selection to produce the (anticipated) optimal treatment outcome. Our regimen selection system consists of a treatment objectives classifier, fuzzy finite state machine models for treatment regimens, and a genetic-algorithm-based optimizer. The optimizer enables the system to either emulate an individual doctor's decision-making or generate a regimen that simultaneously satisfies diverse treatment preferences of multiple physicians to the maximum extent. We used the optimizer to automatically learn the values of 26 parameters of the models. The learning was based on the consensus of AIDS specialists A and B on this project, whose exact agreement was only 35%. The performance of the resulting models was first assessed. We then carried out a retrospective study of the entire system using all the qualifying patients treated in our institution's AIDS Clinical Center in 2001. A total of 35 patients were treated by 13 specialists using the regimens (four and eight patients were treated by specialists A and B, respectively). We compared the actually prescribed regimens with those selected by the system using the same available information. The overall exact agreement was 82.9% (29 out of 35), with the exact agreement with specialists A and B both at 100%. The exact agreement for the remaining 11 physicians not involved in the system training was 73.9% (17 out of 23), an impressive result given the fact that expert opinion can be quite divergent for treatment decisions of such complexity. Our specialists also carefully examined the six mismatched cases and deemed that the system actually chose a more appropriate regimen for four of them. In the other two cases, either would be reasonable choices. Our approach has the capabilities of generalizing, learning, and representing knowledge even in the face of weak consensus, and being readily upgradeable to new medical knowledge. These are practically important features to medical applications in general, and HIV/AIDS treatment in particular, as national HIV/AIDS treatment guidelines are modified several times per year.

Medical informatics: reasoning methods

The progress of medical informatics has been characterized by the development of a wide range of reasoning methods. These reasoning methods are based on organizing principles that make use of the various relations existing in medical domains: associations, probabilities, causality, functional relationships, temporal relations, locality, similarity, and clinical practice. Some, such as those based on associations and probabilities have been developed to the point where there are off-the-shelf tools available for the researcher to develop new decision support tools. Others such as temporal relations require more effort to use effectively. Even so, we have learned the importance of a separate explicit representation of the domain knowledge and have considerable experience and an impressive armamentarium with which to face the new milieu provided by the Internet.

A computer-based flowcharting system for clinical protocols

In medicine, scientific and technological developments for investigation and treatment are proceeding at an ever increasing rate. Protocols for patient management are becoming ever more complicated. Rational design and scientific evaluation of protocols requires precise documentation. We propose a notation based on a series of simple flowcharts which describe procedures in increasing detail. Each step in a flowchart is justified by a reasoned argument, possibly including reference to published articles. General correctness requirements of a protocol include, for example, conformity with known indications and contraindications for investigations and treatments. These correctness requirements can be specified declaratively in mathematical logic and justified by reasoned argument. This makes the correspondence between a protocol and its scientific foundation even more explicit. Flowcharts of this nature are more easily created and modified using a personal computer. Furthermore, use of the computer enables a protocol to be checked automatically against its specification for more rapid identification of errors during development and maintenance of the protocol. We present the structured design of our flowcharting system in the 'Z' specification language, and we examine the practicality of our approach by means of a case study; the management of infertility. Our flowcharting system may also have application outside medicine where it is necessary to describe formal protocols for complex procedures.

Decision support in healthcare

To address the recognized problems associated with information overload and limited human memory, computer-based systems which help healthcare providers use information to make better decisions have been developed and implemented. These decision aids are designed to improve the quality and reduce the cost of healthcare. Currently, the most widely used computer application is to simply provide needed facts about the patient in an organized and timely fashion. Additionally, healthcare workers can access literature, ask questions of aggregates of patient data for clinical or administrative decisions, receive warnings or suggestions when the patient's data satisfy certain logical rules receive critiques when proposing therapies or ordering diagnostic tests, receive guidelines for standards of care, access programs which analyze tradeoffs and likelihoods of alternative outcomes (decision analysis) and receive lists of differential diagnoses. Given this wonderful panoply of capabilities, the question becomes 'why aren't more people using these aids and what are the demonstrated benefits of such capabilities?' In this paper we review the types of decision aids which have been successfully implemented and the challenges to implementation (knowledge representation, connections to databases, need for comprehensive, coded databases and evaluation of benefits).

Computer-assisted evaluation of adverse events using a Bayesian approach

The differential diagnosis of idiosyncratic adverse drug reactions (ADRs) is complex because for each adverse event there are many possible drug and nondrug causes. Recently efforts have been made to computerize causality assessment methods. A new computerized, user-friendly diagnostic aid for Bayesian assessment of adverse drug events (MacBARDI-Q&A) is described. This computer program performs a differential diagnosis on cutaneous reactions suspected of being drug-induced. The authors' results indicate that development of a valid, simple and user-friendly computerized procedure for evaluating adverse drug events is possible. The ongoing development and application of MacBARDI-Q&A and other similar programs should improve the evaluation of putative adverse drug reactions.

Clinician decisions and computers

The clinician's decisions are subject to numerous distorting influences. Computer decision aids can help avoid these distortions by placing the clinician's limited personal experience into broader perspective through comparison with a larger repository of clinically relevant information; by making explicit the assumptions implied by his or her decisions; and by alerting the clinician whenever the decisions made do not appear consistent with these assumptions, with the available information or with the conventional rules of logic. Practical standards of performance with respect to the development, validation and clinical application of these decision aids are still in evolution, however, and a variety of ethical and legal issues have yet to be addressed. Despite the promise of computer decision aids, it remains to be seen whether their diffusion into medical practice will improve the quality and cost of health care.

Diagnosis of complex acid-base disorders: physician performance versus the microcomputer

Patients with acid-base disturbances that are often complex frequently present to the emergency department. The sometimes hectic nature of the ED can preclude the appropriate quantitative analysis required by these disorders, especially when mixed disturbances are present. A computer program using generally accepted acid-base and electrolyte formulae was developed for use on the Apple II+ or IBM-PC microcomputer. Each of a series of 35 acid-base disturbances incorporating single, double, and triple disorders was correctly identified by the computer in less than 45 seconds. Problem sets based on the same 35 disturbances were presented to 21 physician-subjects at various levels of training from the emergency medicine, internal medicine, pediatrics, surgery, and family practice specialties. Although the physicians were given unlimited time and the necessary formulae to reach a diagnosis, they were requested to perform their analyses in the same fashion used in the ED. Although times varied widely, no physician spent more than five minutes on any problem. The physician correct response rates were 86%, 49%, and 17% for single, double, and triple disorders, respectively. The primary disorder correct response rate was 89% for double disorders and 94% for triple disorders. The primary and secondary disorder correct response rate was 58% for triple disorders. The data suggest that the microcomputer may be beneficial in the rapid assessment of complex disorders.

A computer program for the interpretation of hemodynamic pressures

A computer program for the interpretation of data obtained during hemodynamic monitoring suitable for running on many commonly available microcomputers is presented. Potential modifications of the program are discussed.

DIAG: a computer-assisted dermatologic diagnostic system--clinical experience and insight

Over the past few decades, the computer has been utilized in various ways to assist physicians in formulating their differential diagnoses, usually on an experimental basis. Following a review of the three major technics employed in computer-assisted diagnosis, statistical pattern classification, production rules, and cognitive models, a new system (DIAG) is described. The sole function of DIAG is to assist in the formulation of the differential diagnosis of skin diseases. At the beginning of the fourth year of development, the accuracy of the program in reaching diagnoses has been shown, on many occasions, to be comparable to its expert counterparts. However, it has been demonstrated that the system is not utilized within the dermatology clinic routinely. In response to this, aspects of DIAG have been modified in order to improve its clinical acceptability. Until this acceptance is achieved, these tools will not fulfill their potential to affect patient care.

Standardized diagnosis and treatment of fluid, acid-base and electrolyte disorders in the surgical patient with the aid of a programmable pocket calculator

The approach to fluid, acid-base and electrolyte disorders in the surgical patient has been standardized in a pocket calculator program that categorizes these disorders into formal diagnoses and subsequently uses these diagnoses to determine the appropriate therapy. The program was tested in 80 infusion regimens of patients in the intensive care and surgical wards. The advantages of a standardized approach to these disorders, using diagnostic and therapeutic algorithms, are briefly discussed.

Database technology in health care

In this paper we will introduce the concepts of database technology in a way that will make it easy to relate the issues of the technology to problems in health care. After the objectives of the database approach have been defined, the major components of databases and their function will be discussed. The remainder of this paper presents the scientific and operational issues associated with database technology in health care. In the scientific exposition we will begin with the logical design issues, those that assure that the data will reflect the medical environment correctly, and then we will discuss the choices that are available for the physical implementation of a database on a computer system. The operational aspects will range from data entry to output presentation. The importance and growth of these systems has been well documented. Rather than providing a survey of the field, this exposition is intended to link general concepts to the practices observed by us and others.

The medical problem solving process

Understanding the medical problem solving process has implications for medical education and the effectiveness of medical services. Through adaptation to the task at hand the human problem solver is able to ameliorate the effects of inherent limitations. In adapting to the medical problem solving task demands related to diagnosis and therapy, the physician uses the hypotheticodeductive process. The process draws upon the problem solver's disease centered and data centered knowledge and can be made more effective through the use of various heuristic rules and strategies that the physician develops with increasing expertise. Additional and more refined modes of laboratory support for the medical problem solver can evolve through further understanding of the problem solving process.

Clinical judgment and computers

The increasingly frequent application of formal methods, including algorithms and computer programs, to processes that are ordinarily viewed as judgmental seems to be a source of both promise and unease for physicians. A consideration of some of these methods suggests that it may be helpful to attempt to distinguish carefully between judgment and computation. Medical care involves a complex of inferential processes, any of which may be performed as judgments, and some of which may be carried out as a computation. My purpose here is to identify the latter cases. The empirical evidence suggests that such a demarcation is feasible. The most important question appears not to be "Where can we use computers?" but "Where must we use human beings?" Until this matter is more thoroughly explored, tension between physicians and computer advocates will persist.

Clinical information system for junior doctors

The effect of a computer-based clinical information system is described. The information system, called House Officer Information and Scheduling System, was designed to guide junior medical staff in their decisions about the investigation of common medical problems. By comparing the management of similar groups of patients at two hospitals it is demonstrated that the system influences decisions concerning investigations. In particular, the least common condition reviewed was most affected.

Methods and approaches to determine "optimal" quality for clinical chemical data

Although clinical reasoning and decision-making may be very complex, there are methods to perform both piecemeal and more complete optimizations with regard to the use of clinical chemical data and for assessing quality specification. Some methods and approaches are described and illustrated with examples: (i) rules for propagation of errors in simple algebraic/statistical transformations; (ii) systems and sensitivity analysis using biochemical/pathophysiological simulation models; (iii) systems and sensitivity analysis of simulated clinical classification and decision processes. It is concluded that the possible gain of useful information by improving analytical and pre-analytical quality should be related to the often more important aspects of test selection test and test combination, period and frequency of observation, and the use of correct conceptual models for transformation of data.

Computer use in diagnosis, prognosis, and therapy

Computers are used to influence diagnostic and therapeutic decisions. The computer's information-handling capabilities allow it to serve as a reliable extension of the physician's memory and expander of the physician's information and synthesized knowledge resources. Computers have been used to facilitate decisions through organization of patient data, improved classification of patients, decision analysis in clinical settings, and simulation of expert clinical reasoning. Computer programs are more successful in narrow, constrained, single arenas of medicine with much underlying pathophysiologic understanding and where decisions are based largely on hard laboratory data. New models of synthetic reasoning that simulate expert clinical behavior show promise of supporting complicated decisions concerning problems of multiple diseases. All systems are confronted by problems of consensus and authority of the underlying information used.

Towards the simulation of clinical cognition. Taking a present illness by computer

Remarkably little is known about the cognitive processes which are employed in the solution of clinical problems. This paucity of information is probably accounted for in large part by the lack of suitable analytic tools for the study of the physician's thought processes. Here we report on the use of the computer as a laboratory for the study of clinical cognition. Our experimental approach has consisted of several elements. First, cognitive insights gained from the study of clinicians' behavior were used to develop a computer program designed to take the present illness of a patient with edema. The program was then tested with a series of prototypical cases, and the present illnesses generated by the computer were compared to those taken by the clinicians in our group. Discrepant behavior on the part of the program was taken as a stimulus for further refinement of the evolving cognitive theory of the present illness. Corresponding refinements were made in the program, and the process of testing and revision was continued until the program's behavior closely resembled that of the clinicians. The advances in computer science that made this effort possible include "goal-directed" programming, pattern-matching and a large associative memory, all of which are products of research in the field known as "artificial intelligence". The information used by the program is organized in a highly connected set of associations which is used to guide such activities as checking the validity of facts, generating and testing hypotheses, and constructing a coherent picture of the patient. As the program pursues its interrelated goals of information gathering and diagnosis, it uses knowledge of diseases and pathophysiology, as well as "common sense", to dynamically assemble many small problem-solving strategies into an integrated history-taking process. We suggest that the present experimental approach will facilitate accomplishment of the long-term goal of disseminating clinical expertise via the computer.