Overcoming barriers to development of cooperative medical decision support models

Attempts to automate the medical decision making process have been underway for the at least fifty years, beginning with data-based approaches that relied chiefly on statistically-based methods. Approaches expanded to include knowledge-based systems, both linear and non-linear neural networks, agent-based systems, and hybrid methods. While some of these models produced excellent results none have been used extensively in medical practice. In order to move these methods forward into practical use, a number of obstacles must be overcome, including validation of existing systems on large data sets, development of methods for including new knowledge as it becomes available, construction of a broad range of decision models, and development of non-intrusive methods that allow the physician to use these decision aids in conjunction with, not instead of, his or her own medical knowledge. None of these four requirements will come easily. A cooperative effort among researchers, including practicing MDs, is vital, particularly as more information on diseases and their contributing factors continues to expand resulting in more parameters than the human decision maker can process effectively. In this article some of the basic structures that are necessary to facilitate the use of an automated decision support system are discussed, along with potential methods for overcoming existing barriers.

DREAM: a shell-like software system for medical data analysis and decision support

A software system was designed whose aim is to support everyday scientific research of physicians in different fields of medicine. DREAM is a shell-like tool which can be customized embedding in it the desirable structure of a particular medical problem. Various basic statistical analyses are provided along with the decision support capabilities. The decision aid is proposed in two steps--feature selection and classifier design. Genetic algorithm is implemented as the feature selection procedure. The classifier design option includes crisp and fuzzy k-Nearest Neighbors rule and a two-level classification scheme based on majority rule on the votes of several first-level k-Nearest Neighbors classifiers. The system's performance is illustrated with a database from aviation medicine.

Fuzzy multi-level classifier for medical applications

In this paper a fuzzy pattern recognition model is described, which is a tool to handle problems with noncrisp and multi-class membership of the objects. It is oriented to medical diagnostics, where the patients suffer from more than one disease in different degrees. Fuzzy pattern recognition is supposed to fit medical diagnostic problems better than conventional pattern recognition. The design of a multi-level fuzzy decision scheme is considered in order to derive high performance, taking into account expert logic and human experience. Two main topics are discussed--the criterion for evaluation of classification accuracy and the training rule. The implementation of fuzzy multi-level classifier is illustrated with real clinical data.

Diagnosis of acute abdominal pain using a three-stage classifier

The present paper deals with an application of a three-stage classifier based on a decision tree logic to the diagnosis of acute abdominal pain. On the basis of clinical information collected from a series of 476 patients suffering from abdominal pain of acute onset, the method of multistage classifier synthesis is presented. The results of classification accuracy using a modified version of k-nearest neighbours strategy for different features used at interior nodes of a tree are given.

Computer diagnosis in jaundice. Bayes' rule founded on 1002 consecutive cases

Extensive clinical and clinical chemical information was collected from 1002 consecutive jaundiced patients. Initial selection of variables based on Chi 2-tests or Mann-Whitney U-test allowed the removal of 64 of the 107 variables originally collected. A further selection of variables was carried out using a modified version of Bayes' rule thus reducing the number of variables from 43 to 22. Of the 982 patients with a final diagnosis 743 patients (76%) could be classified correctly into one of 13 diagnostic categories. The Bayes' rule was also applied to a test group of a further 110 jaundiced patients and found to perform equally well: of 108 patients with a final diagnosis 81 (75%) were correctly classified. A comparison between the clinician's diagnosis and the computer-aided diagnosis according to Bayes' rule demonstrated agreement with regard to one of the 13 diagnostic alternatives in 734 patients (75%), of whom 81 patients were wrongly diagnosed. In the test group agreement upon diagnosis was found in 80 patients (74%). By plausibly combining the computer-aided and the clinician's preliminary diagnoses, more correct classifications were obtained than with either method alone. Many diagnostic modalities such as ultrasound examination, CT-scan, and direct cholangiography are at hand today for the differential diagnosis of jaundice. Computer-aided diagnosis using Bayes' rule has proved a reliable tool for the clinician and can be used in the planning of a diagnostic strategy for the individual jaundiced patient.

Thyroid disorders: automatic diagnosis in CONSULT I

Automatic diagnosis of thyroid diseases is implemented on CONSULT I, a microcomputer system based on the Patrick model for computer-assisted diagnosis in medicine. The thyroid 'subsystem' consists of 19 classes (diseases) and 16 features (signs, symptoms, laboratory tests). For 76 test cases obtained from patient records (recognition samples), the 'true' class (disease) is decided in the highest 'probability' number in 89% of cases and in the differential diagnosis in 100% of cases. Performance is compared to physicians. Estimation of class-conditional probability densities utilizing equivalence regions in the feature space is discussed.

Symposium on the role of the laboratory in clinical decision making--Part II. Perspectives on clinical decisions

The critical role of expert judgment in clinical decision support has been accommodated in a class of approaches termed knowledge based systems. These have arisen from work in artificial intelligence on expert performance. Some of the perspectives and insights that have been gained from these approaches are briefly discussed.

Statistical pattern classification with binary variables

Binary random variables are regarded as random vectors in a binary-field (modulo-2) linear vector space. A characteristic function is defined and related results derived using this formulation. Minimax estimation of probability distributions using an entropy criterion is investigated, which leads to an A-distribution and bilinear discriminant functions. Nonparametric classification approaches using Hamming distances and their asymptotic properties are discussed. Experimental results are presented.

[Helping physicians arrive at medical diagnoses: how and why (author's transl)]

In the majority of the medical disciplines the diagnosis still constitutes the corner-stone of medicine. Every year this function becomes increasingly difficult and time-consuming and it is felt that the doctor should be able to benefit easily and rapidly from the most recent findings. A simple analysis of the natural process of decision-making, followed by a critical inventory of the various methods used to diagnose, has led us to develop our method for computer-assisted diagnosis (ADM). In our presentation of the methodology of diagnostic decision-making we examine the natural process, the analytical and synthetic methods as well as the difficulties and errors liable to occur. Four methods used in arriving at a diagnosis are analysed and criticized: those based on probability, those based on pattern recognition, those based on logic, and those based on classification methods. Finally a case is made for a computer-assisted diagnostic system based on a logical method of the documentation type.

Pattern-Based Interactive Diagnosis of Multiple Disorders: The MEDAS System

A knowledge-based interactive sequential diagnostic system is introduced which provides for diagnosis of multiple disorders in several body systems. The knowledge base consists of disorder patterns in a hierarchical structure that constitute the background medical information required for diagnosis in the domain under consideration (emergency and critical care medicine, in our case). Utilizing this knowledge base, the diagnostic process is driven by a multimembership classification algorithm for diagnostic assessment as well as for information acquisition [1]. A key characteristic of the system is congenial man-machine interface which comes to expression in, for instance, the flexibility it offers to the user in controlling its operation. At any stage of the diagnostic process the user may decide on an operation strategy that varies from full user control, through mixed initiative to full system control. Likewise, the system is capable of explaining to the user the reasoning process for its decisions. The model is independent of the knowledge base, thereby permitting continuous update of the knowledge base, as well as expansions to include disorders from other disciplines. The information structure lends itself to compact storage and provides for efflcient computation. Presently, the system contains 53 high-level disorders which are diagnosed by means of 587 medical findings.

Computer-aided medical diagnosis: literature review

The difficulty of the medical diagnostic task and the advantages of the computer as an aid in this task are discussed. The general strategy and structure of any computer-aided system is presented, and the relationship of diagnostic accuracy to key variables involved in the development, test and use of a computer-aided diagnostic system is examined. These variables include: the computer algorithm, the source of the information used to develop the data base, the number and type of diseases under investigation, the number and type of indicants used, the source of the test sample, and the source of the validated diagnosis. A table of 58 empirically tested computer-aided medical diagnostic systems is presented; each system is summarised in relation to the variables mentioned above and diagnostic accuracy.

Dynamic decision models for clinical diagnosis

A unified approach to clinical decision-making is presented. This combines partially observable Markovian decision processes (Markov or semi-Markov) with cause-effect models as a probabilistic representation of the diagnostic process. Pattern recognition techniques are used in a first stage of system state identification. This new class of dynamic models has a direct application to medical diagnosis and treatment and specific physiological examples are emphasised. The methodology is given for combining the patient state of health, the clinician's state of knowledge of the cause-effect representation from the observation space (measurements), feature selection using pattern recognition techniques and, finally, the treatment decisions with which to restore the patient to a more desirable state of health. A cost functional for the decision process has then to be optimised according to some pre-assigned objective function (social return from the patient state of health or treatment cost for the patient), when the process has an infinite time horizon.