Towards automated medical decisions

doi:10.1016/0010-4809(72)90080-8

Computers and Biomedical Research

Volume 5, Issue 2, April 1972, Pages 180-189

Computers and Biomed...

https://doi.org/10.1016/0010-4809(72)90080-8 Get rights and content

Abstract

A computer algorithm is presented which will automatically develop a branched testing sequence to aid in medical decision making. The sequence is developed using the symptoms and test results of previously diagnosed patients. Data from one year of Kaiser-Permanente multiphasic screening program were used to demonstrate the technique for identifying patients with previously unrecognized diabetes mellitus. The branched testing sequence and two possible decision rules are presented.

References (15)

G.A. Gorry et al.
Experience with a model of sequential diagnosis
Comput. Biomed. Res.
(1968)
G.O. Barnett
Computer in patient care
N. Engl. J. Med.
(1968)
W.B. Schwartz
Medicine and the computer
N. Engl. J. Med.
(1970)
M.F. Collen
Machine diagnosis from a multiphasic screening program
B. Kleinmuntz
Diagnostic problem solving by computer
Jap. Psychol. Res.
(1965)
A.R. Feinstein
H.L. Bleich
Computer as a consultant
N. Engl. J. Med.
(1971)

There are more references available in the full text version of this article.

Cited by (32)

The effect of noise and biases on the performance of machine learning algorithms
1992, International Journal of Bio-Medical Computing
This paper describes the results of experiments with a machine learning algorithm for the induction of classification trees. We mainly address the impact of noise on the resulting classification tree and on the classification results obtained with the derived tree. We use the domain of the biochemical assessment of thyroid diseases as an example. Some suggestions for quality assessment are outlined that should be available in tools that assist users in deriving classification trees in noisy domains.
A multiclass nonparametric partitioning algorithm
1986, Pattern Recognition Letters
Parametric classification procedures may not be optimal in real classification problems because the properties of the features are not in agreement with the assumptions made. Nonparametric procedures are valuable tools for the design of classifiers in these cases.
This paper reviews a few algorithms from the literature which formed the basis for a procedure with the following properties:
- •
  - it generates a binary decision tree,
- •
  - it can handle mixed data types (discrete and continuous),
- •
  - it performs the feature and criterion selection simultaneously,
- •
  - it can handle multiclass problems,
- •
  - it is interactive and hence enables the user to control the design procedure.
The performance of the algorithm will be demonstrated with a few examples.
Computer-aided medical diagnosis: Literature review
1979, International Journal of Bio-Medical Computing
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.
On discute de la difficulté du travail de diagnostic médical et des avantages de l'ordinateur en tant qu'auxilliaire. On décrit la stratégie générale et la structure de tout système à ordinateur auxilliaire puis on examine la relation entre la précision du diagnostic, les variables fondamentales impliquées dans la conception, l'essai et l'utilisation d'un système à ordinateur auxilliaire. Ces variables comprennent: les algorithmes de calcul, la source d'information utilisée pour construire la base de données, le nombre et le type des maladies étudiées, le nombre et le type des indicateurs, la source de l'échantillon d'essai et celle des diagnostics confirmés. On présente un tableau de 58 systèmes de diagnostic médical à ordinateur auxilliaire; on décrit rapidement chaque système en tenant compte des variables signalées plus haut et de la précision du diagnostic.
Algebraic framework of an optimization problem in diagnosis
1979, International Journal of Bio-Medical Computing
A formal analysis of an optimization problem in medical diagnosis is presented. The mathematical model of the diagnostic process corresponds in this work to an abstract machine definition which is based on the following sets and functions: Set of possible pathological states (diseases) of a patient, set of diagnostic tests (examination methods), set of all possible results of the tests (symptoms and signs), function which chooses successive tests, function performing a test (i.e., assigning to a given test its result), function assigning to each symptom a certain subset of the set of diseases. For a given abstract machine a class of pairs of graphs is formed: their analysis leads to the identification of a class of pairs of subgraphs which corresponds to an optimized diagnostic procedure. The optimization consists of a comparison of the distances between the initial and terminal vertices of the graphs and of a choice of a shortest route to a final diagnosis. For the considered class of pairs of subgraphs a mathematical model of the optimized diagnostic process is reconstructed.
On présente une analyse formelle d'un problème d'optimisation de diagnostic médical. Dans ce travail, le modèle mathématique du processus de diagnostic correspond à la définition d'une machine abstraite fondée sur les ensembles et fonctions suivants: ensemble des états pathologiques possibles d'un patient (maladies), ensemble des tests de diagnostic (méthodes d'examen), ensemble de tous les résultats possibles des tests (symptômes et signes), choix des tests successifs, mise en oleuvre d'un test (c.à.d.) association à un test donné de son résultat, association à chaque symptôme d'un certain sous-ensemble de l'ensemble des maladies. On forme, pour une machine abstraite donnée, une classe de couples de graphes; leur analyse permet d'isoler une classe de couples de sous-graphes conduisant à un processus optimal de diagnostic. Cette optimisation s'effectue par comparison des distances entre les sommets initiaux et finaux des graphes et choix des chemins les plus courts menant au diagnostic final. Un modèle mathématique du processus optimal de diagnostic est reconstruit dans le cas de la classe de couples de sous-graphes considérée.
Perspectives in Clinical Computing
1977, Advances in Computers
This chapter presents a broad overview of the trends that will lead to the enlarged usage of computers in the health sciences. These include the growth of the health care industry itself and the prohibitive cost of continuing to do “business as usual” along with the developing technological foundation upon which more sophisticated systems will be built. The chapter addresses non-fiscal applications of computers to the health care system, including automated interviewing systems, automated multiphasic screening systems, total hospital information systems, automated instrumentation, clinical laboratory systems, transverse axial tomography, and automated diagnostic and clinical consultation systems. The chapter introduces the concept of developing total, computerized clinical information systems and discusses the functions of a total hospital information system (THIS). A total hospital information system incorporates many elements into a single, integrated structure. The key to the system's structure is the individual patient medical record. The structure of a typical THIS includes software packages dealing with administration and fiscal functions such as accounting, accounts payable, budget, accounts receivable, patient census, payroll, and personnel.
A disease-conscious method for sequential diagnosis by use of disease probabilities without assumption of symptom independence
1976, International Journal of Bio-Medical Computing
A method for automatic diagnosis of disease is formulated and applied to a data base of several hundred gastroenterological patients who were each known to have one of six diseases. Application of the method requires no assumptions regarding statistical independence of symptoms. Each disease is associated with its own disease-symptom function, and any order of dependence between the symptoms and each disease may be allowed for. A patient's symptoms are used to determine the value of any specified disease-symptom function. This value is then used to determine the probability that the patient has the corresponding disease. The method is applied to a set of patients and results are compared with those obtained by use of Bayes' theorem. The method is also applied to sequential diagnosis of patients not contained in the initial data base. Additional symptoms are chosen according to their diagnostic value. The entire model is disease-conscious in that disease-symptom functions, disease probabilities and diagnostic values need be evaluated only for those diseases that are considered relevant to the diagnosis.
Une méthode d'aide au diagnostic est formulée et appliquée à un ensemble de plusieurs centaines de malades porteurs d'une affection gastro-entérologique parmi six affections possibles. Cette méthode ne fait pas d'hypothèse sur l'indépendance entre les symptômes. A chaque maladie est associée une fonction maladie-symtôme; et les symptômes que présente un malade, permettent de déterminer la valeur de la fonction. Cette valeur sert à calculer la probabilité que le malade a d'avoir la maladie correspondante.
La méthode a été appliquée à un ensemble de malades et les résultats ont été comparés à ceux obtenus par la formule de Bayes.
La méthode a aussi été utilisée pour le diagnostic séquentiel de malades qui ne faisaient pas partie de l'ensemble des données initiales. Des signes supplémentaires sont choisis suivant leur valeur diagnostic.
L'ensemble du modèle est conscient des maladies les plus probables, dans le sens où les fonctions maladie-symptôme, les probabilités des maladies et les valeurs diagnostiques des signes n'ont besoin d'être calculées que pour les maladies qui paraissent pertinentes pour le diagnostic.

View all citing articles on Scopus

^☆: Supported in part by grants from the Kaiser Foundation Research Institute, and the U.S.P.H.S. through Grant HS 00288 from the National Center for Health Services Research and Development.

^†: Present address of Dr. Gleser is U.S. Army Research Institute of Environmental Medicine, Natick, MA 01760.

View full text

Towards automated medical decisions☆

Abstract

Comput. Biomed. Res.

Computer in patient care

N. Engl. J. Med.

Medicine and the computer

N. Engl. J. Med.

Machine diagnosis from a multiphasic screening program

Diagnostic problem solving by computer

Jap. Psychol. Res.

Computer as a consultant

N. Engl. J. Med.