Statistical methods for multivariate ordinal data which are used for data base driven decision support
AssigneeThe Rockefeller University
InventorKnut M. Wittkowski
Life of this patent
Estimating the term…
Abstract
A method of analysis including an intrinsically valid class of statistical methods for dealing with multivariate ordinal data. A decision support system that can (1) provide automated decision support in a transparent fashion (2) optionally be controlled by a decision maker, (3) provide for an evidence acquisition concept, including automatically increasing the content of an underlying database, and (4) provide a computationally efficient interactive distributed environment. The method is exemplified in the context of assisted diagnostic support.
Claims23
An intrinsically valid statistical method for the analysis by a decision maker of inexact ordinal data having one or more variables to compare populations comprised of at least one object, the method comprising the steps of:
- (a) partially ordering data by determining for all pairs of data the order of a first datum compared to a second datum as (i) superior, (ii) inferior, (iii) equal, or (iv) undecided, wherein for tuples a partial ordering comprises the first datum to be superior if for each variable the first datum is superior or equal, and for at least one variable, the first datum is superior;
- (b) factorizing the partially ordered data;
- (c) generating a score for each datum based on the partial ordering;
- (d) estimating an information content for each of the scores;
- (e) generating a weight for each score based on the information content;
- (f) aggregating the scores and weights of all data using statistical methods for weighted rank scores, wherein the statistical methods comprise ranking, positioning, comparing, discriminating/regressing, and clustering;
- and
- (g) deciding which category the inexact ordinal data should be assigned to, thereby allowing the decision maker to act on the assignment to the category.
A process based on an intrinsically valid statistical approach wherein a decision maker obtains an ordered list of categories to which an entity may be assigned by utilizing a database of reference data sets of known categories and a potentially large set of variables, the process comprising the steps of:
- (a) restricting a database of reference entities of known categories to an ad-hoc database based on a first subset of variables, termed characteristics;
- (b) selecting a set of control categories based on a second subset ofvariables, termed control indicators;
- (c) selecting a set of case categories based on a third subset of variables, termed case indicators;
- (d) selecting a reference population subset for each of the case categories and one reference population for a union of the control categories;
- (e) selecting a set of variables, termed discriminators, specific to a selected case category and the entity's characteristics subset, wherein the entity is positioned with respect to the joint case population and control population;
- (f) determining the entity's score relative to the control population, termed specificity, and the subject's score relative to the case population, termed sensitivity;
- and,
- (g) ordering the categories by utilizing information from all obtained relative positions and consequences of assuming the entity to belong to a particular category.
Dependencies are inferred from the claim text.