×

Computer-implemented medical analytics method and system employing a modified mini-max procedure

  • US 20090259494A1
  • Filed: 05/18/2009
  • Published: 10/15/2009
  • Est. Priority Date: 05/20/2005
  • Status: Active Grant
First Claim
Patent Images

1. A medical analytics method implemented on a computer for diagnosing at least one disease (i) afflicting a patient based on clinical data (m) that excludes subjective qualities of said clinical data (m) and prevalence of said at least one disease (i), said method comprising:

  • a) compiling a knowledge base of disease (i) models exhibiting said clinical data (m);

    b) inputting clinical data present (j) in said patient into said computer;

    c) matching said clinical data present (j) with said clinical data (m) in said knowledge base;

    d) composing a differential diagnosis list of ruled in diagnoses (k) for each of said disease (i) models exhibiting at least one clinical datum (m) matching at least one clinical datum present (j);

    e) computing a probability P(k) for each of said ruled in diagnoses (k) by a mini-max procedure comprising;

    1) obtaining sensitivities S(i)m of each of said clinical data (m) for diseases (i) based on disease (i) models that comprise a total number of disease (i) cases;



    S

    ( i )
    m
    = number



    of



    disease

    ( i )




    cases


    manifesting



    clinical



    datum

    ( m )
    total



    number



    of



    disease

    ( i )




    cases
    ;

    2) computing positive predictive values PP(k)j for clinical data present (j) supporting each of said ruled in diagnoses (k) as follows;



    PP

    ( k )
    j
    = S

    ( k )
    j
    S

    ( 1 )
    j
    +

    + S

    ( k )
    j
    +

    + S

    ( n )
    j
    ,
    where S(k)j are sensitivities of each clinical datum present (j) to said diagnoses (k), and n is the number of said ruled in diagnoses (k);

    3) assigning as probability P(k) of each ruled in diagnosis (k) the maximum value among said positive predictive values PP(k)1 through PP(k)z;


    P(k)=max(PP(k)1, PP(k)2, . . . , PP(k)2),where z is the number of said ruled in diagnoses (j); and

    f) displaying said differential diagnosis list and said probability P(k) for each diagnosis (k) on a display for medical analytics.

View all claims
  • 1 Assignment
Timeline View
Assignment View
    ×
    ×