Bioinformatic approach to disease diagnosis

• US 8,005,627 B2
• Filed: 09/08/2007
• Issued: 08/23/2011
• Est. Priority Date: 09/08/2006
• Status: Active Grant

• Current Assignee: New York Medical College (Touro College & University System)
• Original Assignee: Richard Porwancher
• Inventors: Porwancher, Richard
• Primary Examiner(s): Zeman; Mary K
• Application Number:
  US11/852,283
• Publication Number:
  US 20080064118A1
• Time in Patent Office: 1,445 Days
• Field of Search: None
• US Class Current: 702/20
• CPC Class Codes: G16B 20/00 ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
• CPC Class Codes: G16B 40/00 ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/50 for simulation or modelling of medical disorders
• CPC Class Codes: Y02A 90/10 Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Abstract

A multivariate diagnostic method based on optimizing diagnostic likelihood ratios through the effective use of multiple diagnostic tests is disclosed. The Neyman-Pearson Lemma provides a mathematical basis to produce optimal diagnostic results. The method can comprise identifying those tests optimal for inclusion in a diagnostic panel, weighting the result of each component test based on a multivariate algorithm described below, adjusting the algorithm'"'"'s performance to satisfy predetermined specificity criteria, generating a likelihood ratio for a given patient'"'"'s test results through said algorithm, providing a clinical algorithm that estimates the pretest probability of disease based on individual clinical signs and symptoms, combining the likelihood ratio and pretest probability of disease through Bayes'"'"' Theorem to generate a posttest probability of disease, interpreting that result as either positive or negative for disease based on a cutoff value, and treating a patient for disease if the posttest probability exceeds the cutoff value.

Claims

1. A method for constructing a multivariate predictive model for diagnosing a disease for which a plurality of test methods are individually inadequate, said method comprising:

(a) performing a panel of laboratory tests for diagnosing said disease on a test population comprising a statistically significant sample of individuals with at least one objective sign of disease and a statistically significant control sample of healthy individuals or persons with cross-reacting medical conditions;

(b) generating, by a computer, a score function from a linear combination of said test panel results, said linear combination expressed as β

^TY , wherein D is the disease;

Y₁, . . . , Y_k is a set of K diagnostic tests for D;

Y is a vector of diagnostic test results {Y₁, . . . , Y_k};

D′

=not D;

β

is a vector of coefficients {β

₁, . . . , β

_k}for Y; and

β

^T is the transpose of β

;

(c) performing, by the computer, a receiver operating characteristic (ROC) regression or alternative regression technique of the score function, wherein the test panel is selected and β

coefficients are calculated simultaneously to maximize the area under the curve (AUC) of the empiric ROC as approximated by;

2. The method of claim 1, wherein t₀ is the maximum false-positive rate desired by a physician interpreting the tests and is a multiple of 1/n^H, and the β

coefficients and test panel are chosen simultaneously through partial ROC regression, so that the largest area below the partial ROC curve for the (1−

t₀) quantile of individuals without D is generated, wherein β

^T Y_j >

c and the survival function of patients without disease with a score of c, S_H(c), is equal to t_o.

3. The method of claims 1 and 2, wherein the ROC curve is smoothed using the sigmoid function:

S(x)=1/[1+exp(−

x)],

wherein bias is decreased in estimating x values close to zero by introducing a series of positive numbers σ

_n, satisfying the condition that σ

_n approaches zero as n approaches infinity, such that S_n(x)=S(x/σ

_n), wherein the optimal β

is determined using the sigmoid approximation as the sigmoid maximum rank correlation estimator;

4. The method of claim 1, wherein the optimized score function β

^TY generates a score, c_i, for each patient i with D, and c_j, for each control patient j, wherein the likelihood ratios for scores c_i, and c_j,P(c_i,/D)/P(c_i/D′

)and P(c_j /D)/P(c_j /D′

), respectively, are monotone increasing in patients with D.

5. The method of claims 1, 2 and 4, wherein when there is insufficient data to determine the pretest risk that a patient has a disease D and a laboratory reports the likelihood ratio and cutoff value for that patient'"'"'s test results directly to the physician, the cutoff value for the likelihood ratio is determined by observing the likelihood ratio resulting in 99% specificity in a control population of patients and likelihood ratios that exceed the cutoff value thus derived indicate that there is a high probability of disease D.

6. The method of claim 1, the pretest risk of D is calculated using an individual'"'"'s clinical signs and symptoms

7. The method of claim 1, wherein the pretest risk of D is calculated using the distribution of disease manifestation in the population from which said individuals are selected.

8. The method of claim 1, wherein the posterior odds of D are calculated by multiplying the pretest odds of D by the likelihood ratio associated with the score generated by the patient'"'"'s test results;

and where the posterior odds of D are converted into the posttest probability of D by calculating odds/[1+odds].

9. The method of claim 1, wherein an ROC regression approximation is performed in step (c) and is selected from logistic regression, log-likelihood regression, linear regression, or discriminant techniques.

10. The method of claim 1, further comprising the step of substituting at least a portion of said optimal methodology in another multivariate regression technique using less optimum methodology.

11. The method of claim 10, wherein said multivariate regression technique is selected from logistic regression, log-likelihood regression, linear regression, or discriminant techniques.

12. A diagnostic test panel for diagnosing a disease for which a plurality of test methods are individually inadequate comprising laboratory tests selected by the method of claim 1.

13. The diagnostic test panel of claim 12, wherein said disease is Lyme Disease and said test panel includes a plurality of tests, one or more of which are selected from the group consisting of a test for the V1sE1 IgG antibody, a test for the C6 IgG antibody, a test for the pepC10 IgM antibody and a test for an antibody to the BmpA peptide.

14. A computer based method for diagnosing a disease for which a plurality of test methods are individually inadequate, said method comprising:

combining, by a computer, weighted scores from a panel of laboratory test results;

comparing, by the computer, the combined weighted results to a cutoff value; and

displaying by the computer, a diagnosis based on said comparison to said cutoff value,

wherein said laboratory tests, the weighting assigned thereto and cutoff value above which individuals tested have said disease are determined by the method of claim 1, wherein said disease is Lyme Disease.

15. The computer-based method of claim 14, wherein said test panel comprises a plurality of tests, one or more of which are selected from the group consisting of a test for the V1sE1 IgG antibody, a test for the C6 IgG antibody, a test for the pepC10 IgM antibody and a test for an antibody to the BmpA peptide.