Methods and Systems for Evaluating Interaction of Medical Products and Dependence on Demographic Variables

• US 20080215402A1
• Filed: 09/22/2006
• Published: 09/04/2008
• Est. Priority Date: 09/22/2005
• Status: Active Grant

• Current Assignee: UBC Late Stage, Inc. (The Cigna Group)
• Original Assignee: UBC Late Stage, Inc. (The Cigna Group)
• Inventors: Hochberg, Alan M., Pearson, Ronald K.
• Granted Patent: US 8,170,889 B2
• Time in Patent Office: Days
• US Class Current: 705/7
• CPC Class Codes: G06Q 10/10 Office automation; Time management
• CPC Class Codes: G06Q 30/0204 Market segmentation
• CPC Class Codes: G16H 10/60 for patient-specific data, e.g. for electronic patient records
• CPC Class Codes: G16H 70/40 relating to drugs, e.g. their side effects or intended usage

Abstract

A computer implemented method of evaluating interaction two or more medical products or conditions, in the context of a reference condition, based on a dataset of patient records includes developing a reference contingency lattice of lattice subsets from a subset of the patient records containing the reference condition, wherein the lattice subsets correspond to the number of patient records that contain each of the two or more medical products or conditions and combinations thereof. A plurality of comparison contingency lattices are developed, each of the comparison contingency lattices developed from a respective random subset of patient records not containing the reference condition, where each of the comparison contingency lattices include lattice subsets that correspond to the number of patient records that contain each of the two or more medical products or conditions and combinations thereof. The lattice subsets of the reference contingency lattice are compared to the corresponding lattice subsets of the comparison contingency lattices to determine whether a lattice subset is indicative of an interaction between the medical products or conditions with respect to the reference condition.

Claims

1. A computer implemented method of evaluating interaction of two or more medical products or conditions, in the context of a reference condition, based on a dataset of patient records, comprising:

developing a reference contingency lattice of lattice subsets from a subset of the patient records containing the reference condition, wherein the lattice subsets correspond to the number of patient records that contain each of the two or more medical products or conditions and combinations thereof;

developing a plurality of comparison contingency lattices, each of the comparison contingency lattices developed from a respective random subset of patient records not containing the reference condition, wherein each of the respective random subsets is of the same size as the subset of patient records containing the reference condition, and wherein each of the comparison contingency lattices include lattice subsets that correspond to the number of patient records that contain each of the two or more medical products or conditions and combinations thereof; and

comparing the lattice subsets of the reference contingency lattice to the corresponding lattice subsets of the comparison contingency lattices to determine whether a lattice subset is indicative of an interaction between the medical products or conditions with respect to the reference condition.

2. The method of claim 1, wherein the step of comparing the lattice subsets comprises calculating z-scores for lattice subsets of the reference contingency lattice based on the corresponding subsets in the comparison contingency lattices.

3. The method of claim 2, wherein the significance level of any of the z-scores is adjusted by a Bonferroni correction based on a plurality of comparisons.

4. The method of claim 1, wherein the medical products comprise drugs, medications or vaccines.

5. The method of claim 1, wherein the medical products comprise medical devices.

6. The method of claim 1, further comprising displaying the reference contingency lattice with subsets highlighted based on indication of interaction between the medical products or conditions.

7. The method of claim 1, wherein a contingency lattice comprises subsets having a partial order such that between any two subsets there exists a least upper bound subset and a greatest lower bound subset that are a part of the contingency lattice.

27. A computer readable data storage means containing program code recorded thereon for implementing the method steps of claim 1.

28. A computing system having a system memory containing computer readable data storage means containing program code recorded thereon for implementing the method steps of claim 1.

8. A computer implemented method of evaluating interaction of two or more medical products or conditions, in the context of a reference condition, based on a dataset of patient records, comprising:

developing a reference orthogonal contingency lattice of lattice subsets from a subset of the patient records containing the reference condition, wherein the lattice subsets correspond to the number of patient records that contain each of the two or more medical products or conditions and combinations thereof but that exclude all the other medical products or conditions and combinations thereof, respectively;

developing a plurality of comparison contingency lattices, each of the comparison contingency lattices developed from a respective random subset of patient records not containing the reference condition, wherein each of the respective random subsets is of the same size as the subset of patient records containing the references condition, and wherein each of the comparison contingency lattices include lattice subsets that correspond to the number of patient records that contain each of the two or more medical products or conditions and combinations thereof but that exclude all the other medical products or combinations thereof, respectively; and

comparing the lattice subsets of the reference contingency lattice to the corresponding lattice subsets of the comparison contingency lattices to determine whether a lattice subset is indicative of an interaction between the medical products or conditions with reference to the reference condition.

9. The method of claim 8, wherein the step of comparing the lattice subsets comprises calculating z-scores for lattice subsets of the reference orthogonal contingency lattice based on the corresponding subsets in the comparison contingency lattices.

10. The method of claim 9, wherein the significance level of any of the z-scores is adjusted by a Bonferroni correction based on a plurality of comparisons.

11. The method of claim 8, wherein the medical products comprise drugs, medications or vaccines.

12. The method of claim 8, wherein the medical products comprise medical devices.

13. The method of claim 8, further comprising displaying the reference orthogonal contingency lattice with subsets highlighted based on indication of interaction between the medical products or conditions.

14. The method of claim 8, wherein a contingency lattice comprises subsets having a partial order such that between any two subsets there exists a least upper bound subset and a greatest lower bound subset that are a part of the contingency lattice.

15. The method of claim 8, further comprising:

determining a number, N, of patient records in the dataset containing indications of each of the two or more medical products;

of patient records in the dataset containing indications of each of the two or more medical products, determining a number, Nx, of patient records containing indications of the reference condition;

within a each of a plurality of random samples of size N of patient records of the dataset, the patient records of the respective random samples containing indications of each of the two or more medical products and combinations thereof but that exclude all the other medical products and combinations thereof, comparing a number of such patient records of each random sample containing the reference condition to Nx; and

determining whether a second-order bystander effect exists between the reference condition and any of the two or more medical products or combinations thereof.

16. The method of claim 15, further comprising graphically displaying the number of patient records containing indications of each of the two or more medical products and combinations thereof but that exclude all the other medical products and combinations thereof of each random sample.

17. A computer implemented method of evaluating interaction of two or more medical products, in the context of a reference condition, based on a dataset of patient records, comprising:

determining a number, N, of patient records in the dataset containing indications of medical product A and medical product B;

of patient records in the dataset containing indications of medical product A and medical product B, determining a number, Nx, of patient records containing indications of the reference condition;

among patient records of a first random sample of size N of patient records of the dataset, the patient records of the first random sample containing medical product A and not medical product B, comparing a number of such patient records containing the reference condition to Nx;

among patient records of a second random sample of size N of patient records of the dataset, the patient records of the second random sample containing medical product B and not medical product A, comparing a number of such patient records containing the reference condition to Nx;

among patient records of a third random sample of size N of patient records of the dataset, the patient records of the third random sample containing neither medical product A or medical product B, comparing a number of such patient records containing the reference condition to Nx; and

determining whether a second-order bystander effect exists between the reference condition and either of medical product A and medical product B.

18. The method of claim 17, wherein the respective acts of comparing include comparing at least 100 first random samples, at least 100 second random samples and at least 100 third random samples.

19. The method of claim 18, further comprising graphically displaying the number of patient records of each of the first random sample, second random sample and third random sample containing indications of the reference condition.

20. The method of claim 17, wherein the two or more medical products comprise drugs, medications or vaccines.

21. The method of claim 17, wherein the two or more medical products comprise medical devices.

22. A computer implemented method of evaluating dependence on demographic variables, in the context of a medical product and a reference condition, based on a dataset of patient records, comprising:

determining a number, N, of patient records in the dataset containing indications of both the medical product and the reference condition;

for a continuous demographic variable, determine an average of the demographic variable of patient records in a first random sample of size N5 the patient records of the first random sample containing an indication of the medical product and not containing an indication of the reference condition;

for the continuous demographic variable, determine an average of the demographic variable of patient records in a second random sample of size N, the patient records of the second random sample containing an indication of the reference condition and not containing an indication of the medical product;

for the continuous demographic variable, determine an average of the demographic variable of patient records in a third random sample of size N, the patient records of the third random sample containing neither an indication of the medical product or an indication of the reference condition;

for a categorical demographic variable, among patient records in a fourth random sample of size N, determine a fraction of patient records for which the categorical demographic variable has a specified value, the patient records of the fourth random sample containing an indication of the medical product and not containing an indication of the reference condition;

for the categorical demographic variable, among patient records in a fifth random sample of size N, determine a fraction of patient records for which the categorical demographic variable has a specified value, the patient records of the fifth random sample containing an indication of the reference condition and not containing an indication of the medical product;

for the categorical demographic variable, among patient records in a sixth random sample of size N, determine a fraction of patient records for which the categorical demographic variable has a specified value, the patient records of the sixth random sample containing neither an indication of the medical product or an indication of the reference condition; and

determining whether a relationship exists between the continuous demographic variable or categorical demographic variable, the reference condition and the medical product.

23. The method of claim 22, wherein the respective acts of determining an average include determining averages of at least 100 first random samples, at least 100 second random samples and at least 100 third random samples.

24. The method of claim 23, further comprising graphically displaying the averages of the at least 100 first random samples, the averages of the at least 100 second random samples and the averages of the at least 100 third random samples.

25. The method of claim 22, wherein the respective acts of determining a fraction include determining fractions of patient records for which the categorical demographic variable has the specified value in at least 100 fourth random samples, at least 100 fifth random samples and at least 100 sixth random samples.

26. The method of claim 25, further comprising graphically displaying the fractions of patient records for which the categorical demographic variable has the specified value for the at least 100 first random samples, the fractions of patient records for which the categorical demographic variable has the specified value for the at least 100 second random samples and the fractions of patient records for which the categorical demographic variable has the specified value for the at least 100 third random samples.