Support vector regression for censored data

1. A computer-implemented method of producing a model for use in predicting time to occurrence of a health-related condition, the method comprising:

performing with a data regression and analysis device:

obtaining multi-dimensional, non-linear vectors of information indicative of status of multiple test subjects, at least one of the vectors being right-censored, lacking an indication of a time of occurrence of the health-related condition with respect to the corresponding test subject; and

performing regression using the vectors of information to produce a kernel-based model to provide an output value related to a prediction of time to the health-related condition based upon at least some of the information contained in the vectors of information;

wherein for each vector comprising right-censored data, a censored-data penalty function is used to affect the regression, the censored-data penalty function being different than a non-censored-data penalty function used for each vector comprising non-censored data;

wherein performing the regression comprises using penalty functions that include linear functions of a difference between a predicted value of the model and a target value for the predicted value, and wherein a first slope of the linear function for positive differences between the predicted and target values for the censored-data penalty function is lower than a second slope of the linear function for positive differences between the predicted and target values for the non-censored-data penalty function.

2. The method of claim 1 wherein the regression comprises support vector regression.

3. The method of claim 1 wherein the first slope is lower than the magnitude of a slope of the linear function for negative differences between the predicted and target values for the censored-data penalty function.

4. The method of claim 1 wherein the second slope is greater than the magnitude of a slope of the linear function for negative differences between the predicted and target values for the non-censored-data penalty function.

5. The method of claim 1 wherein the first slope is lower than the magnitude of a third slope of the linear function for negative differences between the predicted and target values for the censored-data penalty function and the magnitude of a fourth slope of the linear function for negative differences between the predicted and target values for the non-censored-data penalty function.

6. The method of claim 1 wherein performing the regression comprises using penalty functions that include epsilon values which control how much deviation between predicted and target values is tolerated before a penalty is assessed, wherein a epsilon value for the censored-data penalty function is different than a epsilon value for the non-censored data penalty function.

7. The method of claim 6 wherein a epsilon value of the censored-data penalty function for positive differences between the predicted and target values is greater than at least one of the magnitude of a epsilon value of the censored-data penalty function for negative differences between the predicted and target values, the magnitude of a epsilon value of the non-censored-data penalty function for negative differences between the predicted and target values, and a epsilon value of the non-censored-data penalty function for positive differences between the predicted and target values.

8. The method of claim 6 wherein the magnitude of a epsilon value of the non-censored-data penalty function for negative differences between the predicted and target values is greater than a epsilon value of the non-censored-data penalty function for positive differences between the predicted and target values.

9. The method of claim 1 wherein the model provides an output value indicative of at least one of a time to occurrence of a health-related condition and a probability of occurrence of the health-related condition.

10. The method of claim 1 wherein the vectors include categories of data of clinical/histopathological data, biomarker data, and bio-image data from a computer image of tissue.

11. Apparatus for producing a model for use in predicting time to occurrence of a health-related condition, the apparatus comprising:

a data regression and analysis device configured to:

obtain multi-dimensional, non-linear vectors of information indicative of status of multiple test subjects, at least one of the vectors being right-censored, lacking an indication of a time of occurrence of the health-related condition with respect to the corresponding test subject; and

perform regression using the vectors of information to produce a kernel-based model to provide an output value related to a prediction of time to the health-related condition based upon at least some of the information contained in the vectors of information;

wherein for each vector comprising right-censored data, a censored-data penalty function is used to affect the regression, the censored-data penalty function being different than a non-censored-data penalty function used for each vector comprising non-censored data;

wherein the data regression and analysis device is configured to perform the regression using penalty functions that include linear functions of a difference between a predicted value of the model and a target value for the predicted value, and wherein a first slope of the linear function for positive differences between the predicted and target values for the censored-data penalty function is lower than a second slope of the linear function for positive differences between the predicted and target values for the non-censored-data penalty function.

12. The apparatus of claim 11 wherein the regression comprises support vector regression.

13. The apparatus of claim 11 wherein the first slope is lower than the magnitude of a slope of the linear function for negative differences between the predicted and target values for the censored-data penalty function.

14. The apparatus of claim 11 wherein the second slope is greater than the magnitude of a slope of the linear function for negative differences between the predicted and target values for the non-censored-data penalty function.

15. The apparatus of claim 11 wherein the first slope is lower than the magnitude of a third slope of the linear function for negative differences between the predicted and target values for the censored-data penalty function and the magnitude of a fourth slope of the linear function for negative differences between the predicted and target values for the non-censored-data penalty function.

16. The apparatus of claim 11 wherein the data regression and analysis device configured to perform the regression is configured to use penalty functions that include epsilon values which control how much deviation between predicted and target values is tolerated before a penalty is assessed, wherein a epsilon value for the censored-data penalty function is different than a epsilon value for the non-censored data penalty function.

17. The apparatus of claim 16 wherein a epsilon value of the censored-data penalty function for positive differences between the predicted and target values is greater than at least one of the magnitude of a epsilon value of the censored-data penalty function for negative differences between the predicted and target values, the magnitude of a epsilon value of the non-censored-data penalty function for negative differences between the predicted and target values, and a epsilon value of the non-censored-data penalty function for positive differences between the predicted and target values.

18. The apparatus of claim 16 wherein the magnitude of a epsilon value of the non-censored-data penalty function for negative differences between the predicted and target values is greater than a epsilon value of the non-censored-data penalty function for positive differences between the predicted and target values.

19. The apparatus of claim 11 wherein the model provides an output value indicative of at least one of a time to occurrence of a health-related condition and a probability of occurrence of the health-related condition.

20. The apparatus of claim 11 wherein the vectors include categories of data of clinical/histopathological data, biomarker data, and bio-image data from a computer image of tissue.

21. A computer-implemented method of producing a model for use in predicting time to occurrence of a health-related condition, the method comprising:

performing with a data regression and analysis device:

obtaining multi-dimensional, non-linear vectors of information indicative of status of multiple test subjects; and

performing regression using the vectors of information to produce a kernel-based model to provide an output value related to a prediction of time to the health-related condition based upon at least some of the information contained in the vectors of information;

wherein performing the regression comprises using penalty functions that include linear functions of a difference between a predicted value of the model and a target value for the predicted value, and wherein at least two different penalty functions are used to affect the regression comprising:

a first penalty function used to affect the regression for vectors of information for which a positive difference is present between the predicted and target values; and

a second penalty function, different than the first penalty function, used to affect the regression for vectors of information for which a negative difference is present between the predicted and target values;

wherein the first penalty function is used to affect the regression for at least one of the following types of vectors of information (i) right-censored vectors of information, lacking an indication of a time of occurrence of the health-related condition with respect to the corresponding test subject, for which a positive difference is present between the predicted and target values and (ii) non-censored vectors of information for which a positive difference is present between the predicted and target values; and

the second penalty function is used to affect the regression for at least one of the following types of vectors of information (i) right-censored vectors of information for which a negative difference is present between the predicted and target values and (ii) non-censored vectors of information for which a negative difference is present between the predicted and target values.

22. The method of claim 21 wherein the magnitude of a slope of the linear function for the first penalty function is different than the magnitude of a slope of the linear function for the second penalty function.

23. The method of claim 22 wherein the first penalty function is used for the right-censored vectors of information for which a positive difference is present between the predicted and target values, the second penalty function is used for the right-censored vectors of information for which a negative difference is present between the predicted and target values, and the slope of the linear function for the first penalty function is less than the magnitude of the slope of the linear function for the second penalty function.

24. The method of claim 22 wherein the second penalty function is used for the non-censored vectors of information for which a negative difference is present between the predicted and target values, the first penalty function is used for the non-censored vectors of information for which a positive difference is present between the predicted and target values, and the magnitude of the slope of the linear function for the second penalty function is less than the slope of the linear function for the first penalty function.

25. The method of claim 21 wherein each of the first penalty function and the second penalty function includes an epsilon value which controls how much deviation between the predicted and target values is tolerated before a penalty is assessed, wherein the magnitude of the epsilon value for the first penalty function is different than the magnitude of the epsilon value for the second penalty function.

26. The method of claim 25 wherein the first penalty function is used for the right-censored vectors of information for which a positive difference is present between the predicted and target values, the second penalty function is used for the right-censored vectors of information for which a negative difference is present between the predicted and target values, and the epsilon value for the first penalty function is greater than the magnitude of epsilon value for the second penalty function.

27. The method of claim 25 wherein the second penalty function is used for the non-censored vectors of information for which a negative difference is present between the predicted and target values, the first penalty function is used for the non-censored vectors of information for which a positive difference is present between the predicted and target values, and the magnitude of the epsilon value for the second penalty function is greater than the epsilon value for the first penalty function.

28. A computer program product producing a model for use in predicting time to occurrence of a health-related condition, the computer program product residing on a non-transitory computer readable storage, medium, the computer program product comprising computer- readable, computer-executable instructions for causing a computer to:

obtain multi-dimensional, non-linear vectors of information indicative of status of multiple test subjects; and

perform regression using the vectors of information to produce a kernel-based model to provide an output value related to a prediction of time to the health-related condition based upon at least some of the information contained in the vectors of information;

wherein performing the regression comprises using penalty functions that include linear functions of a difference between a predicted value of the model and a target value for the predicted value, and wherein at least two different penalty functions are used to affect the regression comprising:

a first penalty function used to affect the regression for vectors of information for which a positive difference is present between the predicted and target values; and

a second penalty function, different than the first penalty function, used to affect the regression for vectors of information for which a negative difference is present between the predicted and target values;

wherein the first penalty function is used to affect the regression for at least one of the following types of vectors of information (i) right-censored vectors of information, lacking an indication of a time of occurrence of the health-related condition with respect to the corresponding test subject, for which a positive difference is present between the predicted and target values and (ii) right non-censored vectors of information for which a positive difference is present between the predicted and target values; and

the second penalty function is used to affect the regression for at least one of said the following types of vectors of information (i) right-censored vectors of information for which a negative difference is present between the predicted and target values and (ii) non-censored vectors of information for which a negative difference is present between the predicted and target values.

29. The computer program product of claim 28 wherein the magnitude of a slope of the linear function for the first penalty function is different than the magnitude of a slope of the linear function for the second penalty function.

30. The computer program product of claim 29 wherein the first penalty function is used for the right-censored vectors of information for which a positive difference is present between the predicted and target values, the second penalty function is used for the right-censored vectors of information for which a negative difference is present between the predicted and target values, and the slope of the linear function for the first penalty function is less than the magnitude of the slope of the linear function for the second penalty function.

31. The computer program product of claim 29 wherein the second penalty function is used for the non-censored vectors of information for which a negative difference is present between the predicted and target values, the first penalty function is used for the non-censored vectors of information for which a positive difference is present between the predicted and target values, and the magnitude of the slope of the linear function for the second penalty function is less than the slope of the linear function for the first penalty function.

32. The computer program product of claim 28 wherein each of the first penalty function and the second penalty function includes an epsilon value which controls how much deviation between the predicted and target values is tolerated before a penalty is assessed, wherein the magnitude of the epsilon value for the first penalty function is different than the magnitude of the epsilon value for the second penalty function.

33. The computer program product of claim 32 wherein the first penalty function is used for the right-censored vectors of information for which a positive difference is present between the predicted and target values, the second penalty function is used for the right-censored vectors of information for which a negative difference is present between the predicted and target values, and the epsilon value for the first penalty function is greater than the magnitude of the epsilon value for the second penalty function.

34. The computer program product of claim 32 wherein the second penalty function is used for the non-censored vectors of information for which a negative difference is present between the predicted and target values, the first penalty function is used for the non-censored vectors of information for which a positive difference is present between the predicted and target values, and the magnitude of the epsilon value for the second penalty function is greater than the epsilon value for the first penalty function.

35. Apparatus for producing a model for use in predicting time to occurrence of a health-related condition, the apparatus comprising:

a data regression and analysis device configured to:

obtain multi-dimensional, non-linear vectors of information indicative of status of multiple test subjects; and

perform regression using the vectors of information to produce a kernel-based model to provide an output value related to a prediction of time to the health-related condition based upon at least some of the information contained in the vectors of information;

wherein performing the regression comprises using penalty functions that include linear functions of a difference between a predicted value of the model and a target value for the predicted value, and wherein at least two different penalty functions are used to affect the regression comprising:

a first penalty function used to affect the regression for vectors of information for which a positive difference is present between the predicted and target values; and

a second penalty function, different than the first penalty function, used to affect the regression for vectors of information for which a negative difference is present between the predicted and target values;

wherein the first penalty function is used to affect the regression for at least one of the following types of vectors of information (i) right-censored vectors of information, lacking an indication of a time of occurrence of the health-related condition with respect to the corresponding test subject, for which a positive difference is present between the predicted and target values and (ii) non-censored vectors of information for which a positive difference is present between the predicted and target values; and

the second penalty function is used to affect the regression for at least one of the following types of vectors of information (i) right-censored vectors of information for which a negative difference is present between the predicted and target values and (ii) non-censored vectors of information for which a negative difference is present between the predicted and target values.

36. The apparatus of claim 35 wherein the magnitude of a slope of the linear function for the first penalty function is different than the magnitude of a slope of the linear function for the second penalty function.

37. The apparatus of claim 36 wherein the first penalty function is used for the right-censored vectors of information for which a positive difference is present between the predicted and target values, the second penalty function is used for the right-censored vectors of information for which a negative difference is present between the predicted and target values, and the slope of the linear function for the first penalty function is less than the magnitude of the slope of the linear function for the second penalty function.

38. The apparatus of claim 36 wherein the second penalty function is used for the non-censored vectors of information for which a negative difference is present between the predicted and target values, the first penalty function is used for the non-censored vectors of information for which a positive difference is present between the predicted and target values, and the magnitude of the slope of the linear function for the second penalty function is less than the slope of the linear function for the first penalty function.

39. The apparatus of claim 35 wherein each of the first penalty function and the second penalty function includes an epsilon value which controls how much deviation between the predicted and target values is tolerated before a penalty is assessed, wherein the magnitude of the epsilon value for the first penalty function is different than the magnitude of the epsilon value for the second penalty function.

40. The apparatus of claim 39 wherein the first penalty function is used for the right-censored vectors of information for which a positive difference is present between the predicted and target values, the second penalty function is used for the right-censored vectors of information for which a negative difference is present between the predicted and target values, and the epsilon value for the first penalty function is greater than the magnitude of the epsilon value for the second penalty function.

41. The apparatus of claim 39 wherein the second penalty function is used for the non-censored vectors of information for which a negative difference is present between the predicted and target values, the first penalty function is used for the non-censored vectors of information for which a positive difference is present between the predicted and target values, and the magnitude of the epsilon value for the second penalty function is greater than the epsilon value for the first penalty function.