Clinical trials and health outcomes research differ in important ways that impact statistical modeling approaches. Clinical trials typically use homogeneous samples and focus on a single endpoint, while health outcomes data is heterogeneous with multiple endpoints. Predictive modeling techniques used in health outcomes research, like those in SAS Enterprise Miner, are better suited than traditional methods as they can handle complex real-world data without strong assumptions and more accurately predict rare events. Validation of models on separate test data is also important for generalizing results.

1. Clinical Trials Versus Health Outcomes Research: SAS/STAT Versus SAS Enterprise Miner Patricia B. Cerrito [email_address] University of Louisville

19. Distribution of Patient Stays

20. Normal Estimate

22. Kernel Density Estimation

24. Kernel Estimate of Length of Stay

27. Sample Size=5

28. Sample Size=30

29. Sample Size=100

30. Sample Size=1000

31. Confidence Limit The confidence limit excludes much of the actual population distribution

32. Confidence Limit With Larger n

36. Regression Equation

38. Simple Regression

39. Classification Table

40. Classification With 3 Variables continued...

41. Classification With 3 Variables

46. 50/50 Split in the Data Filter data to mortality outcome Filter data to non-mortality outcome Use PROC SURVEYSELECT to extract a subsample of non-mortality outcome Append the mortality outcome data to subsample

47. 75/25 Split in the Data

48. 90/10 Split in the Data

51. Sampling Node

52. Misclassification in Regression

53. ROC Curves

55. Rule Induction Results

56. Variable Selection

58. ROC Curves