Biomedical subjects
Publications and source records attributed to Linda K Goodwin.
This paper investigates whether HIPAA de-identification requirements--as well as proposed AAMC de-identification standards--were met in a large clinical data mining study (1997-2001) conducted at Duke University prior to the publication of the final rule. While HIPAA has improved de-identification standards, the study also shows that privacy issues may persist even in de-identified large clinical databases.
Data mining is a research method that is increasingly being used to predict clinical outcomes, for example, cancer or AIDS survival, diagnostic accuracy in abdominal pain or brain tumors, and much more. In clinical practice, predicting which patients will deliver preterm versus full term remains a complex clinical problem for families and the healthcare system. Exploratory data mining was used for predicting birth outcomes in a racially diverse sample (n = 19,970). Duke University provided data (1622 variables) for data mining methods that found 7 demographic variables yielded .72 area under the curve for receiver operating characteristic (ROC) analyses, suggesting that a parsimonious set of preterm birth outcomes predictors may be possible. Improved prediction is needed for interventions to be appropriately targeted for improved birth outcomes management.