Expert Choice 2000 Serial Key
Download ✶ DOWNLOAD
Expert Choice 2000 Serial Key
The expert will identify the particular population to which the data is being compared. The expert will consider the characteristics of the data set being de-identified (e.g., number of fields, number of records, maximum variables), the information in the data set, and the population to which it is being compared. The data set may have many attributes or information fields and it is important to understand the basis for the assertion that the health information can be linked to only one person. Examples of fields that may contribute to uniqueness are: name; date of birth; Social Security number; or address.
The expert will compute a measure of uniqueness based on the number of unique combinations of data values (of any type) that can be derived from the data set. That is, the number of records that can be derived when individual values are combined in different ways. Different statistical or scientific principles may be used to compute the measure of uniqueness. The expert should consider and document any bias in the measure of uniqueness based on type or size of the population to which the data being de-identified is compared, for example, race, gender, or age. Finally, the expert will assess the accuracy of the measure of uniqueness using error rates estimated from the data set, and compare those results to error rates derived from the same data set after making the underlying assumptions for estimating the error rates explicit.
The expert will determine the likelihood that a record in the data set is expected to be unique (i.e., unlikely to identify more than one person). Statistical and scientific methods are usually used to compare data values in the data set with records in a similar data set. For example, the expert would check whether the data values in the data set are unique (i.e., the frequencies of the values were not derived from the data set being compared). The expert would also verify that the data fields being compared were also used to generate a data set for comparing with the data set to be de-identified. A data set may include many such fields and the expert would consider a variety of ways to compute uniqueness for the records in the data set. Once the expert has made a determination that the health information is unique, the expert will compare the unique measure with a reasonable estimate of how many people share the same values for the information, or characteristics that the data show they have in common. Typically, the data set being de-identified is compared to a wider population, for example, a state, a zip code, an entire city, or more than a state.
A technical expert who has extensive experience in designing and implementing de-identification methodologies, and is familiar with the types of statistical analysis that are likely to be applied during the de-identification process, may be deemed an expert under the de-identification standard. The expert would have both the knowledge and the expertise to implement such statistical analyses and determine the best methods to use to create the de-identified data set. Experts may be found among those with extensive experience in using statistical or scientific principles to design and implement de-identification methodologies. Relevant experience may include experience as a data analyst or a statistician. While statistical experience may be sufficient, the expert should have a strong understanding of the principles of database design and other aspects of designing a data set that would aid in the evaluation of the identification risks that the expert determines to be associated with the information in the data set.
As described in the de-identification standard, the expert may be required to evaluate the risks associated with the identified information and determine whether the potential for re-identification is greater than very small. Once the expert determines that the re-identification risk is greater than very small, the expert will be required to perform or assist in the performance of data manipulations, which may include, but are not limited to, removal of the unique identifier from the data or reduction of the number of unique identifiers in the data. Experts may use statistical methods to evaluate the identification risk associated with records in a data set. The expert also may perform two-sample concentration tests to determine whether the identifiers are all expected to be unique or whether the probabilities that some data elements are expected to be unique are too similar to happen by chance. If the concentration tests show a statistically significant difference in these probabilities, the expert may infer that certain data elements are not expected to be unique.