Scan Statistics

---

Fig 12. Scan Statistics Window

LabScanner provides a variety of statistics for the evaluation of the performance of rules and rule sets. After a scan has been carried out, choosing Scan Statistics from the Scan menu will calculate statistics for all tests and rules included in the scan and display them in the Scan Statistics window (Fig 12, right). If you have very long rules names, you may need to widen the window a bit to see all the text. The output in the Scan Statistics window can be exported as a text file using the export button at the lower left. This file can be imported into a text editor or word processor for final formatting and printing, or it can be imported into a spreadsheet for analysis. The numbers in the file are tab-delimited, so they will sort to their own column in a spreadsheet, or they can be formatted with a tab in a word processor.

The statistics calculated include overall numbers of patients, events, tests and rules. The number of values outside of the reference ranges indicated in the rule sets are also reported. The overall number of patterns identified, the number of patients who were flagged as having one or more patterns, and information about the persistence of the patterns are reported. Similar total numbers are reported separately for each test specified by a rule set, with the addition of the average number of tests per patient. Within each type of test (i.e., rule set), the number of patients, patterns and pattern persistence information is reported for each rule that was used in the analysis.

LabScanner's notion of persistent patterns has been discussed previously in the Rules Management and Patient Display sections. Persistent patterns are those in which multiple overlapping event sequences are flagged for the same rule and thus the criteria for that rule are satisfied over an extended span of time and events. LabScanner calculates the "persistence events (#)" as the average number of events from the start event to the end event of a persistent sequence (inclusive) for the set of persistent patterns. The "persistance time" is the average time between the start and end events of the persistent sequence for the set of persistent patterns. These persistence characteristics represent a metric that may be useful both in evaluating the extent of certain types of problems and the effectiveness of interventions designed to alleviate problems. For example, effective intervention related to a particular problem pattern should be visible in subsequent data sets as a decrease in the fraction of patients with persistent patterns of that type as well as a decrease in the average number of events and time in the persistent patterns.