The Value of Automated Data Mining

Abstract

It’s the late ’70s. A couple of things are heavy on the minds of biomed managers: how to live without disco and how to figure out trends with medical equipment based on information inside file cabinet equipment history folders. Decisions were likely made by gut feelings, and asking peers what they thought should be done regarding equipment replacement and department budgets. Now it’s the ’80s. Along comes the computer, with its 8-inch floppy disk drives. The first computerized maintenance management system (CMMS) programs are written and implemented. These applications stored basic equipment inventory information, would possibly queue up scheduled inspection tasks, and kept some very basic service and inspection history. At least it was easier to print hospital medical device inventories on that rockin’ dot matrix printer. The ’90s come bounding in, and CMMS applications have become a bit more elaborate, utilizing coding to flag important service history entries, such as use errors and certain types of failures. Asset records were coded for key information management, such as service coverage methods. With this coding, more functional reports could be written to focus in on service issues and asset management planning. But this still required printing reports (on that blazin’ laser printer), and then paging through to mark up the reports with action items. Asset records lacking key information reduced the value of the database for benchmarking cost of service or replacement planning. In order to identify asset records that needed data completion, one would have to print reports, page through them, and hand them to others to gather and input these key data such as acquisition cost and purchase date. The year 2000 comes and goes. Society did not collapse. CMMS applications introduce scheduled alerts that can be configured to mine data from equipment history and asset records. These scheduled alerts can deliver well-written reports via e-mail directly to the staff members that need information to act on. Automated data mining can release biomed managers and technical staff members from the tedious review of asset inventories and histories that was once required to make effective equipment management decisions. Well-written queries can be used to key in on records that need action. These queries can be launched close to real-time so action may be taken to correct data issues or to provide communication to technical staff and their customers. Data queries may be utilized in stand-alone reports, built within reporting modules now common in CMMS applications, or sometimes written as TransactSQL language and scheduled for automated execution within the database. A very simple example of a T-SQL query that will identify active asset records missing purchase cost is shown below: