Technology Behind The CANS-MCI

A Secure Central Database Hosted on the Web
allows weighted scores derived from longitudinal data to compare the most recent test results to previous test administrations and contribute to increasingly accurate predictions. The centralized database, hosted on a web server, was developed in SQL to allow numerous, simultaneous, and secure queries to be handled across the web. This allows score transformations, based upon longitudinal validation data and revised norms, to be applied retroactively to previously obtained scores.

Security issues were addressed in a Phase 2 SBIR grant. A working database model which resides on Screen's local computer borrows data from the web-hosted database so that data can be manipulated in user-friendly applications facilitating sophisticated queries and the design of predictive and statistical formulas.

The programming necessary to swap the SQL data into Access in order to facilitate calculations and graphical representation of data contributes to the overall design of our custom-built application. That application will soon generate immediate sophisticated text and graphical reports via the Internet to the office or clinic from which test results are sent.

Information Architecture Ensuring Subjects' Confidentiality
Prior to each testing session, the following bits of information are collected from test subjects: the last four digits of the social security number*; first initial; last initial; year of birth; month of birth; gender.

A patient userID is created that can be incorporated into a testing facility's records-keeping methods. Prior to the start of any tests, the userID/password for that particular testing facility must be submitted and authenticated.

The combination of anonymous identifying tidbits that creates a userID is complex enough to inform our application that a testee is unique; we then determine if the person has tested before. The only connection between test data and real identity exist in the testing facility's own records system. Reports on screen or on paper use an anonymous user id only.

*No Screen, Inc. database collects or stores personally identifying information. The application itself contains the incomplete identifier tidbits mentioned above. These are indexed to the test data by simple integer IDs. Each clinical testing site may secure its own method for tabulating the IDs against real names and addresses. Test versions in languages other than American English have been developed in which other 4-number codes are used instead of the 4-number partial Social Security code in the US. Each facility has its own database from which reports can be written in the language of test administration or in another language chosen by the testing facility.