Research Information System
Thesis Type: Doctorate
Institution Of The Thesis: Gazi University, Turkey
Approval Date: 2018
Thesis Language: Turkish
Student: Deniz İlhan Topcu
Supervisor: Özlem Gülbahar
Open Archive Collection: AVESIS Open Access Collection
Abstract:Autoverification is the process of evaluating and validating laboratory results using predefined computer-based algorithms without human interaction. By using autoverification, all reports are validated according to the standard evaluation criteria and the number of reports per laboratory specialist is reduced. This allows the specialist to spend more time to examine results that require special attention. In this study, we aimed to develop an expert system for the evaluation and validation of clinical laboratory results which can also help users to easily create rules and to evaluate its performance. The autoverification software was developed using Microsoft Visual C#, CLIPS and R programming languages and was named as myODS. The performance of the software was evaluated by testing the rules defined in myODS using simulated and actual patient results. Simulation patients were generated by the VIA module to check all the rules. After the simulation study, actual patient results were used to test software performance. For actual patient testing, 3 188,095 results from 2015 to 2017 were used from Gazi University Health Research and Application Hospital. Criteria-, test-, and report-based evaluations were performed with actual patient testing. A total of 586 reports which were evaluated by seven users were compared with myODS evaluations. Validation decision limit check was found to be the most important step determining the validation rate. Report-based validation ratio was found as between 34.4% and 56.5%. This rate can be considered as fair for an autoverification system at the beginning. Time spent for report validation will be reduced as one third with this system. In our opinion, further studies for defining rules are required to increase the efficiency and reliability of autoverification systems.