Using semi-automated approach for mapping local russian laboratory terms to loinc

Abstract

Manual mapping of laboratory data to Logical Observation Identifiers Names and Codes (LOINC) requires a major effort. Application of the LOINC mapping assistant RELMA V.6.6 can reduce the effort required for mapping. The goal of the paper s to study the potential of semi-automated mapping of Russian laboratory terms to LOINC. We performed semi-automated mapping of the 2563 terms from two clinics in Russia. The first step was automatic mapping using RELMA V.6.6 and LOINC V.2.48 Russian translation by Yaroslavl state medical academy. The second step was a manual expert mapping. To evaluate the correctness of mapping we randomly selected 50 of the most commonly (from the first 20 %) and 50 of the most rarely (from the last 20 %) used terms from each clinic (in total 200 terms from each clinic). This sample of 200 terms was reviewed by two experts. The paper presents the results of semiautomatic mapping of Russian laboratory terms to LOINC. Two clinics (A and B) and a laboratory service participated in the project. We were able to map 86 % (Clinic A) and 87 % (Clinic B) of laboratory terms and 99 % of terms used in 2014. In total 2372 out of 2563 were mapped. The required effort was reasonable and the price of mapping and maintenance was considered as relatively low in comparison to manual methods. RELMA V.6.6 and LOINC V.2.48 offer the opportunity of a low effort LOINC mapping even for non-English languages. The study proved that the mapping effort is acceptable and mapping results are on the same level as the manual mapping.