C. Stephanidis (Ed.): Posters, Part II, HCII 2013, CCIS 374, pp. 197–201, 2013.
© Springer-Verlag Berlin Heidelberg 2013
Drug Name Similarity Index for Sound-Alikeness
Tomoyuki Nagata
1,*
, Masaomi Kimura
2
, Michiko Ohkura
2
, and Fumito Tsuchiya
3
1
Graduate School, Shibaura Institute of Technology, Tokyo, Japan
ma12078@shibaura-it.ac.jp
2
Shibaura Institute of Technology, Tokyo, Japan
{masaomi,ohkura}@sic.shibaura-it.ac.jp
3
International University of Health and Welfare, Japan
iuhwfumito@gmail.com
Abstract. Drug name confusion is one of major medical errors. Some similar
drug names can cause medical accidents. In order to solve this problem, the
Ministry of Health, Labour and Welfare developed the drug name database sys-
tem to prevent from authorizing drugs whose names are similar to existing drug
names. Previous studies have been proposed the drug name similarity index
based on look-alikeness. Despite of these efforts, the studies do not take
account of drug name confusion caused by sound-alike. In this paper, we pro-
posed the phonetic similarity index based on the features used in articulatory
phonetics.
Keywords: Drug name confusion, Sound-alikeness, Name similarity index,
Medical error, Human error.
1 Introduction
Though drugs are to promote people’s health, they can be toxic if used improperly.
People who use them are not perfect and can cause a medical error. One of the prom-
ising ways to prevent such an error is to device the systems which reduce human er-
rors. Drug name confusion is one of major human errors related to safe use of drugs.
Actually, some drug names are so similar that it makes medical experts confused
unintentionally. For example, at some hospital in Japan, Succin, instead of Saxizon,
was mistakenly administered to a patient and it unfortunately made him die. In order
to prevent such accidents, the Ministry of Health, Labour and Welfare developed the
system, Medicine Similar Name Search Engine, in order to prevent authorizing drugs
whose names are similar to existing drug names. The system computes similarity of
names based on character matching, such as edit distance and head and tail cosine
similarity (htco). However, the system is inadequate to compute similarity more close
to human sensitivity. Because the system computes similarity based on only matching
character but without consideration the character shapes and the pronunciation. For
example, and are also drug names in Japanese. The exist-
ing system computes the similarity of them low because the characters of them are
almost different. However, because their shapes are very similar, people feel that