Automatic Leaflet Optimizer

Despite the legislative efforts to improve the readability of patient information, different surveys have shown that respondents (which often do not have any medical background) still have problems understanding the information. Empirical research has shown that the editing of the leaflet text mainly suffers from three important shortcomings: 1) The text is often needlessly lengthened by redundant repetitions. EXAMPLE Beromun will not be prescribed and given by your doctor (...) if you are pregnant or planning to become pregnant. You must not breast feed for at least seven days after receiving Beromun. (...) If you are pregnant you should inform your doctor who will then not prescribe Beromun for you. (...) If you are breast-feeding you must not breast-feed your child for at least seven days after receiving treatment with Beromun. [section 2] The leaflets still contain many scientific terms which remain unexplained. In case these terms are accompanied by a popular variant or a definition, this happens in an unsystematic way. EXAMPLE Special precautionary measures (e.g. blood tests) may be appropriate if you take potassium supplement, potassium-containing salt substitutes, potassium-sparing medicines (diuretics, such as certain “water” tablets), lithium containing medicines, or digoxin together with Micardis. The antibiotic rifampicin has been shown to reduce the blood concentration of VIRAMUNE. Your doctor may consider using rifabutin instead. (...) Cimetidine, macrolides (e.g. clarithromycin) and the antifungal drug fluconazole have been shown to increase blood concentrations of Viramune. The distinction between informative texts, instructions and risk warnings is often unclear. EXAMPLE Co-administration of Viramune oral suspension and efavirenz is not recommended since co-administration does not improve efficacy over either medication alone may increase the risk of side effects. Given this problem’s social and economic relevance, we aim to develop an Automatic Optimizer of Patient Information (ABOP). It is an authoring environment which guides and supervises the author through the creation of new leaflet texts as well as the adaptation of existing leaflets. The main objective is not only to produce more consistent and comprehensible patient information but also to do so in a less time consuming way. Apart from guiding the author through the different formal aspect possibilities, ABOP should also give support as regards content and the problems mentioned above: 2) Redundancy shall be eliminated. Scientific terminology shall be reduced as much as possible and should a popular counterpart exist, it shall be given priority over the scientific term. If there is no popular term available, the scientific term shall be explained by means of a definition or an explanation. 3) The different illocutionary forms shall stand out clearly (e.g. through layout) and phrasings shall be linked to a fixed content, thus leaving no doubt about the risk level. In order to be able to offer this support as regards contents, text mining is needed. We aim to reach this goal by using language technology software for syntactic and semantic analysis. The impact of all of these methods will be measured by a quality check on three levels: the readability, comprehensibility and usability of the patient information leaflet.