Validation of the TeleForm scan workflow in the GNC health study on the example of the questionnaire on physical activity

Electronic data capture (EDC) is an important tool for the digitalisation of paper-based documents such as questionnaires and for the identification of errors before values are finally saved in a database. The data acquisition software TeleForm is one example for an EDC system which is used to digitise paper-based documents. TeleForm checks the data of the scanned document and gives indications of possibly incorrectly read data. In the German National Cohort (GNC) this software is among other things applied to digitalise questionnaires.The following questions are addressed in this article: Is the scan workflow referring to the questionnaires in the GNC and in particular the data acquisition software TeleForm (with the settings chosen for the GNC) reliable? How much loss of data quality is acceptable to reduce the amount of work? Can artificial intelligence replace human inspection sufficiently or will the latter continue to play an indispensable role in the scan workflow of the GNC in the future? By answering these questions, the strengths and the limitations of the scan workflow in the GNC using the TeleForm software will be discussed.The current work uses data collected in the GNC centre in Dusseldorf. 300 questionnaires on physical activity were validated and checked twice, first by the system TeleForm and second by a visual assessment. The data acquisition software TeleForm shows high error rates in interpreting free text fields as well as in reading handwritten numbers. Especially the digit “0” was misinterpreted most often.In order to save time and thus make work easier, some shortcomings must be remedied. This can be achieved, for example, by putting special emphasis on the expansion of the reading areas of TeleForm and on the improved reproduction and reading of numerical values.