Methods to find food-health relations for personalized advice

G. Simsek-Senel, M.J. Ummels, M.J. Timmer, J.L. Top

Research output: Book/ReportReportProfessional


Digital personalised dietary advice is increasingly popular. However, reference to scientific knowledge is often missing or only given indirectly. As a consequence, new insights are either neglected or taken up without adequate scientific rigour. In the Personalized Nutrition and Health (PNH) Project we aim to strengthen dietary advice for the consumer using scientific evidence. This advice is based on the ‘Dutch Dietary Guidelines 2015’, published by the Health Council of The Netherlands (Gezondheidsraad), extended with more recently found evidence. The Dietary Guidelines document presents detailed food-health relations using scientific evidence from literature, including the resulting guidelines. An accompanying document describes the approach that was followed to find evidence in literature and explains a detailed methodology for assessing the evidence. The entire process of finding and assessing literature is done manually by the Health Council committee. However, the discovery of relations between food and health continues over time and the process of assessing the scientific evidence demands large use of resources. We submit that automated reasoning will make the process of finding and assessing the scientific evidence more effective and will help keeping scientific evidence up-to-date. We present the steps that the official authorities take to draw conclusions from scientific studies on foodhealth relations. Based on this analysis, we propose potential routes to automating at least parts of the process. First, the detection of relevant publications can be further automated. Secondly, the most important cause-effect relations can be extracted from these publications using techniques from automated summarization. These steps require creating dedicated controlled vocabularies. Finally, the process of grading the level of evidence can partly be automated using current practice in this area. The most challenging step will be to include contextual information, which is required for matching the evidence found with conditions of the real life case for which advice is sought.
Original languageEnglish
Place of PublicationWageningen
PublisherWageningen Food & Biobased Research
Number of pages29
Publication statusPublished - 2021

Publication series

NameReport / Wageningen Food & Biobased Research


Dive into the research topics of 'Methods to find food-health relations for personalized advice'. Together they form a unique fingerprint.

Cite this