Prediction and App Programming | NAMIBIO

The overall aim is the development and implementation of prediction models for quantifying the individual risk of developing food allergies (FA) early in life. The approach is based on large datasets consisting of both clinical data such as clinical parameters, demographics, lifestyle and psychometric scores, and epigenetic data. To create a tool that benefits parents and health care providers in preventing and timely tolerance induction of FA, these models then will be embedded into an easy-to-use App for parents and a professional App for doctors, respectively.

1. In a first step, a simple prediction model for quantifying the individual risk for FA based on one cohort study will be developed as a Pilot App for parents with the aim to evaluate in a field trial parents´ acceptability and usability (SP4).
2. The pilot prediction model will then be refined using harmonized data from all contributing cohorts (SP1). Furthermore, based on the results from the field trial with parents (SP4), the App will be adapted to parents’ needs and expectations, and will be filled with educational material provided by SP5 resulting in the ready to use Parent App.
3. For the Professional App, an advanced computer algorithm for individual prediction of FA will be developed by use of both clinical data and additional epigenetic data (SP2) and will be implemented in accordance with health care providers’ needs (SP4).

1. All available datasets will be analyzed with respect to occurrence of FA.
2. Prediction models will be developed based on state-of-the-art artificial intelligence approaches, e.g. machine learning, combined with common statistical analysis methods and pharmacometrics model approaches for feature selection.
3. Each prediction model will be triangulated against theory-based models from the literature to increase both credibility and validity of the results.

Principal Investigators
Prof Dr med Sven Wellmann
Prof Dr med Marc Pfister

Team
Britta Steffens
Marvin Häberle