iComplai PestiPredict - Advanced Pesticide Risk Prediction for Sourced Ingredients

Session: Technical Session 7 – Data Science and Risk Assessment

Date: Thursday, 2 May 2024
Time: 09:00 - 09:15 (CEST)
Location: London (Hilton Geneva Hotel & Conference Centre)

Title: iComplai PestiPredict - Advanced Pesticide Risk Prediction for Sourced Ingredients

Presenter: Asli Solmaz-Kaiser (CEO)
Author: Janosch Peters (CTO)

Abstract

Introduction: As seen in the case of ethylene-oxide where thousands of products were recalled worldwide in 2021 and 2022, the pesticides that are forbidden and typically not included in the multi-modal tests can cause completely unexpected recalls. Given their significant carcinogenic potential, the control of pesticide residues is vital for safeguarding public health.

Purpose: The primary objective of PestiPredict is to equip food manufacturers with a tool that can predict pesticides above the permitted maximum residue levels that are not known to exist in certain raw materials using machine learning: The so-called “unknown-unknowns”.

Methods: iComplai PestiPredict integrates a robust machine learning (ML) algorithm that analyzes a comprehensive set of data including over 300 million residue testing results, over 40.000 pesticide related authority notifications worldwide and regulatory landscape. Utilizing advanced data analytics and machine learning, the system predicts the likelihood of which pesticide may appear on which raw material, for a total of app. 5.000 raw material and origin combinations, allowing for targeted and timely testing and quality control measures. The system also analyzes the gap between the user’s existing test protocols and the predicted pesticides, reducing the analysis time for the user.

Results: The use of AI-infused capabilities demonstrated measurable quantitative and qualitative results:

1. Based on the raw material data availability, the accuracy rate of pesticide risk prediction reaches up to 96%. (Example: peppers from Turkey)

2. Prevention and saving of potential food recall of around €10 million

3. The time and consequently cost saving of a risk analyst approximating around €30k per year and person

Significance: The presented study enables food producers to switch from a reactive to proactive anticipation of potential pesticide residue risks and effectively evaluate their possible effects, thereby reducing the risk of recalls and consumer health issues.

Read more about pesticide risk prediction here.

Convenors: Mariem Ellouze (Digital Food Systems, Institute of Food Safety and Analytical Science, Digital Food Safety Department, Nestlé Research Center), Monika Trzaskowska (Warsaw University of Life Sciences - SGGW; Institute of Human Nutrition Sciences)

Register here.