NEWS RELEASES - JUN 2026

Can we predict how fast a bioplastic disappears? A new AI tool says yes

Researchers at the Agricultural University of Athens (AUA), a partner in the ANIPH project, published a study that helps predict how quickly biodegradable bioplastics break down in the environment.

Plastic production has exploded over the last seventy years, growing from 1.5 million tonnes in 1950 to 359 million tonnes in 2018. More than 60% of the plastic waste from our homes is single-use food packaging made from petroleum-based plastics. On top of that, the world now uses up to 400 million tonnes more plastic per year than it produces. Together, these trends put real pressure on the environment.

Polyhydroxyalkanoates (PHAs) are part of the answer. These are bio-based plastics, produced by microbial fermentation rather than refined from oil, and designed to biodegrade naturally once they’re no longer needed. This is exactly the principle behind ANIPH’s wound dressings and packaging: materials that come from nature and safely return to it.

Knowing that a bioplastic will biodegrade isn’t the same as knowing how fast. To answer this question, the Agricultural University of Athens team conducted extensive research, recently published in the article A Data-Driven Framework for Predicting PHBV Biodegradation-Induced Weight Loss Based on Laboratory and Real-Environment Condition Tests.

The researchers focused on PHBV (poly(3-hydroxybutyrate-co-3-hydroxyvalerate)), one of the most widely studied PHA materials, and started by building a large, carefully checked database of 1,467 individual, time-tracked weight-loss measurements. Afterwards, the team trained and tested predictive models using two machine-learning methods (Random Forest and XGBoost), a type of approach known as a QSAR model, which learns the statistical link between a material’s properties, its environment, and how much weight it loses over time as it biodegrades. Finally, they validated the models rigorously, including testing them on data the models had never seen before. Both achieved strong accuracy, explaining more than 92% of the variation in weight loss on test data (and over 96% on training data).

According to the data, three factors stood out as the most important in predicting weight loss:

  1. Exposure time: how long the material has been in the environment
  2. Degradation environment: whether it’s in soil, the sea, fresh water, or compost
  3. Hydroxybutyrate (HB) ratio: a measure of the polymer’s chemical composition

The team turned the model into a ready-to-use web application on the Jaqpot computational platform (jaqpot.org), openly accessible to the scientific community through the ANIPH virtual organisation. This means researchers, manufacturers, and innovators can now estimate how a given PHBV formulation is likely to biodegrade in a chosen environment, supporting smarter, faster, evidence-based design of biodegradable plastics.

Read the full article here.

Cover photo by chaiyananuwatmongkolchai from Pixabay