When a tumor manages to nestle itself in the brain of a living being, it has – from the tumor's perspective – done something particularly clever. It has hidden itself behind one of the most powerful barriers with which the body protects its most important organs: the blood-brain barrier, a highly selective filter that only allows certain substances to pass through. Most drugs are not among them, though. It is therefore a major challenge for biomedical research to find an effective chemotherapy for brain tumors.

In recent years, researchers have found a promising ally: nanotechnology. Materials in the nanoscale can, figuratively speaking, take on the role of postmen who deliver active ingredients to the right address. Since nanoparticles are unimaginably small – about 500 times smaller than the diameter of an average human hair – some manage to pass through the body's protective barriers without damaging them. To stick with the example of brain tumors, nanoparticles could transport chemotherapeutical drugs across the blood-brain barrier, where they can fight the brain tumor.

In search for the right nanomaterial

However, depending on the task they have to perform, nanoparticles must have very specific properties: Depending on their shape, material composition and size, they are distributed very differently in the body and accumulate in different organs. It is therefore important to find out which particles perform their task best and cause as little damage as possible. Until now, researchers have mainly used mouse models to answer these questions: They administered various types of nanomaterials to mice and then examined how these were distributed in the mouse body and what side effects they caused. However, these animal studies are not only complex, time-consuming and expensive, they also raise ethical issues. It is not without reason that Swiss animal welfare legislation requires the number of animal experiments carried out to be kept to the minimum necessary.

AI mouse with a decisive advantage

Empa researcher Jimeng Wu, a doctoral student in Empa’s «Nanomaterials in Health» and «Technology and Society» labs, has therefore developed a virtual mouse that uses AI to perform these tests in a much more time-efficient manner. Wu based this so-called physiologically based pharmacokinetic model (PBPK model) on 18 mouse studies: data from experiments conducted by various research teams on living mice. She also integrated a statistical method, Bayesian analysis with Markov chain Monte Carlo simulations, into her model.

The result is a virtual mouse that can be administered – also virtual – nanoparticles. The model then calculates their distribution in the mouse's body based on properties such as size, coating and surface charge. Compared to a traditional PBPK model, which is calibrated for a single substance at a time, Wu's AI mouse has a decisive advantage: ‘The model can adapt its parameters to the measurable properties of the respective nanoparticle,’ explains Jimeng Wu. The tool owes this ability to the multivariate linear regression model, a machine learning approach.

Contribution to Safe and Sustainable by Design

‘This AI-supported screening tool allows researchers to virtually test which type of nanoparticles are best suited for a specific task before they even manufacture these particles,’ explains Jimeng Wu. This not only saves time, but also costs, because it provides a decision-making aid before a costly clinical trial is started.

‘The model thus contributes to the concept of Safe and Sustainable by Design (SSbD),’ adds Peter Wick, who is supervising Jimeng Wu's doctoral thesis together with his colleague Bernd Nowack. This is because the virtual mouse increases the safety of new materials or therapies even before they are developed. However, he points out that the data set used to train the model is still very limited: So far, only 18 peer-reviewed papers with sufficient data quality have been found. ‘In many studies, the properties of the nanoparticles used are not described in adequate detail,’ he notes. The task now is to feed the virtual mouse with additional study data and verify it in order to further increase the reliability of the predictions. ' Our long-term goal is to shorten the process of developing nanomedicine materials all the way to their use as a drug in patients, while ideally being able to avoid animal testing‘, he emphasizes.

Adapting the model for human diseases

Jimeng Wu's future research will also focus on a so-called ‘bridge strategy’ to transfer the principle of her in silico model to human research. To this end, she plans to embed the principles of the virtual mouse in a human PBPK model. Unlike her simulated mouse, which only calculates the distribution of nanoparticles in the liver, kidneys, lungs and spleen, a human in silico model could also be used to study sensitive target organs – for example, to investigate the extent to which certain nanoparticles can cross the blood-brain barrier. Even the brain tumour mentioned at the beginning would no longer feel safe behind this barrier – nanoparticles could act as ‘postmen’ and deliver a package containing a targeted dose of chemotherapy.