ENACT Métasurfaces pilotées par l'IA pour les applications biomédicales : Optimisation des biointerfaces fonctionnalisées par des polymères pour le diagnostic avancé et la biosurveillance

Keywords

Cancer Diagnostics, Metasurfaces, Artificial Intelligence, Convolutional Neural Networks, Optimization, Biointerfaces

Subject details

'This PhD offer is provided by the ENACT AI Cluster and its partners. Find all ENACT PhD offers and actions on https://cluster-ia-enact.ai/.' The ability to engineer protein-surface interactions is crucial in designing next-generation biosensors and diagnostic platforms for cancer detection. Among the promising routes are metamaterials as they provide novel functionalities beyond what is possible by conventional materials in the field of sensors. In the biomedical field, the development of metasurfaces has been focused on the manipulation of electromagnetic and acoustic waves and drug delivery. In this project, a powerful AI-driven experimental approach will be adopted to design metasurfaces with targeted bio-interaction properties for cancer diagnostics. The grafting of polymers onto surfaces plays a crucial role in tailoring interfacial properties, enabling precise control over surface energy, wettability, and biomolecular interactions. The protein-polymer interactions are typically governed by principles of thermodynamic equilibrium, steric exclusion, and electrostatics, while grafting density plays a major role in controlling protein adsorption through brush effects and volume exclusion. Additionally, surface energy regulates protein adhesion, either promoting or inhibiting their interaction depending on surface chemistry and wettability. To date, those properties are largely studied on flat, patterned, or 3D (nanoparticles, nanords, etc.) rationally designed surfaces7. However, the development of metasurfaces should provide new insight in polymer-protein interaction and allow advancing beyond the current state-of-the-art in this field. Metasurfaces, functionalized with responsive polymers, can offer a powerful means to enhance biomarker specificity, minimize biofouling, and improve detection sensitivity—all critical challenges in medical diagnostics and therapeutic monitoring. Unlike conventional surfaces, metasurfaces can be designed to manipulate chemical properties at the nanoscale. Accordingly, this project will focus on short-range order and hyperuniform model derived organization of metasurfaces, fabricated by electron beam lithography (EBL) and functionalized via “grafting-from” polymerization, offering a unique platform for controlling protein adhesion, molecular selectivity, and surface energy. In parallel, Machine Learning (ML) techniques will be employed to bridge the gap between the characteristics of the synthesized metasurfaces and the polymer-protein interactions, as measured via different techniques within the project. Unsupervised learning techniques will allow identifying the most influencing surface characteristics to the measured interaction forces, while supervised ML techniques will be tested to map this connection and create a predictive model that will be subsequently employed for optimal metasurface designs that will enhance biomarker detection sensitivity. The most suitable techniques will depend on the quantity and quality of the obtained data, as well as on their nature. For example, the deployment of convolutional neural networks will be investigated for the analysis of the topology of the grafting surface, combined with more traditional regression techniques. Finally, an effort will be placed on including existing knowledge in the ML development pipeline to guide the learning process and include an explainability aspect in the developed models. Overall, this work will generate curated datasets linking metasurface patterning, polymer grafting, and protein adsorption, paving the way for new AI developments in nanomaterials science for health applications.Dr. Alem has extensively studied the behavior of grafted responsive polymers. Dr. Picciali will contribute through her expertise in ML and optimization techniques while Dr. Meimaroglou, will act as a connecting link through his double expertise in polymer science and modeling of physicochemical systems, including data-driven approaches.