Japanese researchers have developed an artificial intelligence designed hydrogel that combines strong underwater adhesion with self-healing properties, marking a significant step forward in the use of AI for advanced materials discovery. The breakthrough could open new possibilities across healthcare, robotics and industrial applications where conventional adhesive materials often struggle in wet environments.
The research, published in Nature, demonstrates how machine learning can accelerate the development of high performance soft materials by identifying optimal polymer combinations that would have taken considerably longer to discover through traditional laboratory experimentation. The scientists integrated data mining, laboratory testing and AI models to create a hydrogel capable of maintaining strong adhesion even underwater while repairing itself after sustaining damage.
Hydrogels are water-rich polymer materials already used in products ranging from contact lenses to wound dressings and tissue engineering. However, existing hydrogels typically face limitations when exposed to water, where adhesive strength declines significantly. The newly developed material addresses this challenge by achieving underwater adhesive strength exceeding 1 megapascal, representing an order of magnitude improvement over many previously reported underwater adhesive hydrogels and elastomers.
According to the researchers, the AI driven design process began by analysing naturally occurring adhesive proteins found in marine organisms. Instead of manually testing thousands of chemical combinations, the team mined protein databases to identify recurring molecular patterns associated with strong underwater adhesion. Those insights were then translated into synthetic polymer formulations, creating an initial dataset of 180 bio-inspired hydrogels for experimental evaluation. Machine learning models subsequently refined the formulations to identify the most effective combinations, significantly accelerating material optimisation.
Beyond its adhesive performance, the hydrogel can recover after physical damage through its self-healing capability, allowing it to restore structural integrity without external intervention. Researchers believe this characteristic could improve durability in environments where repeated mechanical stress is unavoidable, reducing maintenance requirements and extending product life.
The material could have applications across several industries. In healthcare, it may support the development of advanced wound dressings, surgical adhesives, wearable medical devices and tissue engineering solutions that require reliable performance on wet biological surfaces. In robotics, particularly soft robotics, the hydrogel could enable more flexible and resilient components capable of operating in underwater or high moisture environments. The researchers also highlighted potential uses in marine engineering, industrial sealing and deep sea exploration equipment.
The project also reflects a broader trend in scientific research where artificial intelligence is increasingly being used to accelerate discoveries beyond software applications. Rather than relying solely on conventional trial and error experimentation, researchers are using AI to identify promising material combinations, predict performance and reduce development timelines. Similar approaches are beginning to influence areas such as battery technology, semiconductor development and pharmaceutical research.
While further validation will be required before commercial deployment, the study highlights the growing role of AI as a research partner in materials science. As machine learning becomes more deeply integrated into scientific workflows, researchers expect AI assisted material discovery to shorten innovation cycles and enable the development of next generation products that combine performance, durability and sustainability.