In a recent review article published in the journal The intelligence of natural medicineresearchers at the California Institute of Technology discussed the involvement of artificial intelligence (AI) technologies in the creation of next-generation electronic skin (e-skin) and the analysis of health data collected by e-skin.
Background
E-skin is defined as integrated electronics that mimic and surpass the functions of human skin. E-skins are flexible and comfortable, and therefore can be placed at various locations on the robotic and human body to continuously and non-invasively record biosignals. E-skins are typically used as human-machine interfaces in smart bandages, bracelets, tattoo-like stickers, textiles, rings, masks, and personalized smart socks and shoes.
Although electronic skins have facilitated large-scale collection of health data through real-time recording, analysis and interpretation of health information remains time-consuming and difficult. Various machine learning algorithms have already been used in recent multimodal e-skin platforms for data analysis. Recent advances in big data and digital medicine have enabled artificial intelligence technologies to optimize electronic skin design and create personalized health profiles.
Application of artificial intelligence technologies in electronic appearance design
Reproducing the vital properties of human skin in artificial leather remains problematic, primarily due to many unsolved material issues. Artificial intelligence was designed to optimize material discovery and sensor design for autonomous re-engineering of new e-skin patches.
Natural materials such as cotton and silk are classic base materials for e-skin design due to their biocompatibility and economy. However, lack of stretchability and tunability are the main disadvantages of these materials. The synthesized soft materials showed promising results in accurate signal collection. However, these materials require further verification of biocompatibility and safety.
As a branch of artificial intelligence, machine learning can identify promising materials with targeted properties and optimize material synthesis. AI can be used to select and optimize manufacturing methods based on material properties. In addition, machine learning can be used for quality control during mass production as well as for e-skin design optimization.
Machine learning can explore kirigami designs for adaptive three-dimensional electronic surfaces and pixelated planar elastomeric membranes more effectively than mechanical simulations. This type of e-skin conformation is required for curved surfaces.
For noisy and discrete material experiment data with high variability, it is necessary to preprocess the data by interpolating missing data and realigning biased training sets. A more standardized data set and material pipeline is currently needed to accelerate material development and discovery.
Application of AI technologies to signal processing
Machine learning algorithms are able to analyze data quickly and robustly, and can improve data quality through signal noise suppression, multi-source separation, and artifact removal. Machine learning also has the ability to improve the sensitivity and specificity of e-skin sensors with respect to the target biomarker. For biochemical sensors involving enzymes with a narrow working range, machine learning algorithms can overcome signal saturation and calibrate nonlinear sensors in a dynamic test environment.
Motion artifacts are responsible for background noise in e-skin. Machine learning can facilitate the collection of accurate data by compensating for noise and imperfections in wearable sensors. Through iterative data-driven analysis of detection results, AI-based platforms can improve the detection capabilities of biosensors.
Electronic skins with artificial intelligence for human-machine interfaces
Artificial intelligence technologies play an extremely important role in bridging the gap between human-machine interactions. Artificial intelligence can rapidly analyze and interpret multimodal data obtained from e-skin patches to manipulate robotics and provide human assistance.
Artificial intelligence-powered haptic sensors used in skin-based human-machine interface electronic systems can quickly capture complex hand movements and transmit physical information to a computer system, facilitating associated robotics to perform various tasks such as breathing, shape and object detection. identification.
Robotic prosthetics designed to rehabilitate the movement of people with disabilities can use electronic skins to extract movement data and machine learning algorithms to analyze and control robotic operations.
Electronic skins with artificial intelligence for diagnosis and treatment of diseases
Electronic skin with artificial intelligence is a promising approach for highly accurate diagnosis of cardiac complications. Electronic skins with artificial intelligence can quickly detect small and progressive cardiovascular changes over time, which can facilitate early automatic diagnosis.
AI-powered electronic skins can be used to monitor stress hormone levels in real-time to predict mental health issues. AI-powered multimodal electronic skins have the potential to model risk associations and predict mental health outcomes by identifying previously unrecognized associations between health patterns and stress risk factors.
AI-powered electronic skins can be used to monitor multiple biological parameters, and machine learning algorithms can be used to analyze electronic skin-derived data for biomarker prediction. Electronic monitoring of drugs and metabolism on the skin can also facilitate personalized therapy. AI-powered electronic skins can be used to assess pharmacokinetics and pharmacodynamics to personalize drug dosage.
Data availability and security are major challenges associated with the clinical application of AI-based e-skins. Therefore, strict regulations are required for the adoption of AI-based models in medical practice. Additionally, AI-based models can make mistakes. It is therefore essential to ensure the extent to which people can trust the predictions generated by artificial intelligence.