Application of Deep Learning of Artificial Neural Networks in Monitoring the Sight System State in Long-Duration Space Missions

Abstract

Integration of artificial intelligence in cosmonaut's health monitoring system is a top priority trend in space medicine aimed to enhance crew safety in long-duration space missions. The cutting-edge diagnostic tools aboard the ISS Russian segment such as digital funduscopy, and optic CT of the retina and optic disk allow of piling up statistics of pre- in- and post-flight examinations for calculating the basic block for AI applications. Timely detection of visual dysfunctions in autonomous remote space missions with the help of AI algorithms will increase efficiency of prevention and therapy. Analysis of the experience with convolutional neural networks inophthalmology suggested a CNW architecture for detecting eye grounds pathologies by DF automatic analysis. A developed binary classifier demonstrated sensitivity, specificity and precision of 88.6%, 85.2% and 87%, respectively. In remote missions, the healthcare and therapy system should embody the concept of intelligence telemedicine circuit with artificial neural networks being the pivot. Digital ophthalmology with integrated CNW techniques can be useful also for early diagnostics of eye grounds pathologies within the program of exploration crews screening.