Exploring Innovative Therapies for Post-Menopausal Hot Flush Relief

Abstract

Menopause is an inevitable stage in normal human aging, affecting the quality of life of millions of individuals with a uterus all over the world. The most immediate and patient-reported 'unbearable' symptoms are hot flushes (HFs), generating a large decrease in the quality of life. Among the current treatments of HFs, only estrogen therapy with or without progestins has satisfactory efficacy. However, estrogen/progesterone therapies have significant side effects, stimulating uterine and breast cell proliferation. Despite extensive efforts to develop novel therapies, the lack of animal model(s) that naturally recapitulate the symptoms and pathophysiology of human menopause has had a tremendous negative impact on the success of these efforts, as only primates exhibit menopausal HFs. We have developed an advanced translational animal model from the old world, which undergoes a menopausal process hormonally identical to humans and includes methods to detect HFs with non-invasive thermal imaging. The imaging data is autonomously analyzed using Convolutional Neural Network (CNN) models programmed in the Python programming language. The result is an efficient CNN algorithm capable of detecting primate thermal facial features and automated HF detection, with a computational load allowing analysis to parallel image acquisition (i.e., 24 hours of imaging data takes less than 25 minutes to process). The entire image processing pipeline operates automatically using deep learning techniques. The algorithm has been extensively trained to effectively consider various parameters of an image, including body position, lens distortions, and facial features, to accurately determine the temperature across the monkey's face. By analyzing temperature variations across different parts of the monkey's face, our algorithm can accurately detect and identify key facial features that play crucial roles in heat dissipation with negligible false positives/negatives. This research heavily relies on data analysis due to the millions of infrared images collected by 24/7 video imaging of the subjects. Through this comprehensive analysis, we will be able to provide novel observations on the occurrence and patterns of HFs in monkeys, shedding light on potential underlying factors and implications. Our approach demonstrates the efficacy of deep learning and CNNs in autonomous systems for extracting meaningful information from complex datasets, providing a solid foundation for further research in this field. This approach will help evaluate the efficacy of novel strategies to prevent HFs.