DEVELOPMENT AND COMMERCIALIZATION OF COMPUTATIONAL METHODS TO ACCELERATE DERMATOLOGICAL DISCOVERY

Engineering notes

Engineering notes will be added by the aipentium editorial team.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为大语言模型、生成式AI、ChatGPT相关技术、计算机视觉、深度学习等高价值论文补充中文说明。

Original abstract

The skin is the largest and outermost organ that surrounds the human body and provides a physical barrier that protects the more delicate internal tissues from the harsh surrounding environment. Over the course of evolution, specialized structures have also arisen in the skin, including sweat and sebaceous glands, nails and claws, and hair follicles. The hair follicle is a multi-lineage mini-organ that is a critical component in defining mammalian species. The gain of function derived from hair and fur benefited these species for millennia due to its insulative and protective properties, mechanosensing function, and ability to serve as camouflage. Current approaches to assess these fibers are generally non-quantitative or are low throughput due to technical limitations of ‘splitting hairs’. We developed a novel deep learning-based computer vision approach for the high throughput quantification of individual hair fibers at a high resolution. Our innovative computer vision tool can distinguish and extract overlapping fibers for quantification of multivariate features including length, width, and color to generate single-hair phenomes (shPhenome) of diverse conditions across the lifespan of mice. Using our tool, we explored the effects of hormone signaling, genetic modifications, and aging on hair follicle output. These analyses revealed novel hair phenotypes as a result of endocrinological, developmental, and aging-related alterations in the fur coats of mice. This tool was then adapted to capture more nuanced hair phenotypes, such as curvature, in a variety of mouse models harboring genetic perturbations. Commercialization of the deep hair phenomics technology presented a unique opportunity to pursue research translation, transforming fundamental biological research into marketable products. Much like the traditional academic route, the path to commercialization requires consistent and continuous iteration, testing of hypotheses, and allowing data to influence and lead subsequent actions. However, as opposed to the pursuit of knowledge, translating research into market-ready solutions requires a focus on consumer needs and the identification of market opportunities. Additionally, the shift in incentive structure requires commercial ventures to prioritize intellectual property protection, alter capital and resource allocation, and seek alternative fundraising sources. The entrepreneurial foundation built through this process culminated in the filing of a provisional patent and establishment of a startup company, kerat Inc. This journey illuminated the multifaceted challenges inherent in closing the gap between academic innovation and commercially viable goods, requiring simultaneous advancement of technical capabilities, adaptation to market dynamics, and fabrication of business infrastructure. Overall, the culmination of work presented in this dissertation sheds light on the importance of quantitatively assessing hair fiber properties to capture the biological phenomena that underlie this fundamental feature of mammalian skin. These findings also highlight the capabilities of this methodology for identifying perturbations in normal growth, potentially indicative of disease or dysfunction. Lastly, this showcased the potential of this technology’s commercial use cases in both the cosmetics and diagnostics industries.

Links and sources

Comments