AI has already become a silent force behind the scenes in many of our interactions with the digital world . Catalyzed by the success of deep learning and increasingly accessible open source frameworks for data science and machine learning (e.g., TensorFlow, scikit-learn, and Keras), AI is on an unprecedented path toward democratization. However, democratization is also the first step toward commoditization. Standout AI companies will increasingly require access to proprietary datasets to build specialized models in vertical use cases.
Recursion Pharmaceuticals applies computer vision and machine learning to create a high-dimensional feature space in which to evaluate cellular health broadly across hundreds of disease states. Blake Borgeson and Nan Li offer a technical overview of how Recursion leverages cellular phenotyping for drug discovery. Blake and Nan first outline a brief history of the development and emergence of cellular morphological profiling and describe the principles supporting the ability to turn a biological question of interest into a data science problem. They then explore the deep learning tools and architectures Recursion uses to level up the effectiveness and flexibility of imaging-based approaches to biology, paying special attention to the power of combining convolutional neural networks with the company’s growing mass of rich proprietary cellular imaging data and to efforts that may in the future enable a transition away from brute force search toward predictive drug discovery. Blake and Nan conclude with a summary of early medical discoveries enabled by the platform and what this means for the future of healthcare.
Blake Borgeson is the cofounder and CTO of Recursion Pharmaceuticals, where he is leading the computational development of a drug discovery platform combining high-throughput experiments and machine learning that is capable of finding potential treatments for hundreds of diseases rapidly and in parallel. Previously, he researched and built real-time navigation software for surgical procedures at the M.E. Müller Institute in Bern, Switzerland and cofounded BuildASign.com, an ecommerce company that currently employs over 350 in Austin, Texas. Blake holds a PhD in bioinformatics from UT Austin’s Marcotte Lab, where his research used machine learning to exploit new experimental techniques in rapidly mapping protein complexes, and a BS in electrical engineering from Rice University.
Nan Li brings a mixed technology, investing and entrepreneurial background to Obvious Ventures. He is also an adjunct lecturer at Stanford on venture capital. Nan has been a venture investor and advisor for the past five years, working with companies applying technology toward solving big problems. Previously, he managed early-stage tech investments for Eric Schmidt’s Innovation Endeavors; led product, operations, and finance at Gigwalk, a mobile, crowdsourced data and analytics company funded by Greylock Partners and August Capital; was a VC at Bain Capital Ventures; was a management consultant at Bain & Company; and served as a PM at Microsoft. Nan holds a BSE in computer science engineering from the University of Michigan. He grew up in Detroit after emigrating from China. He enjoys music, photography, culture, puzzles, all Detroit sports, and general nerdom.
Help us make this conference the best it can be for you. Have questions you'd like this speaker to address? Suggestions for issues that deserve extra attention? Feedback that you'd like to share with the speaker and other attendees?
Join the conversation here (requires login)
©2017, O'Reilly Media, Inc. • (800) 889-8969 or (707) 827-7019 • Monday-Friday 7:30am-5pm PT • All trademarks and registered trademarks appearing on oreilly.com are the property of their respective owners. • firstname.lastname@example.org