Amazon, UCLA announce recipients of gift awards for applications of AI in healthcare
The awards support four research projects exploring the intersection of AI and health care.
Amazon and the UCLA Samueli School of Engineering have announced four gift award recipients via the Science Hub for Humanity and Artificial Intelligence. These awards support projects that explore the practical applications of artificial intelligence (AI) in healthcare, emphasizing the potential of advanced technology to address global health issues.
UCLA and Amazon established the Science Hub in October 2021 to support interdisciplinary academic research, education, and outreach efforts in areas of mutual interest around AI and its benefits. This collaboration supports UCLA faculty and graduate student research with both gift and sponsored funding, emphasizing projects that confront humanity’s most significant challenges. The goal of the Science Hub is to find solutions that benefit society, with particular attention to matters of bias, fairness, accountability, and responsible AI.
The following four research projects are being supported:
Adrian Au, STAR (Specialty Training and Advanced Research) Program residency, The UCLA Retinal Biobank
“Age-related macular degeneration (AMD) presents a significant public-health challenge, causing substantial patient suffering and imposing substantial societal burdens. Recognized as a complex, polygenic disease, AMD has been the subject of numerous genome-wide association studies (GWAS) that have demonstrated specific genetic variations associated with increased risk,” Au wrote in the project abstract. “Nevertheless, these studies rely on outdated AMD criteria within homogeneous populations and employ low-resolution retinal-imaging techniques."
“Our proposal seeks to enhance our understanding of AMD genetics by establishing a robust genotype-phenotype database,” Au continues. “We intend to create a centralized repository for clinical, anatomic, and genomic data while developing convolutional neural networks capable of analyzing retinal images without being constrained by traditional AMD definitions.” This initiative holds the potential to enhance the quality of care provided to AMD patients.
Jonathan Kao, associate professor of electrical and computer engineering and principal investigator at the Neural Engineering and Computation Lab, “High-performance, non-invasive brain-machine interfaces using shared autonomy”
“Millions live with paralysis. But there are no widespread devices that significantly improve the quality of life for people with paralysis. One promising approach, brain-machine interfaces (BMIs), decode neural activity (reflecting thoughts and intentions) into actions, enabling users to use computers, move robotic arms, or communicate via speech. But a significant limitation is that the most effective BMIs require neurosurgery, limiting widespread use.
“A goal of our lab is to make effective non-invasive BMIs, where neural signals are recorded without neurosurgery. These signals have a significantly worse signal-to-noise ratio than invasive signals, meaning non-invasive BMIs are difficult to control. To overcome this, we use AI to increase performance. This AWS-UCLA award supports research into how to control robotic arms effectively with noisy and low-information inputs.”
Ricky Savjani, radiation oncology residency, “Elucidating the geography of cancers: How anatomical spatial distributions influence oncological outcomes”
“Over an entire career, an oncologist will treat thousands of patients with cancer,” Savjani wrote in his abstract. “Physicians gain expertise on what treatments work vs. what causes more harm to patients than good. However, this clinical experience is difficult to access, quantify, and teach. What if vital information from every cancer patient ever treated could be accessed instantly?"
“Together with Amazon, we are building an oncological visual-search database for a variety of solid tumors,” Savjani continues. “This transcends text-based spreadsheets and manuscripts to allow direct interrogation of clinical responses in an intuitive WebGL viewer. Our approach utilizes fast, deep-learning registration frameworks to harmonize individual patient data onto a common template. Clinicians will then be able to utilize this tool in real time to optimize treatment decisions together with patients.”
Ying Nian Wu, professor of statistics and Amazon Scholar, “Molecule design by latent-space energy-based modeling”
“Our work is about molecule design for drug discovery,” wrote Wu. “In drug discovery, it is of vital importance to find or design molecules with desired pharmacologic or chemical properties, such as high drug-likeness and binding affinity to a target protein."
“It is challenging to directly optimize or search over the drug-like-molecule space, since it is discrete and enormous, with an estimated size on the order of 1033 molecules,” Wu continues. “We propose a probabilistic generative model to capture the joint distribution of molecules and their properties. We also propose an algorithm to gradually shift the distribution to molecules with desired properties.”