October 24, 2022
CUNY scientists developed CODE-AE to aid in precision medicine.
Researchers have developed an AI model capable of evaluating drug compounds to predict how well they work on humans.
Dubbed CODE-AE, the system was able to find personalized drug treatments for more than 9,000 patients that could improve their conditions.
The model was designed by a research team from the City University of New York (CUNY), with their findings published in Nature Machine Intelligence.
"Our new machine learning model can address the translational challenge from disease models to humans,” said Dr. Lei Xie, the paper’s author and professor of computer science, biology and biochemistry at the CUNY Graduate Center and Hunter College.
“CODE-AE uses biology-inspired design and takes advantage of several recent advances in machine learning. For example, one of its components uses similar techniques in deepfake image generation.”
Finding drug compounds with optimal efficacy specific to patients is a challenge for most clinical trials. Early drug testing on humans is unethical and could even be harmful. Tissue or cell models are used to test the therapeutic effects of drug compounds as surrogates for patients.
Testing toxicity and drug efficacy in humans isn’t always feasible. Drug effects in disease models aren’t always accurate when it comes to predicting how well the drug molecules will work on real patients.
Another challenge is how an ML model ‘learns’ if it doesn’t have enough patient data.
"Although many methods have been developed to utilize cell-line screens for predicting clinical responses, their performances are unreliable due to data incongruity and discrepancies," said You Wu, CUNY Ph.D. student and the paper’s co-author.
“CODE-AE can extract intrinsic biological signals masked by noise and confounding factors and effectively alleviated the data-discrepancy problem.”
The AI model helps predict human responses from patient-specific drug compound molecules simply by looking at cell-line compound screens.
In the future, scientists will find a way to use models like CODE-AE to predict the effect of novel compounds based on its concentration and how it metabolizes in a patient’s body.