NIH-Funded Research Aims to Outsmart Lyme Disease with AI
A new research project, funded by the National Institutes of Health (NIH), is harnessing artificial intelligence to develop more effective treatments for Lyme disease. The team includes a cross-disciplinary group of experts: Bree Aldridge (Professor at the School of Medicine and School of Engineering), Trever Smith II (Research Assistant Professor at the School of Medicine), Hu, and Farhat.
The project’s goals are ambitious: to screen a broader range of chemical compounds, design new potential drugs, and, crucially, determine the precise mechanisms by which effective drugs selectively kill Lyme bacteria.
"If a new compound causes cell death patterns similar to a known cell wall-acting agent, it is inferred to have a similar mechanism."
The foundation of this work is DECIPHAER, an AI tool developed from Aldridge's prior research on tuberculosis. The tool uses a method called morphological profiling. This involves imaging bacterial cells after they have been treated with a compound to observe structural changes.
DECIPHAER employs a "guilt by association" algorithm. By comparing the structural damage caused by a new compound to the damage caused by known drugs, the AI can infer how the new compound works. For instance, if a compound makes a cell look like it was attacked by a cell wall-acting agent, it likely has a similar mode of action.
The team plans to take this a step further by integrating multi-omics data. This deeper layer of analysis will provide more detailed mechanistic insights. It will allow researchers to predict a compound's molecular effects from images and understand how it behaves under different conditions or in combination with other drugs.
This predictive capability is key. It may enable the team to design more effective Lyme disease drugs by understanding their biological impact before extensive lab testing. The ultimate goal is to turn data into a roadmap for safer, more targeted therapies.