Amid Boston Overdose Crisis, a Pair of Harvard Students Are Bringing Narcan to the Red Line


At First Cambridge City Council Election Forum, Candidates Clash Over Building Emissions


Harvard’s Updated Sustainability Plan Garners Optimistic Responses from Student Climate Activists


‘Sunroof’ Singer Nicky Youre Lights Up Harvard Yard at Crimson Jam


‘The Architect of the Whole Plan’: Harvard Law Graduate Ken Chesebro’s Path to Jan. 6

HMS Researchers Use Machine Learning to Recommend Drugs for Alzheimer’s Disease Clinical Trials

Harvard researchers at the Medical School are using machine learning to recommend drugs for Alzheimer's disease clinical trials.
Harvard researchers at the Medical School are using machine learning to recommend drugs for Alzheimer's disease clinical trials. By Jonathan G. Yuan
By Christie K. Choi and Yuen Ting Chow, Crimson Staff Writers

Harvard Medical School researchers developed a strategy that uses machine learning to recommend drugs as candidates for clinical trials for Alzheimer’s disease, according to an article published in the journal Nature Communications last month.

A team of researchers from the Medical School’s Laboratory of Systems Pharmacology and Massachusetts General Hospital created the new method, known as Drug Repurposing In Alzheimer’s Disease. DRIAD uses gene expression data obtained from tissues affected by Alzheimer’s and drug signatures used to evaluate the progression of disease.

The research team hypothesized that if a drug signature can effectively predict disease progression, the drug can be used to alter disease progression, according to Clemens B. Hug, a research associate at the Medical School and the co-first author of the article.

One of the paper’s authors, Mark W. Albers, an assistant professor of neurology at HMS, said DRIAD was built based on data from 1,500 patients experiencing Alzheimer’s disease at different stages.

With this data, the team used machine learning techniques to develop a precise statistical analysis.

“There might be some hidden patterns in the data that traditional correlations can’t reveal,” said Artem Sokolov, an HMS instructor and director of informatics and modeling.

Sokolov said there are millions of potential drugs that have been generated and thousands that have been FDA-approved.

“There is no possible way to evaluate all of them in the clinical trial," Sokolov said. "And our hope is that DRIAD can help bubble certain drugs up to the top by saying ‘the genes that these drugs perturb tend to be associated with Alzheimer’s.’”

The group added the drugs to human brain cells in culture and looked for the genes that changed in response to each drug, according to Albers. Those genes were then fed into DRIAD to see if they aligned well with disease progression.

After applying DRIAD to current drugs, the results revealed that top-performing drugs target Janus kinase proteins, which promote pro-inflammatory signals.

“The idea is that stopping these inflammatory signals can help patients to reduce the symptoms,” Hug said.

Still, the only way to determine the effectiveness of those drugs in treating Alzheimer’s is through clinical trials, according to Sokolov.

Albers said one of the drugs — baricitinib — will be the focus of a Massachusetts General Hospital clinical trial this summer.

Sokolov cited an emerging view that Alzheimer’s is not simply one disease, but rather a presentation of multiple conditions that cause neuronal death, such as inflammation or metabolic disruptions.

“It's naive to think that there would be a single drug that can cure all possible underlying causes,” Sokolov said. “The future steps for our project is to begin untangling some of these mechanisms to try to understand why the neurons are dying, but also try to match up specific treatments for specific causes.”

The research created a website where the public can test out DRIAD themselves by inputting gene sets.

Want to keep up with breaking news? Subscribe to our email newsletter.

ResearchHarvard Medical SchoolMedicine

Related Articles

New Machine Learning Model May Help Predict Dementia, Harvard Medical School Researchers Report