Alzheimer’s disease still has limited treatments. But being able to predict when a person with mild cognitive impairment will transition to a more serious dementia can help provide individuals and their families with a timeline and trajectory of their disease, so they can prepare for the future and make financial and care decisions.

Precision Medicine at Work

The disease trajectory is the focus of an ongoing research project at Johns Hopkins’ Richman Family Precision Medicine Center of Excellence in Alzheimer’s Disease. Using electronic health record data from 140,000 patients compiled by the Johns Hopkins’ Precision Medicine Analytics Platform, a data analytics tool, investigators are developing two models to forecast patients’ Alzheimer’s disease trajectory. One model uses de-identified information from patients who come to the Johns Hopkins’ Memory and Alzheimer’s Treatment Center. The other model follows a more general population — patients over age 65 who have seen primary care physicians at Johns Hopkins Community Physicians’ offices in Maryland — data from whom could be used to identify those who could benefit from additional cognitive screening.

For the model that uses data from patients seen in the Memory and Alzheimer’s Treatment Center, the researchers have found that cognitive assessments, diagnoses for depression or other psychiatric conditions, and prescriptions for antidepressants are among the factors that appear to contribute to a patient’s future trajectory, says Roy Adams, an assistant professor of psychiatry and behavioral sciences working on the project.

The model following patients at Johns Hopkins Community Physicians is a different story. “In the primary care model, it’s a little bit more interesting,” he says. “We’re seeing more cardiovascular and metabolic diagnoses and lab results that are playing a role in the predictions. Those are known risk factors for dementia, but one of the important things we’ve found out about these models is that a different set of factors in these two patient populations are most useful.”

For now, the models incorporate information such as patient diagnoses and medications. But the long-term goal is to add in information such as imaging data, clinical notes and genetic screening evidence.

Researchers hope to bring the models into patient care within a year or two. “There is also the potential to use this information for targeting treatments to patients who may benefit,” Adams says, “but we want to do more investigation about how well this type of information predicts response to some of the pharmaceutical treatments for dementia that are most effective for patients with mild cognitive impairment.”

More information about Alzheimer’s disease.