The Potential of AI in Early Detection of Alzheimer's

Scientists at the University of California San Francisco have discovered a method for using machine learning to analyze patient records to predict the onset of Alzheimer’s disease up to seven years in advance.

The two factors that affected the prediction the most were high cholesterol and osteoporosis, which weakens bones in women.

The research shows the potential of applying artificial intelligence (AI) to identify patterns in clinical data, which can subsequently be utilized to search through extensive genetic databases and ascertain the cause of that risk. The researchers believe that in the future, it will speed up the identification and management of complicated illnesses like Alzheimer’s.

This is a first step towards using AI on routine clinical data, not only to identify risk as early as possible but also to understand the biology behind it. The power of this AI approach comes from identifying risk based on combinations of diseases.

Alice Tang, Study Lead Author and MD/PhD Student, University of California San Francisco

The findings appear in Nature Aging.

Clinical Data and the Power of Prediction

Researchers have been looking for biological causes and early indicators of Alzheimer's disease for a long time. Alzheimer’s is a type of dementia that deteriorates memory and is progressive and ultimately fatal. Approximately two-thirds of the 6.7 million Americans who have Alzheimer's are female. The likelihood of contracting the illness rises with age. Although women typically outlive men, this does not entirely account for the higher rate of disease among females.

To find co-occurring conditions in patients diagnosed with Alzheimer’s at UCSF's Memory and Aging Center compared to people without the disease, the researchers used UCSF's clinical database of more than 5 million patients. They discovered that they could predict, with 72 % predictive power, who would develop the disease up to seven years prior.

Numerous factors were predictive in both males and females, such as high blood pressure, elevated cholesterol, and insufficient vitamin D. For men, enlarged prostates and erectile dysfunction were also predictive. However, osteoporosis was a particularly significant predictor for women.

This is not to say that every older woman who has bone disease will develop Alzheimer’s.

It is the combination of diseases that allows our model to predict Alzheimer's onset. Our finding that osteoporosis is one predictive factor for females highlights the biological interplay between bone health and dementia risk.

Alice Tang, Study Lead Author and MD/PhD Student, University of California San Francisco

A Precision Medicine Approach

The researchers used Scalable Precision Medicine Oriented Knowledge Engine (SPOKE), a specialized tool created at UCSF and developed in the lab of Sergio Baranzini, Ph.D., a Professor of Neurology and member of the UCSF Weill Institute for Neurosciences, to better understand the biology underpinning the model’s predictive power.

SPOKE is a database of databases that scientists can utilize to find trends and put therapeutic targets on the molecular level. It identified the widely recognized link between high cholesterol and Alzheimer’s disease due to a variation in the apolipoprotein E gene, or APOE4. However, when merged with genetic databases, it also discovered a connection—via a variation in a little-known gene called MS4A6A-—between osteoporosis and Alzheimer’s disease in females.

Ultimately, the researchers hope the method can be applied to other difficult-to-diagnose conditions, such as endometriosis and lupus.

This is a great example of how we can leverage patient data with machine learning to predict which patients are more likely to develop Alzheimer’s and also to understand the reasons why that is so.

Marina Sirota, PhD, Study Senior Author and Associate Professor, Bakar Computational Health Sciences Institute, UCSF

The National Institute on Aging provided the primary support for this study. The Medical Scientist Training Program provided additional support.

Journal Reference:

Tang, S. A., et al. (2024) Leveraging electronic health records and knowledge networks for Alzheimer’s disease prediction and sex-specific biological insights. Nature Aging. doi.org/10.1038/s43587-024-00573-8.

Source: https://www.ucsf.edu/