Researchers developed a machine learning model trained on a large amount of electronic health record data for finding risk factors that can predict Alzheimer’s disease. Machine learning model is a computer program that can find patterns in or make predictions about previously unseen data, after people trained it with a set of data. They identified weakened bones in women and chest pain in men as the top risk factors that can predict Alzheimer’s development seven years before the diagnosis.

Scientists have been finding tools and methods to predict Alzheimer’s disease before the symptoms appear, such as biological markers, also called biomarkers.

To identify the clinical conditions that can predict Alzheimer’s, in a recent study, a team of researchers used people’s electronic health record data, from the University of California, San Fransisco (UCSF) Medical Center, to set up a machine learning model. 

To train the model and use it to evaluate individuals’ data, the researchers included health records of at least seven years from 749 patients with Alzheimer’s and 250,545 individuals without the condition. They used 70% of the individuals’ data for training, and 30% of the data for finding the predictors of Alzheimer’s. 

The risk factors of Alzheimer’s may be different between men and women. The researchers analysed the data in men and women groups separately to identify the gender-specific risk factors of the disease. They identified that weakened bones, also known as osteoporosis, was the top clinical condition as a risk factor that predicts Alzheimer’s development in woman. It means that women with weakened bones progress to Alzheimer’s faster than those without the condition. Chest pain was the top condition among men. 

To validate these findings, the researchers included and analysed electronic health records from a wider database, University of California Data Discovery Platform. Data on this platform also showed that weakened bones and chest pain were the top predictors of Alzheimer’s for women and men. In addition to the gender-specific predictors, the team validated high level of blood lipid was the top predictor of Alzheimer’s in all individuals, regardless of men or women. 

These findings can help people at risk of Alzheimer’s start seeking professional advice on prevention and early diagnosis of the disease. This also helps researchers identify at-risk individuals for follow-up examinations, or including them in clinical studies that investigate preventive approaches. 

The researchers also showed the power of using a large amount of information stored in the electronic health records together with current technology, machine learning. To make sure the machine learning model is reliable and effective to use in the future, they should continuously update and train this model, because demographic and societal information of people in the database change over time. 

This study was published in Nature Aging. Image credit: Canva

Got any comments or questions on this article? Please leave comments down below, or reach out to me on LinkedIn, X (Twitter), or the CONTACT ME page, I would love to hear thoughts from you. 

Share this article: