By Dr. Frank Appiah
Faculty Member, School of STEM
Alzheimer’s disease is a neurodegenerative disease that typically affects adults 60 years and older. It is a disease that involves the breakdown and malfunction of the neurons in the brain. As a result, Alzheimer’s patients in advanced stages of this disease may not able to perform day-to-day functions, such as walking, talking and enjoying life in general.
Alzheimer’s Disease Takes a Toll on Everyone
According to the Alzheimer’s Association, Alzheimer’s disease affects about six million Americans and this number is projected to be 13 million by 2050. Today, Alzheimer’s is one of the top five leading causes of death in the U.S.
Alzheimer’s disease is also very expensive to manage. The Alzheimer’s Association notes that the cost is over $300 billion per year and that number is projected to rise to $1 trillion by 2050.
This disease takes a toll on those living with the disease, their caregivers and taxpayers. As a result, there is a growing need to stop this disease in its tracks or at least delay its occurrence.
A ton of research, manpower and dollars is being invested to discover ways to either delay the development of Alzheimer’s disease and/or develop a cure. One of the methods being utilized toward finding a cure is using data science tools like mixture models to identify potential candidates before they develop Alzheimer’s disease, so that they can plan better for the future.
What Is a Mixture Model in Data Science?
A mixture model helps us understand the number of subgroups that exists within a distribution. So in this particular case, using a mixture model on the data from participants who have not developed the disease yet helps researchers identify a subgroup, such as individuals who will remain cognitively intact (people with normal functioning neurons) and other subgroups with a higher risk of developing cognitive impairment in the future.
Related: ChatGPT and Its Use by Non-Traditional College Students
Collecting Alzheimer’s Disease Research Data
To understand how the brain undergoes many changes, Alzheimer’s disease research involves following groups of volunteers over long periods of time and examining them at regular intervals, such as every six or 12 months. Today, many institutions including the Alzheimer’s Disease Neuroimaging Initiative (ADNI) participate in research to collect data on the progression of the disease amongst volunteers/patients.
The volunteers for these studies are examined at various sites. The examination tools that are used include (but are not limited to):
- A cognitive test
- A brain scan
- Lumbar puncturing (a spinal tap)
Test scores in these examinations provide medical researchers with a hint of each volunteer’s neuropsychological state of mind, such as the ability to recall and memorize information.
The brain scan and lumbar puncturing provide biomarkers where medical researchers can look for indications of potential plaque formation on the brain. Based on these types of tests, biomarker results and the demographics of volunteers in the study, data analyses can then be performed to unlock a deeper understanding of Alzheimer’s disease, make progress towards the development of a potential cure and create better ways to manage this disease.
Related: The Future of the Metaverse: Separating Fact from Fiction
How the Mixture Model Can Be Applied toward Alzheimer’s Research
The data science mixture model has been applied to Alzheimer’s research data collected by various institutions across the United States and in other countries. For instance, I conducted a 2021 study, along with researcher Richard J. Charnigo, called “A Comparison of Methods for Predicting Future Cognitive Status: Mixture Modeling, Latent Class Analysis, and Competitors.”
In our study, we applied the mixture model to data from cognitively normal volunteers who enrolled in a longitudinal study at the University of Kentucky’s Alzheimer’s Disease Center.
At baseline, those volunteers were cognitively intact or they had not developed Alzheimer’s disease yet. They were part of the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) at the University of Kentucky, and they were followed throughout the rest of their lives until their autopsies. In our study, the data we used (called T scores obtained from administering a test battery to participants) came from the volunteers at enrollment, during yearly visits and after their deaths.
In our analyses, we used mixture models to identify three groups of individuals at baseline as high, medium and low risks to transition to Alzheimer’s in the future. Our use of the mixture model revealed that at baseline – when everyone has no Alzheimer’s disease – we can, in principle, identify who might develop this disease in the future.
Data Science Can Be Helpful in Determining Who May Develop a Disease in the Future
Alzheimer’s disease researchers have made us aware that their data collection methods may provide pathways to use non-invasive tests and help identify potential Alzheimer’s patients when they are still cognitively intact. Early identification of potential Alzheimer’s disease patients will buy time for patients to put their affairs in order and take proactive measures (such as exercising and eating a proper diet). These measures are known to delay the onset of Alzheimer’s disease according to a Journal of Nuclear Medicine article written researchers Stefan Teipel, Deborah Gustafson and others.
Small sample data challenges usually encountered in medical research include a lack of generalization to the larger population. Often, there are only a small number of volunteers that partake in these studies. But as we endeavor to increase our progress in finding a cure for Alzheimer’s disease, using non-invasive tests and a larger selection of participants from a broad spectrum of American society could help generalize research outcomes.
Comments are closed.