AI’s Potential to Spot Alzheimer’s Signs in Health Data 

AI’s Potential to Spot Alzheimer’s Signs in Health Data. Credit | Getty Image
AI’s Potential to Spot Alzheimer’s Signs in Health Data. Credit | Getty Image

United States: Researchers may eventually be able to identify individuals from medical data who are more prone to get Alzheimer’s disease later in life by using artificial intelligence (AI) computer algorithms. 

 An NIA-funded study suggests that this may be accomplished by teaching specific self-educating algorithms or machine learning systems to identify dangers from electronic health records.  

Two-Step Methodology Revealed 

The underlying genetic activity and metabolic pathways that increase an individual’s risk can then be determined using the data. These results, which were published in Nature Aging, imply that this novel two-step method may aid in a better understanding of the risks connected to each individual’s case of Alzheimer’s for both researchers and caregivers. 

AI’s Potential to Spot Alzheimer’s Signs in Health Data. Credit | Getty Image
AI’s Potential to Spot Alzheimer’s Signs in Health Data. Credit | Getty Image

AI Algorithms in Healthcare Data 

Under the direction of experts at the University of California, San Francisco, the researchers initially taught computers using the blood test results, illness diagnoses, and other significant data that were maintained in the electronic health records of 250,545 control subjects and 749 Alzheimer’s patients. To be more precise, the researchers employed random forest algorithms—a class of computer systems that can identify patterns in data. 

Initial Findings and Accuracy 

Initial tests that are suggested that the programs are accurate at predicting Alzheimer’s risk from electronic health records and the programs are almost 70 percent  accurate at showing Alzheimer’s diagnosis seven years before one happened and about 80 percent accurate at predicting one year in advance. 

And surprisingly these predictions improved to gender, race and ethnicity and visit-related information, such as first visit age and years in the health system was incorporated into the programs. 

Expanded Predictive Factors 

Subsequent research and testing revealed that a number of variables, including high blood fat levels (hyperlipidemia), arthritis, and congestive heart failure, were consistently associated with a higher risk of Alzheimer’s disease, independent of how long before the variables were noted in the records. Other factors that were significant for predicting a diagnosis three years ahead of time were back pain, osteoporosis, and dizziness; moderate cognitive impairment and vitamin D insufficiency seemed to be important for predicting a diagnosis one year ahead of time.