UC San Francisco researchers have utilized machine learning to develop a method for predicting Alzheimer’s Disease up to seven years before symptoms appear. By examining patient records, they have identified high cholesterol and osteoporosis (especially in women) as significant indicators. This advancement highlights the potential of artificial intelligence in early disease detection and in gaining insights into the underlying biology of disease risks. The ultimate goal is to enhance the diagnosis and treatment of Alzheimer’s and other complex diseases.
Machine learning, a branch of artificial intelligence, focuses on creating algorithms and models that empower computers to learn from data and make informed predictions or decisions without explicit programming. It plays a crucial role in recognizing data patterns, categorizing information, and forecasting future events through supervised, unsupervised, and reinforcement learning techniques.
Alzheimer’s disease is a degenerative brain condition that leads to a progressive decline in cognitive functions over time. It is the most prevalent form of dementia, constituting 60 to 80 percent of dementia cases. While there is currently no cure for Alzheimer’s, certain medications can help alleviate symptoms.
The key predictors for Alzheimer’s included high cholesterol and osteoporosis in women. This research showcases the potential of artificial intelligence in identifying patterns within clinical data, which can then be applied to genetic databases for determining disease risk factors. The objective is to expedite the diagnosis and treatment of Alzheimer’s and other intricate diseases.
The study, led by Alice Tang, an MD/PhD student at UCSF, marks a significant milestone in leveraging AI for analyzing routine clinical data to not only detect risks at an early stage but also comprehend the associated biological mechanisms. The strength of this AI-driven approach lies in identifying risks based on combinations of diseases.
Published in the journal Nature Aging, the findings underscore the importance of early prediction and understanding of Alzheimer’s Disease, a debilitating condition that impacts millions of individuals, particularly women who represent a significant proportion of affected individuals.
By examining over 5 million patient records from UCSF’s clinical database, researchers were able to predict Alzheimer’s onset with 72% accuracy up to seven years in advance by identifying co-occurring conditions. Factors such as hypertension, high cholesterol, and vitamin D deficiency were predictive for both genders, while conditions like erectile dysfunction and enlarged prostate were specific to men. Osteoporosis emerged as a crucial predictor for women.
The study underscores the significance of combining diseases to enhance predictive modeling for Alzheimer’s. The association between osteoporosis and Alzheimer’s in women, as revealed through genetic databases, sheds light on the intricate relationship between bone health and dementia risk.
The researchers employed SPOKE, a specialized tool developed at UCSF, to delve into the molecular basis of the predictive model. By analyzing public molecular databases, they identified genetic links between high cholesterol and Alzheimer’s, as well as osteoporosis and Alzheimer’s in women. This approach holds promise for diagnosing challenging conditions like lupus and endometriosis.
Funded primarily by the National Institute on Aging, with additional support from the Medical Scientist Training Program and F30 Fellowship, this study exemplifies the potential of combining patient data with machine learning to predict Alzheimer’s onset and gain insights into the underlying biological mechanisms, as stated by Marina Sirota, the senior author of the study.
