Researchers at the University of California, San Francisco, have achieved a medical milestone by using artificial intelligence to predict Alzheimer’s disease up to seven years before symptoms appear. Published in Nature Aging on February 21, 2024, the study showcases the power of AI in identifying early indicators of Alzheimer’s from electronic health records, emphasizing high cholesterol and osteoporosis, especially in women, as key predictors.

The study, led by Alice Tang of UCSF’s Sirota Lab, utilized a machine learning algorithm to analyze vast clinical data, achieving an impressive 72% accuracy in early Alzheimer’s prediction. This approach not only identifies risk factors early on but also offers insights into the disease’s biology.

Key findings include the identification of high cholesterol, vitamin D deficiency, and, in women, osteoporosis as top predictors. The research also revealed gender-specific insights, with erectile dysfunction and enlarged prostate being significant for men. The team used UCSF’s advanced data analysis tool, SPOKE, to explore genetic connections, uncovering new pathways linking Alzheimer’s with other health conditions.

While acknowledging limitations, such as potential biases in health record data, the researchers are optimistic about applying this method to other complex diseases. This study represents a significant step forward in the early detection and understanding of Alzheimer’s, offering hope for earlier interventions and new research directions.