New research identifies four distinct health pathways linked to Alzheimer’s disease

A large study of electronic health records has uncovered four common patterns of medical conditions that tend to unfold in the years leading up to an Alzheimer’s disease diagnosis. These patterns—centered on mental health issues, brain disorders, cognitive decline, and vascular problems—emerged as distinct routes that people often follow before developing Alzheimer’s, offering new insights into how the disease may take shape over time.

The research, published in the journal eBioMedicine, suggests that combinations of health problems and the order in which they appear may carry more predictive power than any one condition alone. By identifying the most common multi-step pathways that precede Alzheimer’s, the study provides a new framework for recognizing who may be at elevated risk earlier and more accurately than existing models.

Alzheimer’s disease is a progressive brain disorder that affects memory, thinking, and behavior. It is the most common form of dementia and typically strikes older adults, gradually eroding their ability to live independently. More than 6.7 million people in the United States currently live with Alzheimer’s, and that number is expected to nearly double by 2050. Despite decades of research, there is still no cure, and current treatments only modestly slow progression. Much of the effort to reduce Alzheimer’s burden has focused on identifying people at high risk early enough to intervene—before major cognitive damage occurs.

Scientists have long known that certain conditions like depression, heart disease, and diabetes are associated with greater Alzheimer’s risk. But many past studies looked at these risk factors in isolation, without considering how they might unfold together across time. This new research takes a different approach by examining the sequences in which diseases appear in people’s medical records before they are diagnosed with Alzheimer’s. Rather than just focusing on single conditions, the researchers tried to map out the typical paths people follow, step by step, on the way to Alzheimer’s.

“We were interested in understanding the healthcare pathway people have before a diagnosis of Alzheimer’s disease. Electronic health records help us identify these pathways over time,” said study author Tim Chang, an assistant professor and Augustus S. Rose Chair in Neurology at UCLA.

The researchers analyzed longitudinal health data from nearly 25,000 patients in the University of California Health Data Warehouse, which collects records from six academic medical centers. These patients were between the ages of 65 and 90 and had multiple medical visits across several years. Alzheimer’s diagnoses were identified using standardized billing codes. The researchers then used statistical modeling and advanced techniques from data science to align and compare the sequences of conditions that occurred before Alzheimer’s onset.

They found that 5762 of the Alzheimer’s patients had at least one multi-step disease pathway made up of three or more health conditions that led up to their diagnosis. Using clustering methods to group similar sequences, they identified four major trajectory types. Each one was defined by a central condition and showed distinct patterns of progression.

One of the pathways centered on mental health, particularly depression. People in this group often had anxiety, high blood pressure, diabetes, or cholesterol issues before being diagnosed with depression. Depression then tended to occur several years before Alzheimer’s. This group was more likely to include women and Hispanic individuals, and their records often showed memory-related symptoms appearing one to three years before Alzheimer’s onset.

Another large group followed a pattern marked by brain disorders classified under encephalopathy—broadly defined as disorders of brain function. Conditions like encephalopathy and cerebrovascular disease appeared more quickly before Alzheimer’s diagnosis in this group, and these individuals had the shortest time between their central diagnosis and the onset of Alzheimer’s. This cluster was also associated with higher mortality, meaning patients tended to die sooner after their Alzheimer’s diagnosis than those in other groups.

The third trajectory was dominated by people who had previously been diagnosed with mild cognitive impairment or other degenerative nervous system diseases. This group largely followed what might be considered the classic path of cognitive decline that eventually meets the threshold for an Alzheimer’s diagnosis. These patients had longer medical histories, and their diagnoses often included minor neurological or systemic problems years before Alzheimer’s was identified.

The final group followed a vascular pathway. These patients tended to have cardiovascular-related conditions such as hypertension, cerebrovascular disease, and anemia, as well as musculoskeletal issues like joint pain. They also had the longest medical records and the highest number of comorbidities, indicating a high overall disease burden. Many of them followed stepwise trajectories that included hypertension, then cerebrovascular problems, and finally dementia.

To determine whether these trajectories actually predicted Alzheimer’s risk better than individual conditions, the researchers tested their findings in a separate control group that did not have Alzheimer’s. They found that people who followed any of the multi-step trajectories were significantly more likely to eventually be diagnosed with Alzheimer’s than people who had one or two of the same conditions but not in that specific sequence. In many cases, the increased risk from following a particular path was higher than the risk conferred by any single condition in the path.

The researchers also examined whether these pathways were likely to reflect causal relationships. Using a machine learning method called causal structural learning, they estimated the directionality of links between conditions. Some pathways showed stronger evidence of step-by-step progression. For instance, in the encephalopathy cluster, diagnoses like kidney failure were often followed by brain disorders, which in turn were followed by Alzheimer’s. This suggests that systemic health issues may contribute to neurological decline.

To ensure that these findings were not specific to the University of California dataset, the researchers repeated the analysis in the All of Us Research Program, a national study with a more diverse population. They applied the same methodology and found that the four clusters were largely replicable in this independent sample. Nearly 90% of the Alzheimer’s patients in the All of Us cohort could be assigned to one of the previously identified disease trajectories. Most of the stepwise sequences were again found to be predictive of future Alzheimer’s diagnoses.

The findings provide evidence that “the order or pattern of when diagnoses are made may influence a person’s risk for Alzheimer’s disease,” Chang told PsyPost.

Although the results are promising, the researchers caution that their study has limitations. The data came from electronic health records, which can miss diagnoses made outside the health system or be affected by variations in how doctors code conditions. The team used strict inclusion criteria to improve accuracy, but it’s possible that some cases were missed or misclassified.

Also, because the dataset excluded people over 90, the findings may not generalize to the oldest adults, who are at the highest risk of Alzheimer’s. Finally, the study relied on observed associations rather than biological markers, so while the trajectories are informative, they don’t prove that one condition causes another.

However, the findings still have important implications for public health and clinical care. By identifying common patterns that lead up to Alzheimer’s, doctors may be able to spot people at higher risk earlier and target them for screening or interventions. For instance, someone with a history of depression followed by metabolic issues might be more closely monitored for signs of cognitive decline. The study also highlights the need to treat chronic conditions—such as cardiovascular disease and diabetes—not just in isolation, but in light of how they may interact and evolve over time to affect brain health.

The study, “Identifying common disease trajectories of Alzheimer’s disease with electronic health records,” was authored by Mingzhou Fua, Sriram Sankararaman, Bogdan Pasaniuc, Keith Vossel, and Timothy S. Chang.