Predictive Prowess: Alzheimer’s and Artificial Intelligence
A recent study published by the Department of Radiology and Biomedical Imaging at the University of California at San Francisco represents a promising breakthrough in research relating to early detection of Alzheimer’s disease. At the core of the study, however, is a familiar yet unlikely trend: artificial intelligence.
The research team developed an algorithm to read and interpret PET scan images with a particular emphasis on monitoring and detecting changes in glucose uptake over extended periods of time. Glucose monitoring has historically been an important predictive factor in formulating a diagnosis of Alzheimer’s. Healthy cells generally display high levels of glucose uptake, indicative of robust cell activity. Conversely, lower glucose uptake suggests cell inactivity or death, for example, as a result of Alzheimer’s.
The slow, progressive nature of Alzheimer’s has historically rendered it difficult for radiologists to observe the subtle changes in glucose levels until symptoms had reached a stage at which they were no longer meaningfully reversible. The team at UCSF tailored the algorithm to detect subtle features that were imperceptible to the human eye.
To achieve this, the algorithm was fed thousands of PET scan images from thousands of patients at all stages of cognitive impairment, from no impairment through to late-stage Alzheimer’s. Over time, the algorithm learned to discern between the particular features of a given scan which were of assistance in predicting the eventual onset of Alzheimer’s and those which were not. At the conclusion of the study, the algorithm had correctly predicted the onset of Alzheimer’s in more than 92% of cases. Importantly, the algorithm was able to predict the onset of Alzheimer’s, on average, more than six years before the symptoms constituting a typical diagnosis had manifested.
Leaving aside the obvious benefits relating to treatment and reversibility, early detection of Alzheimer’s could stand to have numerous applications in the context of succession and estate planning. For example, a predictive diagnosis could spur a testator to take steps to implement a proper estate plan well before his or her capacity to do so could become a concern. In turn, the testator would have the security that their plan of succession would be carried out according to his or her instructions, reducing the risk of contentious post-death litigation.
Thanks for reading.
Please feel free to check out the following blogs on related topics: