Friday , May 14 2021

AI technology helps predict the future diagnosis of Alzheimer's disease

A study published in Radiology showed that a deep learning model predicted Alzheimer's disease with a specificity of 82% and a sensitivity of 100% about 6 years prior to diagnosis using fluorescence imaging studies with fluorescein PET brain.

"Differences in the pattern of glucose uptake in the brain are very subtle and diffuse" Jae Ho Sohn, MD, from the Department of Radiology and Biomedical Imaging at the University of California, San Francisco, said in a press release. "People are good at finding specific biomarkers of disease, but metabolic changes represent a more global and delicate process."

The researchers looked at whether a deep learning algorithm could predict the final diagnosis of Alzheimer's disease among patients who were brain fluorobromodeoxyglucose.

The researchers collected possible fluoride fluoride dextroroglycate PET images from the data set of the Alzheimer's Neuroimaging Initiative, which included 2,109 imaging studies of 1,002 patients and one retrospective independent test set, which included 40 imaging studies from 40 patients. They trained the deep learning algorithm in 90% of the data set, then checked it in the remaining 10% and in the independent test set. The model was rated with sensitivity, specialization and receiver functions.


A deep learning model predicted Alzheimer's disease with a specificity of 82% and a sensitivity of 100% about 6 years prior to diagnosis, using fluorescence imaging studies of 18 fluoredeoxyglucose PET of the brain, according to the findings of the study.


The deep learning algorithm has learned the metabolic patterns associated with Alzheimer's disease, according to the press release. The model achieved the area under the characteristic curve of the 0.98 receptor (95% CI, 0.94-1) when considering its ability to predict the final diagnosis of Alzheimer's disease in the independent test set, Sohn and colleagues said. The algorithm achieved a 82% specificity in 100% sensitivity to disease detection averaging 75.8 months prior to the final diagnosis.

After comparing the performance of the algorithm with that of radiological readers, researchers also found that the algorithm surpassed readers (57% sensitivity and 91% specialization) P <.05).

"We were very pleased with the performance of the algorithm and we could predict every case that progressed to Alzheimer's disease," Sohn said in the release. "If we diagnose Alzheimer's disease when all the symptoms have been manifested, it's too late to intervene. If we can identify it earlier, this is an opportunity for researchers to find better ways to slow down or even stop disease process ". – by Savannah Demko

Revelationsmall: Sohn reports grants from the UCSF. See the full study on the relevant financial disclosures of all other writers.

Source link