In the latest attempt to detect signs of Alzheimer’s disease, a research fellow at MGH's Center for Systems Biology and an investigator at the Massachusetts Alzheimer's Disease Research Center, and his colleagues used deep learning—“a type of machine learning and artificial intelligence (AI) that uses large amounts of data and complex algorithms to train models.”
The study published in PLOS ONE revealed that the novel deep-learning model, Multi-Confound Regression Adversarial Network (MUCRAN) was developed for Alzheimer's disease detection based on data from brain magnetic resonance images (MRIs) collected from patients with and without Alzheimer's disease who were seen at MGH before 2019, according to a press release.
This is not the first study to use AI to detect Alzheimer’s disease. In September 2022, a study published in The Lancet Digital Health, a study conducted by Carol Y Cheung and colleagues described a deep learning model to detect the disease using retinal photographs. For the study, the researchers trained a supervised deep learning algorithm using six retrospective datasets from Singapore, Hong Kong and the UK. In internal validation, the model showed 83.6% accuracy and 82 specificity.
MUCRAN was tested across five datasets including MGH post-2019, Brigham and Women's Hospital pre- and post-2019, and outside systems pre- and post-2019 to check its accuracy in the detection based on real-world clinical data, regardless of hospital and time, reveals the press release.
The research included 11,103 images from 2,348 patients at risk for Alzheimer's disease and 26,892 images from 8,456 patients without Alzheimer's disease. Across all datasets, MUCRAN performed better than the comparative models, both in the age-matched sample with 79.2% accuracy and the whole test set with 90.2% accuracy.
Among the main findings, an important one was its ability to detect Alzheimer's disease regardless of other variables, such as age. "Alzheimer's disease typically occurs in older adults, and so deep learning models often have difficulty in detecting the rarer early-onset cases," Leming said in a price release. "We addressed this by making the deep learning model 'blind' to features of the brain that it finds to be overly associated with the patient's listed age."
Another common challenge in the detection, specifically using real-world data, is dealing with data that are very different from the training set. For instance, a deep learning model trained on MRIs from a scanner manufactured by one company may fail to recognize MRIs collected on a scanner manufactured by another, says Leming.
Although there have been several deep learning studies for Alzheimer’s detection from brain MRIs, this study has made significant leaps towards performing the research in real-world clinical settings as opposed to perfect laboratory settings, he adds. “Our results—with cross-site, cross-time, and cross-population generalizability—make a strong case for clinical use of this diagnostic technology.”