Deep-Learning Algorithm Detects Alzheimer’s Disease with 90.2% Accuracy Up to Five Years Before Diagnosis
Researchers have developed a deep-learning algorithm that can detect Alzheimer’s disease with 90.2% accuracy, up to five years before a diagnosis. This algorithm is based on neuroimaging data and could lead to more accurate diagnoses. The research team, from Massachusetts General Hospital, explored the possibility of using routinely collected brain MRIs to detect dementia, a feat that few studies had accomplished in real-world clinical settings.
Matthew Leming, PhD, a research fellow at Mass General’s Center for Systems Biology, said, “Alzheimer’s disease typically occurs in older adults, and so deep learning models often have difficulty in detecting the rarer early onset cases. We addressed this by making the deep learning model ‘blind’ to brain features that it finds overly associated with the patient’s listed age.”
The research involved 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. The predictive model based on data achieved a score of 0.939 when diagnosing a patient with Alzheimer's disease and 0.906 one year prior to diagnosis. At three years prior to diagnosis, the performance of the model decreased slightly, yielding a score of 0.884 for predicting Alzheimer's disease, and at five years prior to diagnosis, the score dropped further to 0.854.
The team also developed a knowledge-based model that used current scientific evidence, known risk factors for Alzheimer’s disease, and prescriptions for drugs approved to treat the condition and related dementias. This model was then compared with the data-driven model to see which was more effective in predicting Alzheimer’s disease at the time of diagnosis and one, three, and five years before diagnosis.
The data-driven model significantly outperformed the knowledge-driven model and identified multiple additional risk factors that the knowledge-based model did not, such as malaise, fatigue, muscle weakness, and mood disorders. It also found that women who receive preventive health care, including regular medical exams, gynecological exams, and mammogram screenings, had a lower risk of developing Alzheimer’s than their counterparts who do not receive the same care.
The researchers concluded that the model could help identify high-risk individuals for early informed preventive or prognostic clinical decisions.
0. “AI model using health records predicts Alzheimer's disease up to 5 years before diagnosis” McKnight's Senior Living, 1 Mar. 2023, https://www.mcknightsseniorliving.com/home/news/ai-model-using-health-records-predicts-alzheimers-disease-up-to-5-years-before-diagnosis/
1. “Artificial Intelligence Approach May Help Detect Alzheimer’s Disease From Routine Brain Imaging Tests” Neuroscience News, 4 Mar. 2023, https://neurosciencenews.com/ai-alzheimers-22712
2. “AI could detect Alzheimer's disease from brain scans” E&T Magazine, 3 Mar. 2023, https://eandt.theiet.org/content/articles/2023/03/ai-could-detect-alzheimers-disease-from-brain-scans/
3. “Alzheimer's disease may be predicted by using AI and patient medical records” Mental Daily, 27 Feb. 2023, https://www.mentaldaily.com/article/2023/02/alzheimers-disease-may-be-predicted-by-using-ai-and-patient-medical-records
4. “AI Tools Predict Alzheimer's Up to 5 Years Before Diagnosis” HealthITAnalytics.com, 27 Feb. 2023, https://healthitanalytics.com/news/ai-tools-predict-alzheimers-up-to-5-years-before-diagnosis