New technology predicts AD-related cognitive decline 2 years in advance
A novel modeling system can forecast at-risk patients’ brain cognition scores, to determine their level of Alzheimer’s disease (AD)-cognitive deterioration, up to 2 years in advance. The technology may speed up the development of AD therapies by increasing patient recruitment for clinical trials.
New technology, developed by researchers at MIT (MA, USA), has been demonstrated capable of predicting the brain cognition scores of patients at risk of developing Alzheimer’s disease (AD), up to 2 years ahead of their clinical symptom presentation. The model may thus be able to determine if patients will experience clinically significant disease-related cognitive decline; this may aid patient recruitment for clinical trials of novel AD therapies and speed up the discovery of such drugs.
According to a 2018 report from the Pharmaceutical Research and Manufacturers of America (DC, USA), between 1998 and 2017, 146 potential AD drug candidates – that proposed to treat and/or prevent disease progression – failed to be developed. Over this same period, four drug candidates were approved for clinical use; however, these therapies were only symptomatic relievers.
Many researchers suggest that improved identification and recruitment of patients who in the early stages of AD progression – prior to symptom presentation – when therapies prove most efficacious, may greatly improve the drug discovery process for AD therapies.
For the first time, researchers at the MIT Media Lab (MA, USA) describe a machine-learning (ML) model that may aid scientists in identifying this important, narrowly-defined patient cohort.
The novel system operates according to two models; first, researchers trained a ‘population model’. This utilized data from the Alzheimer's Disease Neuroimaging Initiative (ADNI), the world’s largest AD clinical trial dataset. The ADNI comprises cognition scores, MRI scans and other biometric data on 1700 AD and non-AD patients, recorded at biannual doctor visits over 10 years.
The ‘population model’ was trained to observe AD-progression trends and utilize these to predict new patients’ data between their doctor visits.
A second, ‘individual model’, for new study participants, was also developed; this continually updated patients’ cognition score predictions by the ‘population model’, based on newly recorded data for each patient.
The overall system was able to accurately predict patients’ data up to 24 months in advance.
Researchers may be able to employ the technology to identify at-risk patients before the clinical presentation of AD symptoms and recruit these individuals for trials of AD therapies.
Oggi Rudovic, a researcher at MIT Media Lab, commented: “Being able to accurately predict future cognitive changes can reduce the number of visits the participant has to make, which can be expensive and time-consuming. Apart from helping develop a useful drug, the goal is to help reduce the costs of clinical trials to make them more affordable and done on larger scales.”