Deep neural network model can accurately predict the brain age of healthy patients

A research exhibits {that a} deep neural community mannequin can precisely predict the mind age of wholesome sufferers primarily based on electroencephalogram knowledge recorded throughout an in a single day sleep research, and EEG-predicted mind age indices show distinctive traits inside populations with completely different illnesses.

The research discovered that the mannequin predicted age with a imply absolute error of solely 4.6 years. There was a statistically vital relationship between the Absolute Mind Age Index and: epilepsy and seizure issues, stroke, elevated markers of sleep-disordered respiration (i.e., apnea-hypopnea index and arousal index), and low sleep effectivity.

The research additionally discovered that sufferers with diabetes, despair, extreme extreme daytime sleepiness, hypertension, and/or reminiscence and focus issues confirmed, on common, an elevated Mind Age Index in contrast with the wholesome inhabitants pattern.

In keeping with the authors, the outcomes display that these well being situations are related to deviations of 1’s predicted age from one’s chronological age.

Whereas clinicians can solely grossly estimate or quantify the age of a affected person primarily based on their EEG, this research exhibits a synthetic intelligence mannequin can predict a affected person’s age with excessive precision. The mannequin’s precision allows shifts within the predicted age from the chronological age to specific correlations with main illness households and comorbidities. This presents the potential for figuring out novel medical phenotypes that exist inside physiological alerts using AI mannequin deviations.”

Yoav Nygate, Research Lead Creator and Senior AI Engineer at EnsoData, American Academy of Sleep Medication

The researchers skilled a deep neural community mannequin to foretell the age of sufferers utilizing uncooked EEG alerts recorded throughout medical sleep research carried out utilizing in a single day polysomnography. The mannequin was skilled on 126,241 sleep research, validated on 6,638 research, and examined on a holdout set of 1,172 research. Mind age was assessed by subtracting people’ chronological age from their EEG-predicted age (i.e., Mind Age Index), after which taking absolutely the worth of this variable (i.e., Absolute Mind Age Index). Analyses managed for components reminiscent of intercourse and physique mass index.

“The outcomes on this research present preliminary proof for the potential of using AI to evaluate the mind age of a affected person,” stated Nygate. “Our hope is that with continued investigation, analysis, and medical research, a mind age index will someday grow to be a diagnostic biomarker of mind well being, very similar to hypertension is for dangers of stroke and different cardiovascular issues.”

The analysis summary was revealed not too long ago in a web based complement of the journal Sleep and will probably be introduced as a poster starting June 9 throughout Digital SLEEP 2021. SLEEP is the annual assembly of the Related Skilled Sleep Societies, a three way partnership of the American Academy of Sleep Medication and the Sleep Analysis Society.