Abstract:
The brain-age gap is one of the most investigated risk markers for cross-disorder brain changes. While the field is beginning to recognize the importance of quantifying uncertainty of predictions for group-level inference, no models to date provide single-subject risk assessment as needed for clinical application. Here, we present a novel framework combining neural network models with conformal prediction theory which provides statistical guarantees with regard to single-subject uncertainty estimates. This approach enables the calculation of an individual’s probability for accelerated brain-aging. Building on this, we show empirically in a sample of N=16,903 participants that 1. our model outperforms existing brain-age models providing a lower or comparable error, 2. the statistical guarantees regarding uncertainty estimation indeed hold for every participant, and 3. that the derived individual probabilities for accelerated brain-aging are associated with mental health diagnoses, depressive symptom severity, body-mass index, and Alzheimer’s Disease.

Start: December 16th, 2022 at 13:00 (s.t.)

Link: https://wwu.zoom.us/j/63385671839?pwd=TFV3dHJ2aHJqZGYxSGxNdmRtQ1dMQT09