Generalizable and robust deep learning algorithm for atrial fibrillation diagnosis across geography, ages and sexes.

To drive health innovation that meets the needs of all and democratize healthcare, there is a need to assess the generalization performance of deep learning (DL) algorithms across various distribution shifts to ensure that these algorithms are robust. This retrospective study is, to the best of our knowledge, an original attempt to develop and assess the generalization performance of a DL model for AF events detection from long term beat-to-beat intervals across geography, ages and sexes. The new recurrent DL model, denoted ArNet2, is developed on a large retrospective dataset of 2,147 patients totaling 51,386 h obtained from continuous electrocardiogram (ECG). The model's generalization is evaluated on manually annotated test sets from four centers (USA, Israel, Japan and China) totaling 402 patients. The model is further validated on a retrospective dataset of 1,825 consecutives Holter recordings from Israel. The model outperforms benchmark state-of-the-art models and generalized well across geography, ages and sexes. For the task of event detection ArNet2 performance was higher for female than male, higher for young adults (less than 61 years old) than other age groups and across geography. Finally, ArNet2 shows better performance for the test sets from the USA and China. The main finding explaining these variations is an impairment in performance in groups with a higher prevalence of atrial flutter (AFL). Our findings on the relative performance of ArNet2 across groups may have clinical implications on the choice of the preferred AF examination method to use relative to the group of interest.

The rapid developmental rise of somatic inhibition disengages hippocampal dynamics from self-motion.

Early electrophysiological brain oscillations recorded in preterm babies and newborn rodents are initially mostly driven by bottom-up sensorimotor activity and only later can detach from external inputs. This is a hallmark of most developing brain areas, including the hippocampus, which, in the adult brain, functions in integrating external inputs onto internal dynamics. Such developmental disengagement from external inputs is likely a fundamental step for the proper development of cognitive internal models. Despite its importance, the developmental timeline and circuit basis for this disengagement remain unknown. To address this issue, we have investigated the daily evolution of CA1 dynamics and underlying circuits during the first two postnatal weeks of mouse development using two-photon calcium imaging in non-anesthetized pups. We show that the first postnatal week ends with an abrupt shift in the representation of self-motion in CA1. Indeed, most CA1 pyramidal cells switch from activated to inhibited by self-generated movements at the end of the first postnatal week, whereas the majority of GABAergic neurons remain positively modulated throughout this period. This rapid switch occurs within 2 days and follows the rapid anatomical and functional surge of local somatic GABAergic innervation. The observed change in dynamics is consistent with a two-population model undergoing a strengthening of inhibition. We propose that this abrupt developmental transition inaugurates the emergence of internal hippocampal dynamics.