Memory deficits are a hallmark of many different neurological and psychiatric conditions. The Rey–Osterrieth complex figure (ROCF) is the state-of-the-art assessment tool for neuropsychologists across the globe to assess the degree of non-verbal visual memory deterioration. To obtain a score, a trained clinician inspects a patient’s ROCF drawing and quantifies deviations from the original figure. This manual procedure is time-consuming, slow and scores vary depending on the clinician’s experience, motivation, and tiredness. Here, we leverage novel deep learning architectures to automatize the rating of memory deficits. For this, we collected more than 20k hand-drawn ROCF drawings from patients with various neurological and psychiatric disorders as well as healthy participants. Unbiased ground truth ROCF scores were obtained from crowdsourced human intelligence. This dataset was used to train and evaluate a multihead convolutional neural network. The model performs highly unbiased as it yielded predictions very close to the ground truth and the error was similarly distributed around zero. The neural network outperforms both online raters and clinicians. The scoring system can reliably identify and accurately score individual figure elements in previously unseen ROCF drawings, which facilitates explainability of the AI-scoring system. To ensure generalizability and clinical utility, the model performance was successfully replicated in a large independent prospective validation study that was pre-registered prior to data collection. Our AI-powered scoring system provides healthcare institutions worldwide with a digital tool to assess objectively, reliably, and time-efficiently the performance in the ROCF test from hand-drawn images.

Spinal cord interneurons play critical roles shaping motor output, but their precise identity and connectivity remain unclear. Focusing on the V1 interneuron cardinal class we defined four major V1 subsets in the mouse according to neurogenesis, genetic lineage-tracing, synaptic output to motoneurons, and synaptic inputs from muscle afferents. Sequential neurogenesis delineates different V1 subsets: two early born (Renshaw and Pou6f2) and two late born (Foxp2 and Sp8). Early born Renshaw cells and late born Foxp2-V1 interneurons are tightly coupled to motoneurons, while early born Pou6f2-V1 and late born Sp8-V1 interneurons are not, indicating that timing of neurogenesis does not correlate with motoneuron targeting. V1 clades also differ in cell numbers and diversity. Lineage labeling shows that the Foxp2-V1 clade contains over half of all V1 interneurons, provides the largest inhibitory input to motoneuron cell bodies, and includes subgroups that differ in birthdate, location, and proprioceptive input. Notably, one Foxp2-V1 subgroup, defined by postnatal Otp expression, is positioned near the LMC and receives substantial input from proprioceptors, consistent with an involvement in reciprocal inhibitory pathways. Combined tracing of ankle flexor sensory afferents and interneurons monosynaptically connected to ankle extensors confirmed placement of Foxp2-V1 interneurons in reciprocal inhibitory pathways. Our results validate previously proposed V1 clades as unique functional subtypes that differ in circuit placement, with Foxp2-V1 cells forming the most heterogeneous subgroup. We discuss how V1 organizational diversity enables understanding of their roles in motor control, with implications for their diverse ontogenetic and phylogenetic origins.