fig1

Figure 1. Impaired sociability and social novelty recognition in Shank3B KO mice revealed by machine learning-enhanced behavioral analysis. (A) Schematic diagram of the three-chamber test procedure. (B) Heatmaps of movement trajectories of KO and WT mice in three-chamber sociability test. (C) Time spent in each chamber of WT mice during Stage 1. (D) Time spent in each chamber of KO mice during Stage 1. (E) Time spent in each chamber of WT mice during Stage 2. (F) Time spent in each chamber of KO mice during Stage 2. (G) Time allocation percentages across different behaviors in Stage 1. (H) Heatmap showing the transition probability difference during Stage 1. (I) ROC curve demonstrating the classification performance of a behavioral feature-based classifier in discriminating between WT and KO mice in Stage 1. (accuracy = 0.664 ± 0.002, AUC = 0.964 ± 0.007). (J) Time allocation percentages across different behaviors in Stage 2. (K) Heatmap showing the transition probability difference during Stage 2. (L) ROC curve demonstrating the classification performance of a behavioral feature-based classifier in discriminating between WT and KO mice in Stage 2. (accuracy = 0.664 ± 0.002, AUC = 0.964 ± 0.007). n = 12 mice for WT and n = 8 mice for KO. Data presented as means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Comparisons between two groups were performed using two-tailed Student's t-test or Mann-Whitney U test as appropriate. For analyses involving multiple comparisons (behavioral subtype proportions and transition probabilities), p values are nominal; significance was determined using the Benjamini-Hochberg FDR procedure with q < 0.05.