Understanding how psychiatric disorders and chronic pain alter brain connectivity is essential for advancing precision medicine in mental health. This study introduces a normative modeling approach using resting-state functional MRI data to detect group level deviations in functional connectivity associated with schizophrenia, bipolar disorder, ADHD, and chronic pain. By training an encoder-decoder model on healthy individuals, we establish a baseline of typical brain network interactions, allowing us to identify disorder-specific and shared connectivity disruptions. A rigorous model selection strategy is employed to ensure robustness against preprocessing variability, accounting for different denoising techniques, brain parcellation methods, and connectivity estimation approaches. The results reveal distinct yet overlapping neural network dysfunctions: schizophrenia is linked to limbic-somatomotor and attention network disruptions, bipolar disorder exhibits limbic and ventral attention network abnormalities, and ADHD shows default mode and attention network alterations. Notably, chronic pain shares functional connectivity disruptions with bipolar disorder, suggesting common neurobiological mechanisms underlying emotional dysregulation. These findings are consistent with previous research and further support normative modeling as a powerful approach for neuroimaging-based biomarker discovery. By identifying disorder-specific and transdiagnostic connectivity alterations, this method opens new possibilities for personalized diagnosis, refined disorder classification, and targeted treatment strategies in psychiatric and pain-related conditions.