The field of neurocognitive disorder detection and diagnosis is moving towards the development of innovative and non-invasive methods for early prediction and identification of cognitive decline. Recent studies have focused on leveraging machine learning, natural language processing, and neuroimaging techniques to improve diagnostic accuracy and understand the underlying neural mechanisms of neurocognitive disorders. One of the key directions in this field is the use of diffusion-based frameworks and language models to synthesize clinically plausible future representations of neurological data, allowing for real-time risk assessment and high predictive performance. Additionally, researchers are exploring the use of gaze-based neural preliminary diagnosis, which has the potential to provide more objective insights into brain function and cognitive processing. Another area of research is the development of naturalistic language-related fMRI tasks for detecting neurocognitive decline and disorder. These tasks have shown promise in identifying cognitive decline and early neurocognitive disorder, particularly in older adults. Noteworthy papers in this area include those that propose novel diffusion-based frameworks for early MCI conversion prediction, and those that investigate neural responses to affective sentences as signatures of depression. For example, one paper achieves a significant improvement in early conversion accuracy by using a linguistic compass to steer the generation of clinically plausible future sMRI representations. Another paper uses deep learning models to distinguish healthy from depressed participants with high accuracy, highlighting the potential of neural signatures for diagnostic tools.