To explore the use of large language models (LLMs) in improving the diagnosis of language disorders associated with autism by enhancing diagnostic sensitivity and profiling linguistic features specific to these disorders.
Key Findings:
Over a 10% increase in sensitivity and positive predictive value compared to baseline models, indicating significant improvement.
Identification of ten key features of autism-associated language disorders, including echolalia and pronoun reversal, which are critical for accurate diagnosis.
LLMs can assist in recognizing subtle language disorder patterns that traditional methods may overlook, enhancing the overall diagnostic process.
Interpretation:
The findings suggest that LLMs can serve as valuable supplementary tools in the diagnostic process for autism-related language disorders, enhancing both accuracy and efficiency while supporting clinical judgment.
Limitations:
LLMs are not diagnostic tools but rather aids to clinical judgment, highlighting the need for human expertise in the diagnostic process.
The application of LLMs in ASD diagnosis is still an emerging field with limited exploration, necessitating further research to validate their effectiveness.
Conclusion:
LLMs have the potential to improve the diagnostic process for autism-related language disorders, enabling more personalized therapeutic strategies.
