Agent-guided AI-powered interpretation and reporting of nerve conduction studies and EMG (INSPIRE)
Alon Gorenshtein1, Moran Sorka2, Mohamed Khateb3
1Department of Neurology, Rambam Health Care Campus, Haifa, Israel; Azrieli Faculty of Medicine, Bar-Ilan University, Safed, Israel; AI in Neurology Laboratory, Ruth and Bruce Rapaport Faculty of Medicine, Technion Institute of Technology, Haifa 3525408, Israel.
Summary
A new AI tool, INSPIRE, significantly improves the interpretation of electrodiagnostic tests (EDX) for neuromuscular conditions. It enhances accuracy and standardization in EMG reporting, aiding clinicians in diagnosis.
Area of Science:
- Neurology
- Artificial Intelligence
- Medical Diagnostics
Background:
- Electroneuromyography (EMG) and nerve conduction studies (NCS) are crucial for diagnosing neuromuscular disorders.
- Interpretation of these tests can be complex and time-consuming, requiring significant expertise.
- Standardization of EDX interpretation is essential for consistent patient care.
Purpose of the Study:
- To develop and evaluate an AI-powered tool for enhancing and standardizing the interpretation of electrodiagnostic (EDX) tests.
- To leverage generative AI technology for improved accuracy and efficiency in neuromuscular electrophysiology.
Main Methods:
- Three AI model frameworks were developed: Base-LLM, INSPIRE (a multi-agent AI), and INSPIRE-Lite.
- INSPIRE integrates multiple agents to access reference tables and clinical textbooks for comprehensive analysis.
- Performance was assessed using the AI-Generated EMG Report Score (AIGERS).
Main Results:
- INSPIRE achieved 92.2% accuracy in distinguishing normal from abnormal EDX tests, significantly outperforming Base-LLM (62.6%).
- INSPIRE demonstrated superior AIGERS scores in overall performance, finding, clinical diagnosis, and semantic concordance.
- INSPIRE-Lite showed lower scores than INSPIRE in finding and clinical diagnosis domains.
Conclusions:
- The developed AI model effectively integrates patient history, symptoms, and EDX findings for EMG interpretation.
- The model shows superior performance in accuracy and standardization, addressing challenges like data overload and hallucinations.
- This AI tool enhances diagnostic accuracy and efficiency in generating EDX reports.
