A framework for closing the loop between human experts and computational algorithms for the assessment of movement...

Area of Science:

• Neurology
• Biomechanical Engineering
• Computational Medicine

Background:

• Clinical assessment of abnormal neuromechanics often relies on physical limb manipulation, which suffers from low inter- and intra-rater reliability.
• Existing diagnostic methods for movement disorders lack standardization and objective quantitative measures.

Purpose of the Study:

• To formalize a computational framework that integrates clinician expertise with algorithms to enhance diagnostic capabilities during physical manipulation.
• To improve the reliability and accuracy of diagnosing movement disorders through a structured, data-driven approach.

Main Methods:

• Developed a framework that links clinician expertise with computational algorithms for diagnosing movement disorders.
• Utilized an "information map" database to encode movement dynamics and guide the selection of informative probing motions.
• Employed a recursive algorithm that uses measured limb dynamics and clinician input to estimate the probability of specific movement disorders.

Main Results:

• Formalized a novel framework for the standardized assessment of movement disorders.
• Introduced the "information map" as a key component for selecting movements that yield distinguishing diagnostic information.
• The framework enables recursive refinement of diagnostic probabilities based on collected data and clinician input.

Conclusions:

• The proposed framework provides a foundation for a standardized and more reliable clinical assessment of movement disorders.
• Integrating computational algorithms with clinician expertise has the potential to significantly enhance diagnostic accuracy.
• Future work will focus on testing the framework's efficacy in real-world clinical settings.