Each year, an estimated 795,000 individuals in the United States suffer a stroke. Following the stroke, 85 percent of patients have difficulty moving one arm or hand. These after-effects have a staggering impact on quality of life making stroke the leading cause of permanent disability in the United States. Treatment for impairments caused by stroke involves physical rehabilitation. However, these interventions require extensive therapist support. It has been estimated that 65% of the individuals cannot independently adhere to treatment outside of therapy. University of Pittsburgh researchers have solved this problem with a new invention and method that enables objective assessment of stroke severity. An accompanying wearable measurement system is used to measure the patient’s motions while completing the tasks, and these results are used in an algorithm. This algorithm has been developed based on data-driven analysis using machine learning processes to characterize the patient’s motions, including the severity of stroke-induced impairment and relationship to motions of healthy subjects and other stroke-affected patients.
Description
This invention includes a system and a method that enables objective assessment of stroke severity through measurement of the patient’s ability to perform motions that are representative of normal daily activities. Stroke impairments are currently assessed by a clinician using the Fugl-Meyer Upper Extremity (FMUE) Assessment. While the FMUE scoring system is a good indicator of stroke severity, it relies completely on the clinician to complete and is not conducive to providing feedback to the patient or re-evaluation in at-home exercise settings. In addition, it has been pointed out that that the FMUE method has difficulty distinguishing the difference between improvement in motion vs strength and, therefore, compensatory strategies used by the patient to complete tasks can be difficult to detect. Beyond assessment, the method presented enables personalized rehabilitation in a connected at-home system. Based on the patient’s results from the initial testing, the machine-learning based algorithm provides recommendations for rehabilitation exercises that are specifically tailored to the patient’s unique capabilities.
Applications
Individuals recovering from strokes
Advantages
Machine learning processes characterize the patient’s motions
Method uses data to prescribe improved patient-specific rehabilitation
Data is individualized
Invention Readiness
The results are used in an algorithm that has been developed based on data-driven analysis using machine learning processes to characterize the patient’s motions, including the severity of stroke-induced impairment and relationship to motions of healthy subjects and other stroke-affected patients. Finally, the method uses the data, along with data from many other stroke and healthy subjects, to prescribe patient-specific rehabilitation based on that individual’s capabilities.
IP Status
https://patents.google.com/patent/US20240164665A1
