Sensor-Based Stroke Diagnosis Using ML for Rapid Assessment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for diagnosing stroke are inefficient, particularly in acute settings, as they rely on subjective evaluations by specialists and are not comprehensive, leading to potential misdiagnosis and delayed treatment due to the need for expensive and time-consuming MRI scans or cumbersome electroencephalography equipment.
Innovation Solution
A computer-implemented method using sensors to detect neurological symptoms through machine learning models, processing audio, video, and biological data to predict the occurrence and type of stroke in real-time, enabling quick and accurate diagnosis by non-specialist first responders or clinicians.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If sensor-based automated assessment is implemented, then diagnostic speed and accessibility are improved, but measurement precision may be compromised compared to specialist evaluation
Solution Approach 1:
The patent introduces an automated sensor-based assessment system that acts as an intermediary between the patient and the final diagnosis. Multiple sensors (video, audio, force, electrical) capture neurological symptoms, and machine learning models process this data to generate objective measurements. This intermediary system enables rapid screening and continuous monitoring while specialists focus on complex cases, resolving the contradiction between speed and precision.
Solution Approach 2:
The system transforms subjective neurological assessments into objective quantitative parameters through sensor measurements. By converting symptoms like speech patterns, eye movements, and motor responses into measurable data points, the system achieves both rapid automated evaluation and precise objective measurement, eliminating the traditional trade-off between speed and accuracy.
2Measurement precision
If comprehensive neurological assessment is performed, then diagnostic accuracy is improved, but device complexity and cost increase
Solution Approach 1:
The assessment system is divided into modular sensor components (video sensors for facial symmetry, audio sensors for speech analysis, force sensors for motor function, electrical sensors for autonomic function). Each sensor type targets specific neurological domains, allowing comprehensive assessment through coordinated simple modules rather than one complex device. This segmentation maintains diagnostic accuracy while reducing individual component complexity.
Solution Approach 2:
The patent employs a multi-functional sensor platform that can perform diverse neurological assessments using integrated sensor types. The same hardware infrastructure supports multiple assessment protocols and can evaluate different neurological functions, reducing overall system complexity while maintaining comprehensive diagnostic capability across multiple symptom domains.
3Adaptability or versatility
If automated sensor-based diagnosis is used, then accessibility and scalability are improved, but reliability may be reduced compared to specialist evaluation
Solution Approach 1:
The system incorporates continuous feedback loops where sensor data is constantly monitored, machine learning models are retrained with new data, and assessment protocols are refined based on performance metrics. This feedback mechanism ensures the automated system maintains and improves reliability over time while becoming more widely accessible, as the system learns from diverse populations and clinical scenarios.
Solution Approach 2:
The automated assessment system performs preliminary diagnostic evaluation and symptom quantification before specialist review. By pre-processing and standardizing assessments across multiple sites, the system ensures consistent reliable measurements are captured initially, reducing variability and improving overall diagnostic reliability while enabling scalable deployment across resource-limited settings.
Data Source
AI summary
A method, a system, and a computer program product for detecting and/or determining occurrence of a neurological event in a subject. Data corresponding to one or more symptoms, detected by one or more sensors, associated with a subject is received. The sensors include sensors positioned directly on the subject and/or sensors positioned away from the subject. One or more symptom values are assigned to one or more detected symptoms. A severity score for each of the symptoms is determined. The severity scores are determined using one or more machine learning models receiving the assigned symptom values as input. A prediction that the subject is experiencing at least one neurological event and at least a type of the neurological event is generated using a combination of the determined severity scores corresponding to the symptoms. A generation of one or more alerts is triggered based on the prediction. One or more user interfaces are generated for displaying the alerts.