Swallowing function evaluation method based on ai monitoring and related device

Abstract

The embodiment of the application relates to the field of data processing, and provides a swallowing function evaluation method based on AI monitoring and related devices, which comprises the following steps: acquiring throat light contact data, non-contact image data and physiological parameter data; performing feature fusion and derivation on the throat light contact data, the non-contact image data and the physiological parameter data to obtain stroke pathological correlation features; performing swallowing function grading evaluation according to the throat light contact data, the non-contact image data, the physiological parameter data and the stroke pathological correlation features to obtain swallowing function grading results; performing aspiration risk probability evaluation according to the stroke pathological correlation features to obtain an aspiration risk probability; performing rehabilitation progress evaluation according to the stroke pathological correlation features to obtain rehabilitation progress evaluation results; and performing comprehensive evaluation and decision output according to the swallowing function grading results, the aspiration risk probability and the rehabilitation progress evaluation results to obtain swallowing function comprehensive evaluation and decision data, so that more accurate swallowing function evaluation can be realized.

Claims

1. A method for assessing swallowing function based on AI monitoring, characterized in that, The AI-based swallowing function assessment method includes: Acquire light contact data, non-contact imaging data, and physiological parameter data of the larynx; The pathological features associated with stroke are obtained by fusing and deriving features from the light contact data of the larynx, the non-contact imaging data, and the physiological parameter data. Swallowing function is graded and assessed based on the light contact data of the larynx, the non-contact imaging data, the physiological parameter data, and the pathological correlation characteristics of stroke, and the swallowing function grading results are obtained. The risk probability of aspiration is assessed based on the aforementioned pathological characteristics associated with stroke, and the risk probability of aspiration is obtained. The rehabilitation progress is assessed based on the aforementioned pathological correlation characteristics of stroke, and the rehabilitation progress assessment results are obtained. Based on the swallowing function grading results, the probability of aspiration risk, and the rehabilitation progress assessment results, a comprehensive assessment decision is output to obtain swallowing function comprehensive assessment decision data.

2. The swallowing function assessment method based on AI monitoring according to claim 1, characterized in that, The feature fusion and derivation of the light contact data of the larynx, the non-contact imaging data, and the physiological parameter data yields stroke pathological association features, including: Feature extraction was performed on the light contact data of the larynx to obtain basic electromyographic features and basic motor features; Feature extraction is performed on the non-contact image data to obtain the basic image features; Feature extraction is performed on the physiological parameter data to obtain basic physiological features; The swallowing coordination index is obtained by performing multi-dimensional fusion calculations based on the electromyographic characteristics, the motor characteristics, and the physiological characteristics. The swallowing coordination index for multiple consecutive swallowing processes is obtained from the swallowing coordination index to obtain the first continuous swallowing coordination index. A time-series trend analysis was performed on the first continuous swallowing coordination index to obtain dynamic risk trend characteristics; Based on the electromyographic characteristics, the motor characteristics, the imaging characteristics, the physiological characteristics, the swallowing coordination index, and the dynamic risk trend characteristics, the pathological association characteristics of stroke are obtained.

3. The swallowing function assessment method based on AI monitoring according to claim 2, characterized in that, The swallowing function grading assessment is performed based on the light laryngeal contact data, the non-contact imaging data, the physiological parameter data, and the pathological correlation characteristics of stroke, to obtain the swallowing function grading result, including: The stroke pathology-related features were subjected to feature screening to obtain a subset of core swallowing function features; Based on the laryngeal light contact data, the non-contact imaging data, and the physiological parameter data, the dominant swallowing disorder pattern is identified; the dominant disorder pattern includes at least one of oral phase initiation delay, pharyngeal phase weakening triggering, laryngeal incomplete closure, and swallowing-breathing coordination disorder. Based on the dominant obstacle pattern, the weight of each core feature of swallowing function in the subset of core features of swallowing function is dynamically adjusted to obtain a dynamically adjusted weight subset; Based on the dynamically adjusted weight subset, feature weighting is performed on each core swallowing feature in the core swallowing feature subset to obtain a weighted subset of swallowing function features. A linear summation is performed on each weighted value of the swallowing function feature in the weighted value subset to obtain a comprehensive swallowing function score. The swallowing function classification result is obtained by performing a grading threshold mapping process based on the dominant obstacle pattern and the comprehensive swallowing function score.

4. The swallowing function assessment method based on AI monitoring according to claim 3, characterized in that, The assessment of aspiration risk probability based on the pathological correlation characteristics of stroke, to obtain the aspiration risk probability, includes: A subset of aspiration risk-related features is obtained from the aforementioned stroke pathological association features; Each aspiration risk-related feature in the subset of aspiration risk-related features is standardized to obtain a standardized subset of aspiration risk features. For each standardized aspiration risk feature in the standardized aspiration risk feature subset, feature fusion processing is performed to obtain an aspiration risk feature vector; Based on a preset clinical coefficient matrix, the risk linearity calculation of the aspiration risk feature vector is performed to obtain the aspiration linear risk value. Based on the linear risk value of aspiration, the probability conversion of aspiration risk is performed to obtain the initial value of aspiration risk probability. Based on the dynamic risk trend characteristics in the pathological association features of stroke, determine whether there is a temporal abnormality in the current swallowing and obtain the swallowing temporal abnormality risk coefficient. Based on the imaging and motor characteristics in the pathological association features of stroke, the failure probability of critical protective events in the pharyngeal phase is calculated to obtain the failure probability of critical events. Based on the target image basic features corresponding to the pathological association features of stroke within a preset time window after swallowing, the amount of pharyngeal residue is assessed to obtain the residue risk index. The aspiration risk probability is obtained by performing nonlinear coupling calculation based on the initial value of the aspiration risk probability, the swallowing timing abnormality risk coefficient, the critical event failure probability, and the residue risk index.

5. The swallowing function assessment method based on AI monitoring according to claim 4, characterized in that, The rehabilitation progress assessment based on the pathological correlation characteristics of stroke, to obtain the rehabilitation progress assessment results, includes: The current swallowing coordination index corresponding to the current assessment process during the same stroke rehabilitation period and the historical swallowing coordination index corresponding to multiple historical assessment processes are obtained from the stroke pathology correlation features. The current swallowing coordination index and the historical swallowing coordination index are arranged in order of evaluation time to form a time-series coordination index dataset; The slope of the rehabilitation trend is calculated based on the time-series synergistic index dataset to obtain the rate of change of the synergistic index; Based on the assessment interval between the current assessment process and the nearest historical assessment process, the rate of change of the synergy index is converted into a time-based value to obtain the current recovery rate. Based on a preset rehabilitation rate threshold, a status determination process is performed according to the current rehabilitation rate to obtain a rehabilitation progress assessment result.

6. A swallowing function assessment device based on AI monitoring, characterized in that, The device includes: The acquisition module is used to acquire light contact data, non-contact imaging data, and physiological parameter data of the larynx; The first processing module is used to perform feature fusion and derivation on the light contact data of the larynx, the non-contact imaging data and the physiological parameter data to obtain stroke pathological correlation features. The second processing module is used to perform a swallowing function grading assessment based on the laryngeal light contact data, the non-contact imaging data, the physiological parameter data, and the pathological correlation characteristics of stroke, and to obtain the swallowing function grading result. The third processing module is used to assess the probability of aspiration risk based on the pathological correlation characteristics of stroke, and obtain the probability of aspiration risk. The fourth processing module is used to assess the rehabilitation progress based on the pathological correlation characteristics of the stroke and obtain the rehabilitation progress assessment results. The fifth processing module is used to perform a comprehensive assessment and decision output based on the swallowing function grading results, the probability of aspiration risk, and the rehabilitation progress assessment results, to obtain comprehensive swallowing function assessment and decision data.

7. The swallowing function assessment device based on AI monitoring according to claim 6, characterized in that, The first processing module is used to perform feature fusion and derivation on the light contact data of the larynx, the non-contact imaging data, and the physiological parameter data to obtain stroke pathological correlation features, specifically for: Feature extraction was performed on the light contact data of the larynx to obtain basic electromyographic features and basic motor features; Feature extraction is performed on the non-contact image data to obtain the basic image features; Feature extraction is performed on the physiological parameter data to obtain basic physiological features; The swallowing coordination index is obtained by performing multi-dimensional fusion calculations based on the electromyographic characteristics, the motor characteristics, and the physiological characteristics. The swallowing coordination index for multiple consecutive swallowing processes is obtained from the swallowing coordination index to obtain the first continuous swallowing coordination index. A time-series trend analysis was performed on the first continuous swallowing coordination index to obtain dynamic risk trend characteristics; Based on the electromyographic characteristics, the motor characteristics, the imaging characteristics, the physiological characteristics, the swallowing coordination index, and the dynamic risk trend characteristics, the pathological association characteristics of stroke are obtained.

8. The swallowing function assessment device based on AI monitoring according to claim 7, characterized in that, The second processing module is used to perform a swallowing function grading assessment based on the light laryngeal contact data, the non-contact imaging data, the physiological parameter data, and the pathological correlation characteristics of stroke, and to obtain a swallowing function grading result. Specifically, it is used for: The stroke pathology-related features were subjected to feature screening to obtain a subset of core swallowing function features; Based on the laryngeal light contact data, the non-contact imaging data, and the physiological parameter data, identify the dominant swallowing disorder pattern. The dominant disorder pattern includes at least one of oral phase initiation delay, pharyngeal phase weakening triggering, laryngeal incomplete closure, and swallowing-breathing coordination disorder. Based on the dominant obstacle pattern, the weight of each core feature of swallowing function in the subset of core features of swallowing function is dynamically adjusted to obtain a dynamically adjusted weight subset; Based on the dynamically adjusted weight subset, feature weighting is performed on each core swallowing feature in the core swallowing feature subset to obtain a weighted subset of swallowing function features. A linear summation is performed on each weighted value of the swallowing function feature in the weighted value subset to obtain a comprehensive swallowing function score. The swallowing function classification result is obtained by performing a grading threshold mapping process based on the dominant obstacle pattern and the comprehensive swallowing function score.

9. A terminal, characterized in that, The device includes a processor, an input device, an output device, and a memory, which are interconnected. The memory is used to store a computer program, which includes program instructions. The processor is configured to invoke the program instructions to execute the AI-based swallowing function assessment method as described in any one of claims 1-5.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, the computer program including program instructions that, when executed by a processor, cause the processor to perform the swallowing function assessment method based on AI monitoring as described in any one of claims 1-5.

Images