CN-121987457-A - Electromagnetic control eyelid driving device and driving method based on artificial intelligence

CN121987457ACN 121987457 ACN121987457 ACN 121987457ACN-121987457-A

Abstract

The invention belongs to the technical fields of biomedical engineering, flexible electronic technology and artificial intelligence rehabilitation assistance, and provides an electromagnetic control eyelid driving device and a driving method based on artificial intelligence, comprising the steps of collecting multi-mode physiological signals of eyes of two sides of a patient in real time; the method comprises the steps of calculating double-side blink intention probability through a mixed AI prediction model, generating a driving instruction when a multi-frame consistency condition is met and double sides are synchronous, generating a controllable magnetic field based on the driving instruction, driving double-side eyelids to execute opening and closing actions in a non-contact flexible driving mode of magnetic fluid push-pull force, and independently fine-adjusting the magnetic fluid push-pull force of the double-side eyelids based on a double-layer control mechanism combining feedforward and feedback to compensate action difference of the double-side eyelids so as to enable closing amplitude of the double-side eyelids to be consistent. The invention can realize high sensitivity, low delay, strong and robust blink intention recognition and judgment and auxiliary driving function execution, and has good biocompatibility, wearing comfort and clinical application potential.

Inventors

LEI FENGYANG
TANG LIXUE
WANG ZIMENG
CAO XI
HOU SHENGPING

Assignees

北京市眼科研究所
首都医科大学
北京市糖尿病研究所(北京市糖尿病防治办公室)
首都医科大学附属北京同仁医院

Dates

Publication Date: 20260508
Application Date: 20260408

Claims (10)

1. An electromagnetic control eyelid driving method based on artificial intelligence is characterized by comprising the following steps: Step S100, acquiring multi-mode physiological signals of the eyes of the two sides of a patient in real time through a non-invasive sensing mode of skin contact, wherein the multi-mode physiological signals comprise electromyographic signals and skin surface strain signals; step S200, extracting local time characteristics of the multi-mode physiological signals through a mixed AI prediction model, analyzing time sequence dependency and bilateral synchronicity of the multi-mode physiological signals, calculating bilateral blink intention probability, and generating a driving instruction when multi-frame consistency conditions are met and bilateral synchronicity is achieved; Step S300, based on a driving instruction, a controllable magnetic field is generated, and a double-side eyelid is driven to execute opening and closing actions in a non-contact flexible driving mode of magnetic fluid push-pull force, and meanwhile, based on a double-layer control mechanism combining feedforward and feedback, the magnetic fluid push-pull force of the double-side eyelid is independently finely adjusted according to the multi-mode physiological signal in real time so as to compensate action differences of the double-side eyelid, so that closing amplitude of the double-side eyelid is consistent.
2. The method according to claim 1, wherein in the step S200, the hybrid AI prediction model is a hybrid architecture of a convolutional neural network and a long-short-term memory network, and the hybrid AI prediction model further includes a multi-modal fusion layer for weighting and integrating the features of the multi-modal physiological signals through an attention mechanism.
3. The method according to claim 1, wherein in step S200, when the multi-frame consistency condition is satisfied and the double-sided synchronization is performed, generating the driving instruction includes generating the driving instruction when both of the double-sided blink intention probabilities exceed a set threshold and synchronization is achieved within a preset time window, and it is determined that blink intention exists in at least 2 to 3 consecutive analysis time frames.
4. The method according to claim 1, wherein in the step S300, the double-layer control mechanism combining feedforward and feedback includes outputting initial values of bilateral driving currents according to the intensity of the driving command and a pre-calibrated mechanical model, and collecting the multi-mode physiological signals in real time during the execution of the opening and closing actions, and performing independent real-time fine tuning on the driving currents.
5. The method according to claim 4, wherein the step S300 further comprises executing a safety management strategy and an action rhythm management and calibration strategy synchronously during the execution of the opening and closing actions, wherein the safety management strategy comprises dual real-time monitoring and feedback control of the driving current and the temperature, and the action rhythm management and calibration strategy comprises performing action rhythm management and action minimum interval time control and synchronously completing online self-calibration.
6. An electromagnetic control eyelid driving device based on artificial intelligence, characterized by comprising a wearable main body, further comprising: The flexible sensor module is integrated on the wearable main body and used for collecting multi-mode physiological signals of the eyes of the two sides of a patient in real time in a non-invasive sensing mode of skin contact, wherein the multi-mode physiological signals comprise myoelectric signals and skin surface strain signals; The algorithm control module is in communication connection with the flexible sensor model and is arranged in the wearable main body, and is used for extracting local time characteristics of the multi-mode physiological signals through the mixed AI prediction model, analyzing time sequence dependence and double-side synchronism of the multi-mode physiological signals, calculating double-side blink intention probability, and generating a driving instruction when a multi-frame consistency condition is met and double-side synchronism is achieved; The electromagnetic driving module is in communication connection with the flexible sensor module and the algorithm control module and is used for generating a controllable magnetic field based on a driving instruction, driving the double-side eyelid to execute opening and closing actions in a non-contact flexible driving mode of magnetic fluid push-pull force, and simultaneously, independently fine-adjusting the magnetic fluid push-pull force of the double-side eyelid according to the multi-mode physiological signal in real time based on a double-layer control mechanism combining feedforward and feedback so as to compensate action difference of the double-side eyelid and enable closing amplitude of the double-side eyelid to be consistent.
7. The electromagnetically controlled eyelid driving device according to claim 6, wherein the flexible sensor module comprises a strain sensor array and an electrode array integrally formed on a flexible elastomer substrate using a liquid metal screen printing technique, wherein the strain sensor array has a serpentine or wavy liquid metal line structure.
8. The electromagnetically controlled eyelid driving device according to claim 6, wherein the electromagnetic driving module comprises an electromagnet group configured to control a driving current of the electromagnet group by a pulse width modulation manner to generate the gradient magnetic field, and further comprises a current monitoring unit and a temperature sensor.
9. The electromagnetic control eyelid driving device of claim 8, wherein the electromagnetic driving module further comprises an eyelid contact unit and a magnetic fluid unit, the eyelid contact unit comprising a closed microcavity, the magnetic fluid unit being encapsulated within the closed microcavity, the magnetic fluid unit comprising a magnetic fluid comprising ferroferric oxide nanoparticles.
10. The electromagnetic control eyelid driving device of claim 9, wherein an outer wall of the closed microcavity is provided with a limiting structure that limits the excessive flow of the magnetic fluid.

Description

Electromagnetic control eyelid driving device and driving method based on artificial intelligence Technical Field The application relates to the technical fields of biomedical engineering, flexible electronic technology and artificial intelligence rehabilitation assistance, in particular to an electromagnetic control eyelid driving device and an electromagnetic control eyelid driving method based on artificial intelligence. Background Currently, there are significant limitations to clinical treatment and assistance techniques for people with eyelid dysfunction or blink disabilities, mainly manifested in the following aspects: 1. The existing treatment approaches are invasive and lack dynamic adjustment capability-traditional correction of eyelid function is mainly dependent on surgery (such as levator palpebrae shortening, frontal valve suspension, etc.) or mechanical appliances. Although improving eyelid position to some extent, surgery is invasive and irreversible in nature, not only with long recovery cycles, but also with more difficulty in dynamic adjustments following surgery according to patient disease progression or individual needs. While non-invasive mechanical orthotics often provide static support or simple mechanical linkage, the driving mode is stiff, the circadian rhythm and fine dynamics of the natural opening and closing of the eyelid cannot be simulated, and the complex and changeable daily activity requirements are difficult to meet. 2. The control logic of the existing auxiliary devices is non-physiological and has poor individual adaptability, and even if some active auxiliary devices appear in recent years, the core control strategy of the existing auxiliary devices still has fundamental defects. On the one hand, the movement mode is often based on a preset fixed rhythm or a simple instruction, and the natural, smooth and cooperative movement rule of human eyes cannot be simulated, so that the movement mechanical stiffness is caused, and the wearing comfort and long-term compliance of a patient are seriously affected. On the other hand, these devices generally lack the ability to accurately identify the user's active blink intent, failing to achieve "idea-driven" visual control. For patients with impaired binocular function, achieving synchronized and natural binocular cooperative motion is particularly difficult due to lack of reference to healthy side signals and response to differential needs of the affected side. 3. The systematic lack of physiological signal monitoring and closed-loop control capability is that the prior art has two technical bottlenecks in the aspect of constructing a bionic closed-loop control system. First, at the level of signal acquisition, the muscle group driving eyelid movement (such as orbicularis oculi muscle) belongs to fine muscles, and the myoelectricity (EMG) and accompanying skin micro-deformation signals generated by the muscle group are very weak. The traditional rigid sensor is difficult to realize stable and high-sensitivity multi-mode acquisition of the signals on the premise of ensuring noninvasive and flexible wearing. Second, at the control level, although the camera eye tracker can track eye movement, it cannot directly acquire physiological status signals of eyelid dynamic units (muscle and skin), resulting in a lack of critical real-time feedback to the system. The lack of the end-to-end closed-loop capability makes the device unable to adaptively adjust according to the actual working state of the muscle and external interference, and eventually limits the possibility of achieving high-precision and low-delay auxiliary effects. In summary, the current technology has significant shortcomings in the imitative nature of functional reconstruction, the real-time nature of control, the adaptability of individualization and the ability of physiological closed-loop control. Disclosure of Invention In view of the above-mentioned shortcomings of the prior art, the present invention provides an electromagnetic control eyelid driving device and driving method based on artificial intelligence, which utilizes multimode physiological signal fusion and hybrid AI prediction model active prediction, can realize high sensitivity, low delay, strong robust blink intention recognition judgment and auxiliary driving function execution, and has good biocompatibility, wearing comfort and clinical application potential. In order to achieve the above and related objects, the present invention adopts the following technical scheme: The first aspect of the invention provides an electromagnetic control eyelid driving method based on artificial intelligence, comprising the following steps: Step S100, acquiring multi-mode physiological signals of the eyes of the two sides of a patient in real time through a non-invasive sensing mode of skin contact, wherein the multi-mode physiological signals comprise electromyographic signals and skin surface strain signals; Step S2