A physiological sensing system and a physiological signal sensing method

Abstract

This invention relates to a physiological sensing system and a physiological signal sensing method. The physiological sensing system includes a sensing device, a processing device, and a decision-making device. The physiological signal sensing method includes: using the sensing device to sense physiological signals such as sleep, heartbeat, and respiration of an organism; transmitting the physiological signals to the processing device, which includes an AI Model Selector System to identify the physiological signal patterns of the organism; and using the decision-making device, which includes an AI Hybrid Deep Learning Model, to determine whether the organism's vital signs are abnormal and to propose improvement suggestions. The AI ​​Hybrid Deep Learning Model is an AI physiological detection system that combines a multi-modal learning mechanism. Based on the physiological signal patterns identified by the system, it determines whether the organism's state is abnormal, enabling real-time notification and remote monitoring mechanisms to achieve improved preventive medical effects.

Claims

1. A physiological sensing system, comprising: A sensing device for detecting at least one physiological signal of a living being, including one or a combination of sleep, heartbeat or breathing signals, and converting it into a detection signal that can be processed by artificial intelligence; a processing device electrically connected to the sensing device for receiving the sensing data and processing it to generate physiological characteristic data. And a decision-making device electrically connected to the processing device, used to execute artificial intelligence algorithms based on the physiological characteristic data to identify the physiological patterns of the organism's sleep, heartbeat or breathing, and to determine whether the organism has an abnormal physiological state.

2. The physiological sensing system as described in claim 1, wherein the processing device includes an AI Model Selector System for selecting at least one deep learning model for physiological signal analysis, the deep learning model including one or a combination of the following: Convolutional Neural Network (CNN), Long Short-Term Memory Network (LSTM), Bidirectional Long Short-Term Memory Network (BiLSTM), Transformer, and Deep Neural Network (DNN).

3. The physiological sensing system as described in claim 2, wherein the processing device uses a one-dimensional convolutional neural network (1DCNN) to extract features from physiological signals and includes classifying physiological patterns using a temporal model, the temporal model including one or more of the following: a long short-term memory network (LSTM), a bidirectional long short-term memory network (BiLSTM), and a transformer.

4. The sensing system as described in claim 2, wherein the AI ​​Model Selector System of the processing device selects an applicable deep learning model based on the characteristics of the physiological signal, the characteristics including signal length, frequency variation, and noise intensity, and the AI ​​Model Selector System includes training the classification function using meta-learning or continuously optimizing the system selection strategy during data collection using reinforcement learning to ensure the accuracy and computational efficiency of model selection.

5. The sensing system as described in claim 1, wherein the processing device includes a verification feedback system (VFC) for comparing the model prediction results with the actual observation data and adjusting the computational parameters of the artificial intelligence model based on the difference.

6. The physiological sensing system as described in claim 1, wherein the decision-making device includes employing a multi-modal learning model to integrate feature data of different physiological signals and perform health risk assessment.

7. The physiological sensing system as claimed in claim 1, wherein the processing device includes a time-frequency analysis module that enhances attention to timing and spectral features, the time-frequency analysis module including any one of: short-time Fourier transform (STFT), wavelet transform (WT), or Hilbert-Huang transform (HHT).

8. A physiological monitoring system, comprising: A physiological sensing system as described in claim 1; A concealed video intercom device is electrically connected to the physiological sensing system to provide two-way communication functionality; And a remote physiological monitoring device for measuring physiological data of a living organism and transmitting the physiological data to the physiological sensing system, the remote physiological monitoring device including a blood oxygen sensor and a blood pressure monitoring device.

9. A physiological signal sensing method, comprising the following steps: using a sensing device to sense at least one physiological signal of a living organism and generate sensing data; transmitting the sensing data to a processing device and performing signal processing to generate physiological characteristic data; using an artificial intelligence model to analyze the physiological characteristic data to identify the physiological patterns of the organism's sleep, heartbeat, or breathing; and determining whether the organism has an abnormal physiological state based on the analysis results.

10. The physiological signal sensing method as described in claim 9, wherein the method further includes a computational efficiency optimization process, comprising the following steps: Step 1, using a one-dimensional convolutional neural network (1DCNN) to perform preliminary feature extraction on the physiological signal to reduce the input dimension of the subsequent temporal model; Step 2, selecting a convolutional neural network (CNN) or a bidirectional long short-term memory network (BiLSTM) for pattern analysis based on the temporal characteristics of the physiological signal; Step 3, when an abnormal event is detected, activating a Transformer module for deep analysis.

11. The physiological signal sensing method as described in claim 10, wherein the computational efficiency optimization process further includes model compression techniques, including one or a combination of model quantization, knowledge distillation, and depthwise separable convolution.

Images