Search

CN-122024433-A - Intelligent drainage metering early warning device and method based on multi-sensor fusion and LSTM prediction model

CN122024433ACN 122024433 ACN122024433 ACN 122024433ACN-122024433-A

Abstract

The invention relates to the technical field of intersection of medical monitoring equipment and artificial intelligence, in particular to an intelligent drainage metering and early warning device and method based on multi-sensor fusion and an LSTM prediction model. The device comprises a drainage liquid collecting container, a four-dimensional sensor array, a microprocessor, an intelligent alarm and an interface module, wherein the four-dimensional sensor array comprises a weighing sensor, an optical sensor, a capacitance sensor and a miniature pressure sensor, and the microprocessor, the intelligent alarm and the interface module enable an electronic drainage meter to realize high-precision measurement, advanced intelligent early warning, self-adaptive clinical adaptation, safe and efficient data transmission and rapid abnormality correction by fusing various intelligent algorithms, so that the safety and the efficiency of postoperative drainage monitoring are remarkably improved.

Inventors

  • YUAN YING

Assignees

  • 首都医科大学附属北京天坛医院

Dates

Publication Date
20260512
Application Date
20260224

Claims (10)

  1. 1. Intelligent drainage metering early warning device based on multisensor fuses and LSTM predictive model, its characterized in that, the device includes: the inner wall of the drainage liquid collecting container is provided with a super-hydrophobic wall-hanging-preventing coating; the four-dimensional sensor array comprises a weighing sensor, an optical sensor, a capacitance sensor and a miniature pressure sensor, wherein the precision of the weighing sensor is +/-0.1 g, and the miniature pressure sensor is used for monitoring the pressure of fluid in a drainage tube; The microprocessor integrates an ARM Cortex-M7 kernel and an edge calculation acceleration unit, and is configured to run a multi-sensor federal fusion algorithm, an attention-enhanced LSTM prediction algorithm, a VAE-CNN joint anomaly detection algorithm, a wavelet-federal encryption transmission algorithm and a reinforcement learning self-adaptive threshold optimization algorithm; the intelligent alarm comprises a directional loudspeaker and a three-color indicator lamp and is configured to output three-level response signals according to abnormal grades; the interface module supports Bluetooth 5.0, wi-Fi 6 and USB 3.0 data transmission and is configured to interact encrypted data with the hospital information system; The drainage liquid collecting container is connected with the four-dimensional sensor array, the four-dimensional sensor array is connected with the microprocessor, the microprocessor is respectively connected with the intelligent alarm and the interface module, and the intelligent alarm is connected with the interface module.
  2. 2. The apparatus of claim 1, wherein the four-dimensional sensor array is configured to collect data simultaneously every 0.2s, the optical sensor is configured to detect a refractive index of the drainage fluid to distinguish between fluid types, and the capacitive sensor is configured to compensate for ambient temperature disturbances to the metering.
  3. 3. The apparatus of claim 1, wherein the multi-sensor federation fusion algorithm in the microprocessor is configured to receive real-time data of the four-dimensional sensor array, dynamically adjust each sensor weight based on a federal learning model and a recursive least squares algorithm, output fused drainage volume and flow rate, and measure errors less than or equal to + -0.3%.
  4. 4. The apparatus of claim 1, wherein the three-level response signal of the intelligent alarm comprises a blue pre-alarm signal, a yellow pre-alarm signal and a red pre-alarm signal, which correspond to the suspected abnormality, the abnormality to be confirmed and the abnormality to be diagnosed, respectively.
  5. 5. The device of claim 1, further comprising an intelligent negative pressure control module, wherein the intelligent negative pressure control module is linked with the microprocessor and the miniature pressure sensor and is configured to dynamically adjust the negative pressure value according to the pressure in the drainage tube, and the adjustment accuracy is +/-2 mmHg.
  6. 6. The intelligent drainage metering early warning method based on the multi-sensor fusion and the LSTM prediction model is applied to the intelligent drainage metering early warning device based on the multi-sensor fusion and the LSTM prediction model, and is characterized by comprising the following steps: collecting drainage liquid by using a drainage liquid collecting container, and synchronously collecting weight, refractive index, dielectric constant and drainage tube pressure data of the drainage liquid in every preset time by using a four-dimensional sensor array; based on the weight, refractive index, dielectric constant and drainage tube pressure data of the drainage liquid, through The microprocessor runs a multi-sensor federation fusion algorithm, dynamically adjusts the weight of each sensor based on the drainage liquid type, and calculates the real-time drainage volume and flow rate; based on the results of the real-time drainage volume and flow velocity, predicting the drainage flow velocity trend in the future period of time through an attention-enhanced LSTM model, identifying an abnormal mode by combining a VAE-CNN combined algorithm, and triggering the corresponding grade response of the intelligent alarm to obtain related drainage data; And carrying out wavelet transformation compression and homomorphic encryption on the drainage data, and transmitting the drainage data to a hospital information system through an interface module.
  7. 7. The method of claim 6, wherein running a multisensor federal fusion algorithm via a microprocessor based on weight, refractive index, dielectric constant, and intra-drain pressure data of the drainage fluid, dynamically adjusting sensor weights based on drainage fluid type, and calculating real-time drainage volume and flow rate comprises: Initializing the weight of each sensor based on a preset cloud pre-trained global weight model, carrying out local fine adjustment by combining the refractive index and pressure data of the drainage liquid acquired in real time, and calculating the real-time drainage volume and flow velocity; The weight satisfies w1+w2+w3 +w4=1; wherein w1 W4 is the weight of the weighing, optical, capacitance and pressure sensor respectively.
  8. 8. The method of claim 6, wherein predicting the drainage flow rate trend over a future period of time by an attention-enhanced LSTM model comprises: The method comprises the steps of distributing first attention weights to drainage flow velocity time sequences in a first postoperative period through an attention-enhanced LSTM model, and distributing second attention weights to drainage flow velocity time sequences in a second postoperative period; And correcting the prediction result by combining the pressure data of the miniature pressure sensor based on the first attention weight and the second attention weight.
  9. 9. The method of claim 6, wherein the VAE-CNN joint algorithm identifies abnormal patterns, triggers corresponding level responses of intelligent alarms, comprising: Calculating a reconstruction error of the prediction result through the variation self-encoder, and marking the reconstruction error as a suspected abnormal value when the reconstruction error is larger than a first threshold value; extracting high-frequency features of the suspected abnormal values through a convolutional neural network, and matching a complication feature library to realize abnormal type classification; And triggering corresponding grade response of the intelligent alarm based on the abnormal type classification.
  10. 10. The method of claim 6, wherein the wavelet transform compression and homomorphic encryption of the drainage data is transmitted to the hospital information system via the interface module, comprising: compressing the drainage data by wavelet transformation to obtain first compressed data; And encrypting the first compressed data by adopting a homomorphic encryption algorithm, and transmitting the encrypted data to a hospital information system through an interface module.

Description

Intelligent drainage metering early warning device and method based on multi-sensor fusion and LSTM prediction model Technical Field The invention relates to the technical field of intersection of medical monitoring equipment and artificial intelligence, in particular to an intelligent drainage metering and early warning device and method based on multi-sensor fusion and an LSTM prediction model. Background The existing electronic drainage metering equipment has the technical bottlenecks that a single sensor fusion algorithm is limited in adaptability, diversified liquid characteristics such as blood, bile and cerebrospinal fluid are difficult to cover, a traditional prediction model is insufficient in attention degree of clinical key time periods and limited in early warning advance, an abnormal detection can only identify a known mode, the rare complication omission rate is high, the privacy leakage risk exists in data transmission, and the multi-center clinical data collaborative optimization is difficult to adapt. Therefore, the invention constructs a full-link intelligent system of 'perception-analysis-decision-transmission' by fusing front-edge algorithms such as federal learning, deep reinforcement learning, variation self-encoder and the like with four-dimensional sensing hardware, and breaks through the technical limitation. Disclosure of Invention In view of the above, the invention aims to provide an intelligent drainage metering early warning device and method based on multi-sensor fusion and LSTM prediction model, so as to solve the technical problems in the prior art. According to a first aspect of an embodiment of the present invention, there is provided an intelligent drainage metering and early warning device based on a multisensor fusion and LSTM prediction model, the device comprising: the inner wall of the drainage liquid collecting container is provided with a super-hydrophobic wall-hanging-preventing coating; the four-dimensional sensor array comprises a weighing sensor, an optical sensor, a capacitance sensor and a miniature pressure sensor, wherein the precision of the weighing sensor is +/-0.1 g, and the miniature pressure sensor is used for monitoring the pressure of fluid in a drainage tube; The microprocessor integrates an ARM Cortex-M7 kernel and an edge calculation acceleration unit, and is configured to run a multi-sensor federal fusion algorithm, an attention-enhanced LSTM prediction algorithm, a VAE-CNN joint anomaly detection algorithm, a wavelet-federal encryption transmission algorithm and a reinforcement learning self-adaptive threshold optimization algorithm; the intelligent alarm comprises a directional loudspeaker and a three-color indicator lamp and is configured to output three-level response signals according to abnormal grades; The interface module supports Bluetooth 5.0, wi-Fi6 and USB3.0 data transmission and is configured to interact encrypted data with the hospital information system; The drainage liquid collecting container is connected with the four-dimensional sensor array, the four-dimensional sensor array is connected with the microprocessor, the microprocessor is respectively connected with the intelligent alarm and the interface module, and the intelligent alarm is connected with the interface module. Further, the four-dimensional sensor array is configured to collect data synchronously every 0.2s, the optical sensor is used for detecting the refractive index of the drainage liquid to distinguish liquid types, and the capacitive sensor is used for compensating interference of ambient temperature on metering. Further, the multi-sensor federation fusion algorithm in the microprocessor is configured to receive real-time data of the four-dimensional sensor array, dynamically adjust the weight of each sensor based on a federation learning model and a recursive least squares algorithm, output the fused drainage liquid volume and flow velocity, and measure the error less than or equal to +/-0.3%. Further, the three-stage response signals of the intelligent alarm comprise a blue early warning signal, a yellow early warning signal and a red alarm signal, which correspond to suspected abnormality, abnormality to be confirmed and abnormality state to be diagnosed respectively. Further, the intelligent negative pressure control module is linked with the microprocessor and the miniature pressure sensor and is configured to dynamically adjust the negative pressure value according to the pressure in the drainage tube, and the adjusting precision is +/-2 mmHg. According to a second aspect of the embodiment of the present invention, an intelligent drainage metering and early warning method based on a multi-sensor fusion and LSTM prediction model is applied to any one of the above intelligent drainage metering and early warning devices based on a multi-sensor fusion and LSTM prediction model, and the method includes: collecting drainage liquid by using a drainage liquid