CN-122025063-A - Intelligent medical equipment asset management method based on digital twinning

CN122025063ACN 122025063 ACN122025063 ACN 122025063ACN-122025063-A

Abstract

The invention belongs to the technical field of intelligent medical equipment asset management, and particularly relates to an intelligent medical equipment asset management method based on digital twinning. Aiming at the problems of poor real-time performance, low scheduling efficiency, untimely maintenance and the like of the traditional medical equipment management, the invention constructs a hospital BIM model and an equipment life cycle data set, utilizes an improved LSTM network to generate an equipment digital twin intelligent body and embeds the BIM model to form a digital twin model, realizes the real-time positioning of the equipment based on the model, combines health evaluation and improved reinforcement learning to generate a borrowing scheme, dynamically generates a maintenance plan according to an equipment idle busy kurtosis threshold and health degree, and feeds back and updates. The invention realizes the fine management of the whole life cycle of the equipment, improves the scheduling efficiency and the equipment utilization rate, reduces the resource waste and ensures the continuity and the safety of diagnosis and treatment services.

Inventors

YU HAINING
MENG XIANGYUAN
QI LIANG
ZHAO YANFU
ZHANG YANHAI
LIU YUQING
LI YINGXU
TIAN YE
SU ZHONGYUAN
MENG XIAOGUANG

Assignees

山东第一医科大学附属肿瘤医院(山东省肿瘤防治研究院、山东省肿瘤医院)
北京朔方天城智能科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260130

Claims (8)

1. An intelligent medical equipment asset management method based on digital twinning is characterized by comprising the following steps: S1, constructing a BIM model of a hospital space, and collecting operation data of a life cycle of intelligent medical equipment to form a life cycle data set of the intelligent medical equipment; S2, converting the life cycle data set of the intelligent medical equipment into a three-dimensional geometric model by utilizing a geometric modeling technology, and embedding the three-dimensional geometric model into a hospital BIM model to obtain a digital twin model; S3, realizing real-time positioning and tracking of the intelligent medical equipment based on the digital twin model, and carrying out priority evaluation by combining with the life cycle health degree of the equipment to generate a borrowing decision scheme of the intelligent medical equipment; And S4, setting a threshold value of the idle and busy kurtosis of the overall equipment, monitoring the idle and busy state of the overall equipment in real time through a digital twin model, and when the idle and busy kurtosis of the equipment is lower than a corresponding threshold value, combining the health degree of the equipment to generate a maintenance plan, and reversely feeding back a maintenance result to the digital twin body after the maintenance is executed to update the health degree of the equipment.
2. The method for asset management of intelligent medical equipment based on digital twinning according to claim 1, wherein the constructing a BIM model of a hospital space in step S1 comprises: Acquiring point cloud data of a main structure of a target space of a hospital through multiple devices, and unifying the point cloud data acquired by the devices to a global coordinate system of the target space through a time stamp alignment technology to form a global space point cloud data set; Dividing the global space point cloud data set into three levels of a main structure layer, a functional partition layer and a device association layer based on functional attributes of a target space, and giving exclusive semantic labels to point clouds of each level to obtain semantic point clouds; Constructing a scene specification constraint rule base and a semantic point cloud-BIM component mapping rule base in the target field, and automatically converting different semantic point cloud data into corresponding BIM components based on the two rule bases to obtain a hospital BIM model.
3. The digital twinning-based intelligent medical device asset management method according to claim 1, wherein the step S1 of collecting operation data of a life cycle of the intelligent medical device to form a life cycle data set of the intelligent medical device includes: setting an acquisition time point of the whole life cycle of the intelligent medical equipment, and acquiring operation data of the life cycle of the intelligent medical equipment according to the acquisition time point; carrying out data cleaning on the acquired data and carrying out normalization operation on the geometric data and the operation data to obtain normalized characteristic data; Designing a multi-level time window, extracting time sequence association features under different time granularities, quantifying fluctuation rules and association degrees of data in a continuous time sequence, and calculating the time sequence by the following steps: , wherein, For the time t acquisition time and the multi-scale time sequence associated characteristic value with the time window length w, In order to form a window in the inner layer, For normalized operational data at time t-k, Is an inner window centered on t-k The normalized data mean value of the data in the data storage unit, Is a window weight coefficient; Calculating a time sequence residual value according to a preset reference value: , wherein, In order to calculate the resulting time-series residual value, For the set reference value(s), Fusing time sequence associated characteristic values of all window weights for the time t; Normalized processing is carried out on the residual saliency to obtain , wherein, And Respectively the average value and standard deviation of the time sequence residual value in the history state; By using The principle judgment residual significance is normalized to obtain abnormal data; And removing the abnormal data, and complementing the operation data by using a linear interpolation method to form an intelligent medical equipment life cycle data set.
4. The method for asset management of intelligent medical equipment based on digital twinning according to claim 1, wherein the step S2 of converting the life cycle data set of the intelligent medical equipment into a three-dimensional geometric model by using a geometric modeling technique and embedding the three-dimensional geometric model into a hospital BIM model to obtain a digital twinning model comprises: Constructing an improved LSTM network model, and inputting the processed life cycle data set of the intelligent medical equipment into the network model to obtain future change data of the state change of the simulation equipment; constructing the obtained future change data of the state change of the simulation equipment into a three-dimensional model by utilizing the three-dimensional geometric model, and taking the three-dimensional model as a digital twin intelligent body of the intelligent medical equipment; And combining the digital twin intelligent agent of the intelligent medical equipment into the BIM of the hospital to obtain a digital twin model.
5. The method for digital twinning-based intelligent medical device asset management of claim 4, wherein said constructing an improved LSTM network model, inputting the processed intelligent medical device lifecycle data set into the network model to obtain future change data simulating device state changes, comprises: Constructing a residual enhancement cell state module, calculating residual items of the current candidate cell state and the historical cell state, overlapping the residual items with the candidate cell state, and then combining the historical features reserved by the forgetting gate with the current features screened by the input gate to finish updating of the residual enhancement cell state; constructing a prediction stationarity constraint output layer module, performing smoothing treatment on a basic predicted value obtained by mapping the cell state after residual enhancement, and outputting a continuous and steady equipment state predicted result; embedding a residual enhanced cell state module and a prediction stability constraint output layer module into a standard LSTM network architecture to form a complete improved LSTM network model; inputting the pretreated continuous intelligent medical equipment life cycle data set into the improved LSTM network model according to a time sequence order, taking the actual operation monitoring value of the equipment as a label, and optimizing network learning parameters through back propagation until a loss function converges; after training, inputting continuous life cycle data segments to be predicted, and outputting future change data simulating the state change of equipment through residual enhanced cell state updating and moving average stabilization processing.
6. The digital twinning-based intelligent medical equipment asset management method according to claim 1, wherein the step S3 is based on a digital twinning model to realize real-time positioning and tracking of the intelligent medical equipment, and performs priority evaluation in combination with life cycle health of the equipment, and generates a borrowing decision scheme of the intelligent medical equipment, and the method comprises the following steps: selecting any point in the hospital space BIM as an original coordinate to obtain a coordinate system so as to define coordinate values of all intelligent medical equipment; calculating three sections of operation data sequences from the current moment to three preset time points in the future by using a digital twin model, wherein the three preset time points are 24 hours, 48 hours and 72 hours with the current moment as the base point; Simultaneously acquiring actual operation data sequences, respectively calculating the time deviation rates of the predictions of 24 hours, 48 hours and 72 hours and the actual acquisition time, and carrying out time sequence average on all the acquisition time to obtain the time sequence average deviation rates of 24 hours, 48 hours and 72 hours; And respectively calculating three health indexes at preset time through an exponential function: , wherein, The method comprises the steps of judging weight according to predicted time, and weighting three preset time points to obtain life cycle health degree of the equipment; generating a priority index of the equipment according to the life cycle health of the equipment, and taking the highest priority and the nearest scheduling path as optimization targets, wherein the scheduling path is calculated by a constructed coordinate system, and generating a borrowing decision scheme of the intelligent medical equipment by improving a reinforcement learning method.
7. The method for asset management of intelligent medical equipment based on digital twinning according to claim 6, wherein the optimization objective is the highest priority and the nearest dispatch path, the dispatch path is calculated by the constructed coordinate system, and the borrowing decision scheme of the intelligent medical equipment is generated by improving reinforcement learning, comprising: Constructing a state space, taking a global coordinate system established by a hospital BIM model as a reference, extracting multidimensional real-time state parameters of intelligent medical equipment in a digital twin model, and constructing a reinforcement learning state vector The method comprises the steps of including coordinate values of equipment, distance values to a demand point, life cycle health and real-time idle state identification of the equipment; Setting a reward function by carrying out normalization weighting on the highest priority and the latest dispatching path, wherein the priority is the ratio of the current equipment life cycle health value to the maximum life cycle health value, and the dispatching path is the ratio of the 1-dispatching path length to the maximum dispatching path length; Construction of an action space Wherein each action Setting action validity screening rules, and judging the action to be valid action only when the real-time idle state mark of the equipment is empty and matching requirements exist; Constructing a residual enhancement strategy network architecture, setting N full-connection hidden layers, wherein layers 1 to N-1 are residual blocks, and defining residual mapping in the residual blocks Identity mapping , wherein, For the input data of the k-th layer residual block, F () and I () are the residual mapping function and identity mapping function, respectively, and the residual block is output The Nth layer is an output layer, and the N-1 th layer residual block is input to output and the action probability is output through the full connection layer and the Softmax activation function; Calculating accumulated benefits from the current moment to the end of the decision period based on the reward function as a feedback signal for strategy updating; Updating rule based on strategy gradient, mapping parameters to residual blocks of residual enhancement strategy network And (5) performing iterative updating: wherein And (3) iteratively updating the feedback signals updated by the strategy until the network converges, and generating an optimal intelligent medical equipment borrowing decision scheme.
8. The method for asset management of intelligent medical equipment based on digital twinning according to claim 1, wherein the step S4 sets a threshold of the idle kurtosis of the overall equipment, monitors the idle status of the overall equipment in real time through a digital twinning model, and when the idle kurtosis of the equipment is lower than the corresponding threshold, generates a maintenance plan in combination with the health degree of the equipment, comprising: and taking the ratio of the number of the used devices to the total number of the devices as the idle busy kurtosis of the devices, presetting an idle busy kurtosis threshold, and carrying out device maintenance according to the reverse order of the health degree of the devices when the idle busy kurtosis of the devices is lower than the corresponding threshold.

Description

Intelligent medical equipment asset management method based on digital twinning Technical Field The invention belongs to the technical field of intelligent medical equipment asset management, and particularly relates to an intelligent medical equipment asset management method based on digital twinning. Background Along with the improvement of medical informatization and intelligent level, intelligent medical equipment has become a core support of medical treatment services of hospitals, the number and the variety of the intelligent medical equipment continuously increase, the intelligent medical equipment is distributed in all functional areas of the hospitals, the life cycle covers a plurality of links such as operation, scheduling and maintenance, and the intelligent medical equipment has huge data volume and strong time sequence. The traditional intelligent medical equipment asset management relies on manual recording and experience judgment, and has the problems of poor real-time performance, fuzzy positioning, inaccurate data processing and the like, so that the equipment utilization rate is unbalanced, the borrowing scheduling efficiency is low, the maintenance is not timely, the medical resource waste is caused, and the continuity and the safety of diagnosis and treatment service can be influenced. Disclosure of Invention Aiming at the technical problems in the background technology, the invention provides an intelligent medical equipment asset management method based on digital twinning. In order to achieve the above purpose, the technical scheme adopted by the invention comprises the following steps: S1, constructing a BIM model of a hospital space, and collecting operation data of a life cycle of intelligent medical equipment to form a life cycle data set of the intelligent medical equipment; S2, converting the life cycle data set of the intelligent medical equipment into a three-dimensional geometric model by utilizing a geometric modeling technology, and embedding the three-dimensional geometric model into a hospital BIM model to obtain a digital twin model; S3, realizing real-time positioning and tracking of the intelligent medical equipment based on the digital twin model, and carrying out priority evaluation by combining with the life cycle health degree of the equipment to generate a borrowing decision scheme of the intelligent medical equipment; And S4, setting a threshold value of the idle and busy kurtosis of the overall equipment, monitoring the idle and busy state of the overall equipment in real time through a digital twin model, and when the idle and busy kurtosis of the equipment is lower than a corresponding threshold value, combining the health degree of the equipment to generate a maintenance plan, and reversely feeding back a maintenance result to the digital twin body after the maintenance is executed to update the health degree of the equipment. Preferably, the constructing a BIM model of the hospital space in step S1 includes: Acquiring point cloud data of a main structure of a target space of a hospital through multiple devices, and unifying the point cloud data acquired by the devices to a global coordinate system of the target space through a time stamp alignment technology to form a global space point cloud data set; Dividing the global space point cloud data set into three levels of a main structure layer, a functional partition layer and a device association layer based on functional attributes of a target space, and giving exclusive semantic labels to point clouds of each level to obtain semantic point clouds; Constructing a scene specification constraint rule base and a semantic point cloud-BIM component mapping rule base in the target field, and automatically converting different semantic point cloud data into corresponding BIM components based on the two rule bases to obtain a hospital BIM model. Preferably, the step S1 collects operation data of the life cycle of the intelligent medical device to form a life cycle data set of the intelligent medical device, including: setting an acquisition time point of the whole life cycle of the intelligent medical equipment, and acquiring operation data of the life cycle of the intelligent medical equipment according to the acquisition time point; carrying out data cleaning on the acquired data and carrying out normalization operation on the geometric data and the operation data to obtain normalized characteristic data; Designing a multi-level time window, extracting time sequence association features under different time granularities, quantifying fluctuation rules and association degrees of data in a continuous time sequence, and calculating the time sequence by the following steps: , wherein, For the time t acquisition time and the multi-scale time sequence associated characteristic value with the time window length w,In order to form a window in the inner layer,For normalized operational data at time t-k,Is an inner window