Search

CN-122025177-A - Anesthesia clinical data management method and system based on big data acquisition

CN122025177ACN 122025177 ACN122025177 ACN 122025177ACN-122025177-A

Abstract

The invention discloses an anesthesia clinical data management method and system based on big data acquisition, and relates to the technical field of clinical data management, comprising the following steps of acquiring original time signals of a plurality of artificial intelligent chips in the process of parallel processing anesthesia clinical data, synchronously recording clock pulse changes, heart rate sampling rhythms and breath sampling rhythms of the artificial intelligent chips, and establishing a dynamic unified time reference basis; and performing time-sharing mapping on the time pulse difference of each artificial intelligent chip according to the established time reference basis. According to the invention, through a unified time reference and rhythm correction mechanism, the time consistency of the multi-artificial intelligent chip in parallel processing of anesthesia data is ensured, misjudgment caused by signal dislocation is avoided, and the stability of anesthesia monitoring and reasoning is improved. Meanwhile, through continuous buffer regulation and control and cyclic fine adjustment, time synchronization is continuously maintained under high concurrency operation, abnormal risk control is reduced, and safety and controllability of an anesthesia process are enhanced.

Inventors

  • CHEN JUNJIE

Assignees

  • 南通大学附属医院

Dates

Publication Date
20260512
Application Date
20260130

Claims (10)

  1. 1. The anesthesia clinical data management method based on big data acquisition is characterized by comprising the following steps: Collecting original time signals of a plurality of artificial intelligent chips in the process of parallel processing of anesthesia clinical data, synchronously recording clock pulse changes, heart rate sampling rhythm and breath sampling rhythm of each artificial intelligent chip, and establishing a dynamic unified time reference basis; performing time-sharing mapping on the time pulse difference of each artificial intelligent chip according to the established time reference basis, correspondingly comparing the advanced wave crest in the heart rate sampling rhythm with the delayed wave trough in the respiration sampling rhythm, extracting a time offset node from the advanced wave crest and the delayed wave trough in the respiration sampling rhythm, and constructing a dynamic delay compensation reference; Based on a dynamic delay compensation reference, performing smooth adjustment on the rhythm of each artificial intelligent chip in the data writing process, continuously correcting time offset nodes obtained by time division mapping, recovering the time continuity of heart rate sampling rhythm and breath sampling rhythm, and establishing a synchronization precondition of signal fusion; Step four, under the premise of synchronization, continuous buffer regulation and control are carried out on the input sequence of the real-time signal fusion process, and the sampling rhythm after smooth regulation is gradually input into a fusion channel according to the time sequence; And fifthly, performing cyclic fine adjustment on time pulse changes of all the artificial intelligent chips according to operation results of continuous buffer regulation and control, and maintaining synchronization stability among multiple chips based on recovered time continuity.
  2. 2. The method for managing anesthetic clinical data based on big data collection according to claim 1, wherein the step of collecting raw time signals of the plurality of artificial intelligence chips in parallel processing the anesthetic clinical data comprises: Extracting original time signals from time output ends of a plurality of artificial intelligent chips participating in parallel data processing, transmitting clock pulse signals generated by time pulse sources in each artificial intelligent chip to a shared data acquisition port through a time signal extraction path, and recording clock pulse change curves of the artificial intelligent chips with nanosecond time resolution; After the original time signal acquisition is completed, synchronous recording is carried out on the heart rate sampling rhythm and the respiration sampling rhythm generated by each artificial intelligent chip, the heart rate sampling signal and the respiration sampling signal are associated to the time stamp of the corresponding clock pulse signal in the recording process, and the time sequence of three types of data fields is kept consistent; Based on the output of the main time source, continuously tracking the clock pulse change, the heart rate sampling rhythm and the breath sampling rhythm on a unified time axis, and recording the time offset difference value of each artificial intelligent chip by comparing the pulse change trend of adjacent time periods; And integrating the time offset difference values of the artificial intelligent chips according to the time mapping results recorded on the unified time axis to form a continuously variable time curve, and mapping the clock pulse change curve, the heart rate sampling rhythm curve and the respiration sampling rhythm curve into the unified time axis.
  3. 3. The method for managing anesthetic clinical data based on big data collection according to claim 2, wherein the step of performing time-sharing mapping on the time pulse differences of the respective artificial intelligence chips according to the established time reference basis comprises: According to the established time reference basis, comparing the clock pulse changes of each artificial intelligent chip in the same time period, arranging clock pulse signals generated by each artificial intelligent chip in time sequence, and recording phase advance or delay parts existing in the time pulse changes in the corresponding time period in the form of offset values; After the time pulse difference among the artificial intelligent chips is obtained, performing time-sharing mapping on the time pulse difference, dividing a time pulse change interval of each artificial intelligent chip into a plurality of continuous time segments by taking a unified time axis as a reference, and establishing continuous mapping of a multi-chip time relationship in each time segment; When the time-sharing mapping is completed, corresponding comparison is carried out on the advanced wave crest in the heart rate sampling rhythm and the delayed wave trough in the respiration sampling rhythm, a time pairing relation between the heart rate wave crest and the respiration wave trough is established on a unified time axis, and time offset nodes of all the artificial intelligent chips under different time slices are recorded; when the time offset nodes are obtained, a continuous time difference curve is formed according to the arrangement sequence of the time offset nodes, and a dynamic delay compensation reference is constructed according to the curve.
  4. 4. The method for managing anesthetic clinical data based on big data collection according to claim 3, wherein the step of performing smooth adjustment of the rhythm of each artificial intelligence chip during the data writing process based on the dynamic delay compensation reference comprises: Based on a dynamic delay compensation reference, carrying out segmented extraction on time offset nodes of each artificial intelligent chip, and marking corresponding offset intervals on a time axis in a data writing process, wherein the time intervals between adjacent time offset nodes are defined as rhythm adjustment intervals so as to form continuous rhythm sections; When the time offset node subsection extraction is completed, smooth adjustment is performed on the data writing rhythms of all the artificial intelligent chips based on a dynamic delay compensation reference, a heart rate sampling rhythms leading part and a respiration sampling rhythms delaying part are synchronously brought into an adjustment range, and a time stepping relation is gradually adjusted to form a continuous rhythmic state; In the smooth adjustment process, continuously correcting the time offset node obtained by the time division mapping, correspondingly comparing the rhythm section after smooth adjustment with the offset node recorded in the time division mapping, gradually reducing each time difference value, and recovering the time corresponding relation between the heart rate sampling rhythm and the respiration sampling rhythm; When the continuous correction of the time offset node is completed, the integral balance adjustment is carried out on the data writing rhythm of each artificial intelligent chip, the time stepping information after the smooth adjustment and the continuous correction is integrated into a unified time axis, and the time continuity is maintained through continuous mapping.
  5. 5. The anesthesia clinical data management method based on big data collection according to claim 4, wherein in the process of carrying out overall balance adjustment on the data writing rhythms of the artificial intelligent chips, the time stepping information after smooth adjustment and continuous correction is sequentially input according to the time sequence of a unified time axis, the time stepping information is synchronously adjusted through the real-time update of a dynamic delay compensation reference, and the continuous correspondence relationship between the heart rate sampling rhythms and the respiration sampling rhythms in the data writing stage is maintained.
  6. 6. The method for managing anesthetic clinical data based on big data acquisition according to claim 4, wherein the step of performing continuous buffer control on the input sequence of the real-time signal fusion process under the synchronization premise includes: Performing time initialization of an input buffer zone on the heart rate sampling rhythm and the breath sampling rhythm which are subjected to smooth adjustment, arranging sampling data output by a plurality of artificial intelligent chips in time sequence, and setting a continuous buffer zone on an input channel to temporarily store signals so as to align heart rate sampling values and breath sampling values in the same time window on a time axis; Performing continuous regulation and control on the input sequence of the real-time signal fusion process, and controlling the signal input rhythms of all the artificial intelligent chips according to a unified time axis, so that heart rate sampling rhythms and respiration sampling rhythms are sequentially advanced in fixed time steps in adjacent time windows; in the continuous regulation and control process, real-time buffer control is carried out on the sampling rhythm after smooth regulation, the sampling rhythm is input after peak advance time delay occurs in the heart rate sampling rhythm, signals are temporarily stored when delay occurs in the respiration sampling rhythm, and time synchronization input of the two signals is kept; And executing time closed-loop balance adjustment on the input sequence of the real-time signal fusion process, continuously inputting the smoothly adjusted sampling rhythm into the fusion channel according to the time sequence, ensuring the integrity of the time rhythm and realizing signal synchronization in the input stage.
  7. 7. The anesthesia clinical data management method based on big data acquisition according to claim 6, wherein in the time closed loop balance adjustment process, input sequences of heart rate sampling rhythm signals and breath sampling rhythm signals of all artificial intelligent chips are dynamically coordinated according to a unified time axis, a time matching relation between heart rate peaks and breath troughs is kept in each time window, and input operation of a next time window is automatically triggered when signal input is completed.
  8. 8. The method for managing anesthetic clinical data based on big data collection according to claim 6, wherein the step of performing cyclic fine adjustment on time pulse variation of each artificial intelligence chip according to an operation result of continuous buffer adjustment comprises: The method comprises the steps of integrally collecting and grouping time pulse changes of all artificial intelligent chips in a current operation period, and comparing time offset of each artificial intelligent chip with a unified time reference to form an initial data set containing a time offset state; step-by-step cyclic adjustment is carried out on the time pulse change according to the running result of continuous buffer regulation and control, and tiny time differences existing among all artificial intelligent chips are gradually eliminated through adjustment of time output periods, so that a self-adaptive time balance mechanism is formed; performing continuous dynamic tracking and micro-amplitude balancing on the corrected time output, and incorporating micro-time fluctuation of each artificial intelligent chip in continuous operation into a circulation fine tuning process in real time so as to maintain stable extension of time continuity; And (3) performing periodic comprehensive coordination on time pulse change results of all artificial intelligent chips, and realizing time output period matching through uniform mapping of time stepping intervals.
  9. 9. The method for managing anesthetic clinical data based on big data acquisition according to claim 8, wherein when the cyclic fine adjustment of the time pulse variation is performed, the artificial intelligence chips in the time advance state are adjusted by widening the time pulse interval, and the artificial intelligence chips in the time delay state are adjusted by contracting the time pulse interval, so that the time pulse waveforms of the artificial intelligence chips are kept synchronously advancing on the same time axis.
  10. 10. The anesthesia clinical data management system based on big data acquisition is used for realizing the anesthesia clinical data management method based on big data acquisition according to any one of the claims 1-9, and is characterized by comprising a time reference building module, a delay compensation building module, a rhythm smoothing and adjusting module, a signal buffering and fusing module and a synchronous circulation fine adjustment module; The time reference establishing module is used for acquiring original time signals of the plurality of artificial intelligent chips in the process of parallel processing of anesthesia clinical data, synchronously recording clock pulse changes, heart rate sampling rhythms and breath sampling rhythms of the artificial intelligent chips, and establishing a dynamic uniform time reference basis; The delay compensation construction module is used for executing time-sharing mapping on the time pulse difference of each artificial intelligent chip according to the established time reference basis, correspondingly comparing the advanced wave crest in the heart rate sampling rhythm with the delay wave trough in the respiration sampling rhythm, extracting a time offset node from the advanced wave crest and the delay wave trough in the respiration sampling rhythm, and constructing a dynamic delay compensation reference; the rhythm smoothing adjustment module is used for carrying out smoothing adjustment on the rhythm of each artificial intelligent chip in the data writing process based on the dynamic delay compensation reference, continuously correcting the time offset node obtained by the time division mapping, recovering the time continuity of the heart rate sampling rhythm and the respiration sampling rhythm, and establishing a synchronization precondition of signal fusion; the signal buffer fusion module executes continuous buffer regulation and control on the input sequence of the real-time signal fusion process on the premise of synchronization, and gradually inputs the smoothly regulated sampling rhythm into the fusion channel according to the time sequence; And the synchronous circulation fine adjustment module is used for executing circulation fine adjustment on the time pulse change of each artificial intelligent chip according to the running result of continuous buffer regulation and control and maintaining the synchronization stability among the multiple chips based on the recovered time continuity.

Description

Anesthesia clinical data management method and system based on big data acquisition Technical Field The invention relates to the technical field of clinical data management, in particular to an anesthesia clinical data management method and system based on big data acquisition. Background The anesthesia clinical data management based on big data acquisition refers to the process of acquiring physiological parameters (such as blood pressure, heart rate, blood oxygen, brain electricity, respiratory rate, anesthesia medicine dosage, reaction and the like) of a patient in real time through multi-source medical equipment and sensors in the whole anesthesia process, and constructing a high-precision multi-dimensional clinical data set. And (3) performing high-speed processing, noise filtering and feature extraction on the acquired original signals by relying on an intelligent terminal or an edge computing node of the embedded AI chip, and comparing and deep learning modeling on the real-time data stream and a historical case database. The intelligent recognition and prediction of anesthesia depth, vital sign trend and potential risk are realized through AI chip accelerated neural network reasoning, and the anesthetic drug administration strategy and life support parameters are dynamically adjusted. And finally, a traceable, quantifiable and determinable anesthesia clinical big data system is formed in a unified data management platform, and real-time support is provided for accurate anesthesia control, intraoperative safety early warning and an individual anesthesia scheme. The prior art has the following defects: In the process of cooperatively processing anesthesia clinical data by a plurality of AI chips, microsecond delay exists in a link synchronization signal, so that phase dislocation is easily caused in time references among different chips. When the heart rate signal and the respiration signal enter the fusion channel in an asynchronous state, heart rate peaks are overlapped into the respiration signal sequence in advance, and false fluctuation which is inconsistent with the actual respiration rhythm is formed. When the system performs real-time reasoning, the abnormal superposition is misidentified as an apnea event, so that a wrong safety intervention instruction is triggered. Such problems are not common, but once they occur, they are very likely to cause imbalances in anesthesia depth control, disturbance in the rhythm of the breathing machine, and abnormal drug output, and in severe cases, they may cause temporary hypoxia or anesthesia interruption in the patient. The above information disclosed in the background section is only for enhancement of understanding of the background of the disclosure and therefore it may include information that does not form the prior art that is already known to a person of ordinary skill in the art. Disclosure of Invention The invention aims to provide an anesthesia clinical data management method and system based on big data acquisition, so as to solve the problems in the background technology. In order to achieve the above purpose, the invention provides the following technical scheme that the anesthesia clinical data management method based on big data acquisition comprises the following steps: Collecting original time signals of a plurality of artificial intelligent chips in the process of parallel processing of anesthesia clinical data, synchronously recording clock pulse changes, heart rate sampling rhythms and breath sampling rhythms of the artificial intelligent chips, and executing continuous tracking on a unified time axis to establish a dynamic unified time reference basis; performing time-sharing mapping on the time pulse difference of each artificial intelligent chip according to the established time reference basis, correspondingly comparing the advanced wave crest in the heart rate sampling rhythm with the delayed wave trough in the respiration sampling rhythm, extracting a time offset node from the advanced wave crest and the delayed wave trough in the respiration sampling rhythm, and constructing a dynamic delay compensation reference; based on a dynamic delay compensation reference, performing smooth adjustment on the rhythm of each artificial intelligent chip in the data writing process, continuously correcting the time offset node obtained by the time division mapping, and recovering the time continuity of the heart rate sampling rhythm and the respiration sampling rhythm to establish a synchronous premise of signal fusion; on the premise of synchronization, continuous buffer regulation is carried out on the input sequence of the real-time signal fusion process, the sampling rhythm after smooth regulation is gradually input into the fusion channel according to the time sequence, the heart rate wave crest is prevented from being overlapped into the respiratory sampling rhythm in advance, and the generation of false fluctuation is restrained from th