AI-driven clinical research data integration and achievement transformation auxiliary system

Abstract

The application discloses an AI-driven clinical scientific research data integration and achievement transformation auxiliary system, relates to the technical field of medical informatization and artificial intelligence, and comprises an absolute time context reconstruction module, a phase conflict identification module, a trend direction correction module, a time sequence scheduling planning module and a dynamic rhythm regulation module. The application realizes the gradual connection among the absolute time context reconstruction, the phase conflict identification and the trend direction correction, restores the continuous trend of multi-period and multi-source data under the unified reference, and obtains the stable direction structure before the data is collected and exchanged. Meanwhile, through the cooperation of staggered entry, rhythm arrangement and rhythm regulation, the data continuously maintains the consistent direction in the dynamic promotion, the data flow gradually gets rid of the direction superposition caused by the phase misplacement in the promotion process through the alternate actions of mirror back-annotation, phase folding and front unloading according to the time sequence scheduling table, and the trend keeps the real extension in the fusion stage, thereby providing a reliable basis for scientific research analysis.

Description

Technical Field

[0001] The present invention relates to the technical fields of medical informatization and artificial intelligence, and particularly relates to an AI-driven clinical research data integration and result transformation assistance system. Background Art

[0002] An AI-driven clinical research data integration and result transformation assistance system refers to a comprehensive assistance system facing the scenarios of hospital research and translational medicine, which conducts unified integration, intelligent analysis and correlation modeling on clinical data, experimental data, follow-up data, literature materials and multi-source heterogeneous research information through artificial intelligence technology. Based on the capabilities of natural language processing, knowledge graph construction, time series pattern recognition, causal association mining and automated scientific research analysis, this system introduces a self-learning and continuous evolution mechanism. By continuously tracing historical research projects, research path selection, analysis result feedback and result transformation effectiveness, it constantly corrects data association rules, optimizes feature expression methods and strengthens model deduction logic, enabling the system to gradually improve the understanding depth and response accuracy of research questions with the accumulation of scientific research practice. Thus, it can stably transform the scientific research data scattered in different platforms and different formats into structured, computable and traceable knowledge units, and generate a scientific research hypothesis verification path, statistical analysis suggestions, research result interpretation logic and preliminary text materials that can be directly used for paper, fund or result declaration that are more suitable for actual research scenarios under the drive of self-learning, significantly improving the data utilization efficiency, research modeling accuracy and scientific research result transformation speed in the clinical research process, and realizing the full-process intelligent and continuously evolving assistance from clinical data to scientific research output.

[0003] The prior art has the following deficiencies: In the prior art, the convergence of multi-period observation data usually relies on algorithms such as timestamp alignment, sampling frequency unification and feature sequence reconstruction to achieve the fusion of multi-source time series data. However, when there are sampling delays, signal drifts or phase alignment errors between different observation periods, it is extremely easy to出现相位反转现象,使趋势识别模块将真实下降趋势误解为上升趋势,造成关键指标在系统中被以完全相反方向解读。此类相位反转不仅破坏了时间序列的连续性,还会引发模型在特征提取与模式识别阶段的误导性反馈,使干预方向偏离实际生理变化轨迹,机制假设依据被错误构建,研究结论在因果推断与统计验证过程中产生根本性偏差,从而导致整个科研数据分析链条的可靠性与成果转化有效性严重受损。

[0004] The above information disclosed in the background art section is only used to strengthen the understanding of the background of the present disclosure, and thus it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention It should be noted that there is an unclear part in your original text (the garbled part in ""), which I have left as it is in the translation. You may check and correct it for a more accurate translation.

[0005] The object of the present invention is to provide an AI-driven clinical research data integration and result transformation assistance system to solve the problems in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions: An AI-driven clinical research data integration and result transformation assistance system, including an absolute time context reconstruction module, a phase conflict identification module, a trend direction correction module, a timing scheduling and planning module, and a dynamic rhythm regulation module: The absolute time context reconstruction module constructs an absolute time context based on the timestamp differences of multiple data sources, extracts the peak scale and valley scale within each observation period, calibrates the phase start point and phase end point, and generates a complete phase anchor point band for providing a unified trend tracking reference; The phase conflict identification module extends the critical inflection point flow trajectories of each observation period based on the phase anchor point band, identifies the direction mutation regions and generates forward and reverse conflict bands, forming a reverse suspicion window; The trend direction correction module extracts the abnormal segment boundaries according to the reverse suspicion window, establishes a continuous direction guiding line for each time series curve, and generates a direction correction index for re-normalizing the curve direction in the time dimension; The timing scheduling and planning module plans the staggered access order of different data streams based on the direction correction index, performs rhythm arrangement through micro-delay, short silence, and light replay, and generates a timing scheduling table to achieve flexible connection of data streams in both the time dimension and the space dimension; The dynamic rhythm regulation module performs alternating actions of mirror re-injection, phase reversion, and pre-loading during the data convergence stage according to the timing scheduling table, forms an adaptive rhythm regulation loop, and eliminates the superposition of trend reversals from the dynamic dimension and solidifies the true trend direction.

[0007] Optionally, the steps of constructing the absolute time context and generating the phase anchor point band are as follows: Sort out the original time records from multiple data sources, so that the time information under different carriers, different formats, and different relative time systems is incorporated into a single time reference, and reveal the cross-source time differences by comparing time marks, so as to form a continuous time context covering all observation periods; Identify the peak scale and valley scale of each observation period under the constraint of the continuous time context, confirm the high-value interval and low-value interval under the absolute time reference, and provide a structural basis for the calibration of the phase start point and phase end point; Determine the phase start point and phase end point of each observation period with the peak scale and valley scale as the boundaries, so that the phase segments form a phase structure with a front-back relationship along the continuous time context; All phase segments are placed in a continuous time context and connected laterally, so that each observation period forms an overlayable structural strip and generates a complete phase anchor band that can be used across periods to provide a unified trend tracking reference.

[0008] Optionally, when generating the phase anchor band, the peak and valley scales of each observation period are arranged sequentially in the continuous time context to ensure that the peak-valley correspondence remains consistent across different periods. Furthermore, by continuously extending the phase start and phase end points, a tight connection is formed between phase segments, allowing the phase anchor band to exhibit a stable trajectory across the period structure, thereby enhancing the consistency of trend tracking.

[0009] Optionally, the process of forming a reversed suspect window based on the phase anchor point band includes the following steps: Given that the phase anchor point band presents peak scale, valley scale, and phase start and phase end points, the key inflection points of each observation cycle are identified and their corresponding positions in the absolute time context are completed, so that the key inflection points form a continuous flow trajectory along the phase anchor point band. By comparing the flow trajectory horizontally across the cycle range, the trajectories with separated trends will form directional offset characteristics, and each flow trajectory will obtain directional distribution information in the absolute time context. Based on the information on directional distribution, the relationship between the preceding and following directions is sorted out, so that the concentrated area of ​​directional deviation forms a concentrated zone of directional change, and the forward and reverse zones are formed by integrating the consistency of the trend. Based on the forward and reverse bands, boundaries are drawn so that adjacent areas are organized as windows of potential reversal, used to present the risk of trend deflection.

[0010] Optionally, the formation of the reversal suspect window is achieved by defining the intersection edge of the forward and reverse bands at the concentration band of directional change, and rearranging the intersection edge into a single continuous region in the continuous time context of the phase anchor band, so that the directional offset is concentrated in this region, thereby making the trend deflection position clearly expressed in the absolute time frame and used for subsequent directional adjustments.

[0011] Optionally, the steps for extracting the boundaries of abnormal segments and generating a direction correction index based on the inverted suspect window are as follows: After revealing the location of the directional shift by reversing the suspect window, the time segment corresponding to the reversing suspect window is expanded segment by segment in each time series curve, so that the local changes are presented continuously in the absolute time context, and potential boundary clues of abnormal segments are identified accordingly. After identifying potential boundary clues, these clues are combined with the undulating structure of the curve to make the start and end points of the anomalous segments form continuous and comparable boundaries in the absolute time context, and to clearly separate the directional changes within the segments. After the boundary of the abnormal segment is clearly defined, the directional trend of the stable segment is extended from the outer extension of the cycle to the boundary of the abnormal segment, so that the directional offset position is connected with the stable segment, and a continuous directional line containing the stable segment and the adjusted segment is formed. After the direction continuous lines are formed, the direction markers in the lines are matched with the original directions of the time-series curves, so that the direction offset segments can be adjusted under the mapping relationship, and a direction correction index covering the entire curve is generated to achieve direction normalization in the absolute time context.

[0012] Optionally, the generation of the direction correction index is limited to the node order of the continuous direction lines as the sole basis, so that the direction identifier of each node is continuously extended with the adjacent nodes, and the mapping relationship is continuously maintained with the absolute time context, so that the direction unification continues to advance continuously throughout the process.

[0013] Optionally, the steps for planning the staggered entry order of different data streams based on the direction correction index and generating a time-series scheduling table are as follows: After marking the direction adjustment nodes of each time-series curve with the direction correction index, the trend extension method of the data stream is classified according to the distribution of the direction adjustment nodes, so that the data stream forms an initial entry sequence in the absolute time context. Based on the initial entry sequence, a micro-delay is introduced before the data stream reaches the key nodes, and short silences are set at some entry nodes to create time intervals in the pace of the data stream's advancement within the absolute time context. After micro-delay and short silence form the basic rhythm, light playback is introduced to allow the data stream that has completed the directional adjustment to be appropriately pulled back in the absolute time context and to form a flexible connection with the preceding data stream. After the rhythm is choreographed, the specific moments of the data stream in micro-delay, short silence and light playback are organized according to the absolute time context, so that all rhythm actions form a time sequence schedule, which is used to realize the flexible connection of the data stream in the time dimension and the spatial dimension.

[0014] Optionally, after the light replay is completed, the entry rhythm forms a progressive sequence by continuing the micro-delay, so that each data stream maintains a stable distance before the key nodes of the absolute time context, and enters the convergence position in the order of the timing schedule after the short silence ends, so that all data streams maintain directional consistency and continuously follow the extended trajectory of the direction correction index during the rhythm connection process.

[0015] Optionally, the steps for performing alternating actions of mirror injection, phase foldback, and pre-unloading according to the timing schedule table to form an adaptive rhythm control cycle are as follows: After the timing schedule table presents a rhythm arrangement of micro-delay, short silence and light replay, the advancement position of each data stream is confirmed according to the timing schedule table, and mirror back-injection is performed before the data stream reaches its key node specified in the timing schedule table, so that the fluctuation segment after the direction adjustment is back-injected to the previous position in advance. After the mirror injection is completed, phase reversal is performed according to the timing schedule table, so that the data stream in the direction intersection interval rotates back along the direction of the phase starting point, so that the direction acceleration segment and the direction deceleration segment form a continuous transition in the absolute time context; After the phase reversal is completed, the pre-unloading is performed according to the timing schedule table, so that the fluctuation segment with the risk of directional deviation is removed from the propulsion path in advance, reserving space for the subsequent data flow. After the mirror injection, phase reversal and pre-unloading are completed in sequence, the above actions are alternately cycled according to the time window according to the timing schedule, so that the directional deviation is gradually absorbed and the true trend direction is solidified.

[0016] The beneficial effects of the technical solution provided by this invention include at least the following: This invention achieves a unified temporal reference at the structural level by seamlessly integrating absolute timeline reconstruction, phase conflict identification, and trend direction correction. This ensures consistent representation of directional changes within a global framework. By placing peak and trough scales, phase boundaries, inflection point flow directions, and boundary information of anomalous segments within the same temporal context, directional shifts in the data are promptly captured and rearranged under the guidance of a direction correction index, allowing the trend to be restored before convergence. The resulting data foundation not only maintains structural coherence but also ensures sustained trend stability, providing a more reliable trajectory for subsequent scientific analysis.

[0017] This invention, through the integrated linkage of staggered entry, rhythmic arrangement, and cyclical regulation, enables data to possess flexible connectivity during the convergence phase, continuously consolidating directional consistency in a dynamic dimension. By executing alternating actions of mirror injection, phase reversal, and pre-unloading according to the time-series scheduling table, the data flow gradually overcomes the directional superposition caused by phase misalignment during its advancement, allowing the trend structure to continuously converge towards the true direction over time. This process endows the data flow with adaptive adjustment capabilities during the fusion phase, ensuring the stability of the trend direction unaffected by periodic differences, providing solid support for the reliability of research results and the accuracy of trend inference. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This is a schematic diagram of the modules of the AI - driven clinical research data integration and achievement transformation assistance system provided by the embodiments of the present invention. Detailed implementation manners

[0020] Example embodiments will now be described more fully with reference to the accompanying drawings. However, the example embodiments can be implemented in various forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that this disclosure will be more complete and comprehensive, and will fully convey the concept of the example embodiments to those skilled in the art.

[0021] The present invention provides an AI - driven clinical research data integration and achievement transformation assistance system as Figure 1 shown, including an absolute time context reconstruction module, a phase conflict identification module, a trend direction correction module, a time - series scheduling and planning module, and a dynamic rhythm regulation module: The absolute time context reconstruction module constructs an absolute time context based on the timestamp differences of multiple data sources, extracts the peak scale and valley scale within each observation period, calibrates the phase start point and phase end point, and generates a complete phase anchor band for providing a unified trend tracking reference. In order to establish a unified reference that can support trend tracking among multiple data sources with different sources, different acquisition rhythms, and different observation periods, it is necessary to gradually expand the time information and structural fluctuation information, so that each observation period has corresponding comparability on the time scale, thereby providing a solid foundation for the subsequent confirmation of the multi - period trend direction. The specific implementation steps are as follows: Sort the original time records from multiple data sources, so that the time information scattered in different carriers, different formats, and different relative time systems is incorporated into a single time reference. In this process, the time records within each data source are arranged coherently according to the acquisition order, and by comparing the time stamps corresponding to each data record, the time differences between different data sources are revealed, making the inherent acquisition sequence relationship between multiple observation periods clear. Subsequently, through the integration of such cross - source time differences, a continuous time context covering all observation periods is gradually formed, and this continuous time context is used as the basis for subsequent structural scale extraction, so that the data content from different periods is placed under the same absolute time reference.

[0022] After establishing the absolute timeline, it is necessary to identify peak and trough scales that stably represent the fluctuation structure within each observation period. To achieve this, the numerical fluctuations within each observation period are straightened out under the constraints of the absolute timeline, ensuring that the internal fluctuations are consistent with the global time relationship. In this way, the high and low value ranges within each period are clearly defined, allowing the peak and trough positions to be stably confirmed under absolute time reference. After confirming the peak and trough scales, the internal structure of each observation period has a basic framework for longitudinal comparison, enabling it to be identified as structural units with the same meaning in cross-period analysis. Since these scales are formed on the absolute timeline, the peak and trough scales are not only used for locating the internal structure within the current period but also for establishing correspondences with corresponding scales in other periods, providing conditions for the calibration of phase start and phase end points in subsequent steps.

[0023] After establishing the peak-valley scale, it is necessary to further define the phase start and end points of each cycle to clearly delineate the boundaries of the fluctuation units within the cycle. To this end, the peak-valley scale of each cycle is considered a key scale for dividing the internal structure of the cycle, and these scales are used as boundaries to break down the continuous fluctuations within the cycle into several phase segments. Subsequently, by continuously analyzing the arrangement of these phase segments in the absolute timeline, the phase segments of each cycle can be extended sequentially from their starting position to their ending position, forming a phase structure with a clear sequential relationship. In this process, the phase start point of each cycle is determined by the first structural scale in the absolute timeline, while the phase end point is determined by the last structural scale. This phase start and end point derived from the peak-valley scale ensures that each cycle not only has a temporal correspondence but also a similarity in its fluctuation structure, enabling effective comparison of the structures between different cycles.

[0024] After the phase start and end points are determined, phase segments within all periods are horizontally connected on an absolute timeline, forming a continuous structural band under the same time reference. As phase segments from all periods are arranged on the absolute timeline, the fluctuation structure of each period forms a corresponding band on the global time coordinate. These bands, when superimposed, form a phase anchor band usable across periods. This phase anchor band not only includes the peak-valley distribution and phase boundaries of each period but also exhibits a consistent structural trajectory across periods, enabling subsequent trend tracking to be conducted within a unified reference framework. By solidifying the phase anchor band, structural differences between multiple periods are absorbed within the constraints of a unified timeline, allowing subsequent trend direction tracking to overcome the influence of inter-period time offsets and fundamentally improving the consistency of trend understanding during multi-period data fusion.

[0025] The phase conflict identification module extends the flow trajectory of key inflection points in each observation cycle based on the phase anchor point band, identifies regions of sudden directional changes and generates forward and reverse conflict bands, forming a reversal suspected window; Following the generation of phase anchor bands based on the absolute time context, to further reveal the structural changes within each observation period under a unified reference, it is necessary to unfold the flow trajectory of key inflection points with the support of the phase anchor bands. This extends the trend direction within the period to a cross-period framework, creating conditions for identifying regions of abrupt directional changes and ultimately forming a reversal window to characterize potential trend reversal risks. The specific implementation steps are as follows:

[0026] Given that the phase anchor band clearly and completely presents the peak and trough scales, as well as the phase start and end points for each observation period, it is necessary to identify the key inflection points reflecting structural changes in each observation period based on the phase anchor band. This allows these inflection points to be positioned within the absolute time frame, corresponding to the phase anchor band. Through this positioning method, each inflection point is no longer merely a local fluctuation node within a period, but has a clear temporal location within the unified framework of the phase anchor band. Subsequently, by extending these key inflection points outward along the arrangement direction of the phase anchor band, the structural changes represented by each inflection point can be depicted continuously, forming flow trajectories within the absolute time frame that can be used for subsequent analysis. Since these flow trajectories are all based on the phase anchor band, the trend of each trajectory can obtain a consistent reference across multiple periods, allowing for equivalent comparison of the direction of change within different periods.

[0027] After the flow trajectories at key inflection points are extended to the absolute time frame, further attention needs to be paid to the potential directional divergences that may occur during the cross-cycle unfolding process. When the flow trajectories at inflection points in some cycles extend outwards along the phase anchor zone, while those in other cycles converge inwards, a directional difference arises within the same reference frame. To effectively identify this difference in subsequent processing, it is necessary to perform a horizontal comparison of trajectories belonging to the same phase segment but with separate trends, thereby revealing their inherent directional shifts. Through this comparison process, the flow trajectories at inflection points in each cycle are not only located within the phase anchor zone but are also further endowed with directional distribution information, giving them a continuous directional trend within the absolute time frame. This display of trends allows the flow trajectories from different cycles to be placed in mutually referential positions within the cross-cycle frame, providing a necessary foundation for the subsequent identification of directional abrupt change regions.

[0028] With the directional distribution information gradually becoming clear, further analysis of the continuous trend of the flow trajectory is needed to reveal a complete directional extension trend supported by the phase anchor zone. By analyzing the preceding and following extension relationships of each flow trajectory, it can be found that some trajectories exhibit sudden directional deviations during their extension, indicating a sharp reversal of the trend near that point. When such reversals occur not only within a single cycle but also across multiple cycles with similar directional changes in close proximity, these areas can be considered regions with directional abrupt changes. Under the condition of the phase anchor zone as a unified reference, these directional abrupt change regions can be concentrated within the absolute time context, forming concentrated zones of directional changes that can be identified across cycles. After the concentrated zones of directional changes are revealed, these regions need to be further integrated according to the consistency of their trends, resulting in a partitioned distribution of regions with unidirectional and opposite directional changes, thus forming forward and reverse zones, allowing different directional trends to obtain clear boundaries under the same reference coordinates.

[0029] After identifying the forward and reverse bands, it is necessary to further delineate the boundaries of these areas with directional divergence characteristics, enabling them to serve as concentrated spaces for trend reversal risks. During boundary delineation, it is crucial to integrate the concentrated bands of directional changes using phase anchor points, ensuring clear edges between the forward and reverse bands within the absolute timeframe. Subsequently, by organizing the adjacency relationships of the forward and reverse bands, these adjacent areas can be reorganized into reversal potential windows with specific structural attributes. The formation of reversal potential windows allows directional deviations across multiple periods to be aggregated and presented within a unified framework, clearly expressing the potential locations of trend direction deflections and providing necessary basis for subsequent unified handling of trend directions. Through the generation of reversal potential windows, the potential trend reversal risks across multiple periods are placed within an observable and locatable reference structure, allowing subsequent handling of trend reversal issues to proceed under clearer conditions.

[0030] The trend direction correction module extracts the boundary of abnormal segments based on the reversal suspect window, establishes continuous direction guide lines for each time series curve, and generates a direction correction index to re-normalize the curve direction in the time dimension. After establishing the suspected reversal window to indicate cross-cycle directional shifts, to restore the continuous direction of each time series curve under a unified cross-cycle reference, it is necessary to further analyze the fluctuation segments within the time series curves layer by layer, based on the directional conflict characteristics presented in the suspected reversal window. This clarifies the locations where trend shifts may occur, and on this basis, a continuous direction guideline is constructed. Under the new direction reference, a direction correction index is generated for each curve that can be used to re-normalize the direction. The specific implementation steps are as follows:

[0031] After the suspected reversal window has revealed the location of the concentrated directional shift, each time series curve needs to be expanded segment by segment according to the time interval corresponding to the suspected reversal window, so that the local changes of the curve within that segment are presented in a continuous manner. Since the suspected reversal window is usually located in the directional conflict area of ​​multiple cycles, each curve may contain segments within this area that are not easily detected structurally but play a key role in the overall trend. Therefore, in the segment corresponding to the suspected reversal window, it is necessary to clearly divide the continuous changes between adjacent points according to the absolute time context, so that areas with sudden increases in the rate of change, sudden reversals in the trend direction, or rapid accumulation of fluctuation density can be identified as potential boundary clues of abnormal segments. Subsequently, by superimposing the overall range of the suspected reversal window with the local change characteristics within each curve, those local segments that exhibit directional instability in the suspected reversal window can be extracted, thereby providing the necessary structural support for extracting the boundaries of abnormal segments.

[0032] After identifying curve segments with directional instability within the suspected reversal window, it is necessary to further clarify the boundaries of these segments, clearly separating them from the stable extensions of the curve. To this end, the starting and ending points of each fluctuation within a segment can be observed, focusing on the continuity of the trend changes within the curve. This allows these starting and ending points to form continuous and comparable boundaries within the absolute time frame. Subsequently, the potential boundary clues obtained in the previous step are combined with the undulating structure of the curve itself, ensuring that the boundaries not only reflect changes in the intensity of local fluctuations but also accurately represent the positional relationship of the segments within the suspected reversal window. Through this method, the boundaries of anomalous segments are stably defined, allowing directional changes within the segments to be fully expressed within a symmetrical boundary structure, providing a clear structural basis for subsequently establishing continuous directional guidelines. Because this boundary definition process relies on the suspected reversal window as a higher-level reference, the generated boundaries not only have intra-cycle significance but also cross-cycle temporal consistency.

[0033] Once the boundaries of anomalous segments are clearly defined, it is necessary to reconstruct the directional extension of each time series curve within the absolute time frame around these boundaries, ensuring a continuous expression of the curve's directional trend within the same reference framework. To this end, starting from the outer edge of each curve's period, the directional trend of stable segments is gradually extended along the time axis to the boundaries of anomalous segments, creating a continuous directional path for the stable segments. Subsequently, within the anomalous segments, the boundary sequence determined in the previous step is used to moderately smooth out areas with severe directional deviations, allowing them to connect with the trends of stable segments. Through this connection, each curve gradually forms a continuous directional line within the absolute time frame, composed of both stable and adjusted segments. Since these lines are constructed under the constraint of a reversal suspicion window, their directional extension can transcend period differences, gradually making the directional changes of each curve more consistent. With the establishment of continuous directional lines, the trend changes of each curve exhibit a natural extension pattern within the complete time frame, making directional guidance actionable.

[0034] After the continuous direction guideline has been constructed, a direction correction index needs to be generated to enable each time series curve to achieve directional normalization within the absolute time context. To this end, the direction identifier at each time point in the continuous direction guideline needs to be mapped one-to-one with the original direction of the corresponding curve, allowing all segments with offset directions to be repositioned under this mapping relationship. Through this mapping method, index entries for direction adjustment can be generated for each curve, with each index entry corresponding to a specific node in the absolute time context and indicating the appropriate direction extension method. As index entries are generated one by one, the direction correction index gradually forms a direction adjustment sequence covering the entire curve, enabling the curve to undergo directional consistency processing in subsequent steps based on the correction index. Since the direction correction index is built upon the continuous direction guideline, which in turn is based on the boundaries of abnormal segments, the direction correction index has a complete association from local structure to overall trend, allowing each curve to regain a unified directional presentation within the absolute time context, laying the foundation for subsequent trend normalization.

[0035] The timing scheduling and planning module plans the staggered entry order of different data streams based on the direction correction index. It uses micro-delay, short silence and light playback to arrange the rhythm and generate a timing scheduling table, so that the data streams can be flexibly connected in the time and space dimensions. After the direction correction index has re-normalized the directional extension of each time series curve in the absolute time context, in order to ensure that multiple data streams can enter the unified time frame in a coordinated state during subsequent data convergence, it is necessary to further rely on the direction adjustment information revealed in the direction correction index to plan the entry order of each data stream in the absolute time context, so that each data stream can obtain a flexible connection mode in both the time and spatial dimensions. The specific implementation steps are as follows:

[0036] After the direction correction index has clearly marked the direction adjustment nodes of each time series curve in the absolute time context, it is necessary to classify the trend extension methods of each data stream based on these nodes, so that they can form a distinguishable entry sequence in the subsequent entry sorting process. In this process, the marked positions of different data streams in the direction correction index are extracted one by one, allowing the dense and sparse areas of direction adjustment nodes to be visually represented in the absolute time context. This representation method clarifies where each data stream has a direction adjustment need, thereby determining the structural pressure it bears during the overall trend unification process. Subsequently, the structural features reflecting the direction adjustment density in the direction correction index are used as the basis for planning the staggered entry sequence, allowing data streams with more direction adjustment needs to obtain relatively relaxed entry positions in the overall entry process, while data streams with fewer direction adjustment needs can enter the absolute time context at earlier positions. This preliminary sorting method based on the direction correction index provides the foundation for implementing the staggered entry strategy.

[0037] After establishing an initial entry classification based on the direction correction index, it is necessary to further refine the entry rhythm of different data streams within the absolute time frame, ensuring that the entry process not only has temporal intervals but also rhythmic segments. To this end, micro-delays need to be introduced at critical moments when data streams enter the absolute time frame. This allows the data streams to be slightly shifted back in time as they approach the unified reference, preventing a concentrated influx of multiple data streams near the direction conversion node. Under the effect of micro-delays, each data stream has independent adjustment space when entering a critical node, allowing the direction adjustment action indicated by the direction correction index to be completed in a relatively stable context. Subsequently, to ensure that the segments between two micro-delays do not overlap excessively, short silences need to be set at some entry nodes, allowing the data streams to temporarily maintain their position upon arrival at these nodes, thus reserving space for the arrival of subsequent data streams. The introduction of short silences allows for an intermittent distribution of the data stream's progression rhythm along the time axis, thereby enhancing the overall coordination of the entry process.

[0038] After micro-delay and short silence have established a basic rhythmic framework, it is necessary to further introduce light replay on the entry path. This allows some data streams to retract moderately after completing directional adjustments, ensuring a smooth transition between the retracted fluctuation segments and the already adjusted data streams. Light replay not only allows the adjustment nodes in the directional correction index to connect more naturally with subsequent fluctuations but also enables the data streams to present a structure that better aligns with the overall trend after directional unification. With the gradual introduction of light replay, the position of each data stream in the absolute timeline can not only be adjusted forward or backward but also spatially create flexible spacing with other data streams, preventing tight compression between curves after directional adjustments. Through the gradual integration of micro-delay, short silence, and light replay, the staggered entry sequence of each data stream gradually forms a flexible, continuous, and coordinated structure, giving the overall entry process a coordinating characteristic.

[0039] After the choreography has assigned a differentiated entry rhythm to each data stream, these rhythmic actions need to be organized according to the continuous sequence of the absolute timeline, forming a timing schedule for subsequent execution. To do this, the specific moments of each data stream during micro-delays, short silences, and light replays need to be arranged chronologically, ensuring that the positional changes of each data stream within the absolute timeline are recorded in a clear sequence. Subsequently, by superimposing the temporal change sequences of all data streams, the global entry form of the overall data stream within the absolute timeline is presented. In this process, the timing schedule not only records the entry time of each data stream but also its spatial connection method, giving it the ability to reflect the overall rhythmic state. With the generation of the timing schedule, the progression of each data stream within the absolute timeline is fixed with a clear rhythmic sequence, allowing subsequent actions in the data convergence phase to unfold under a clear rhythmic arrangement. This achieves flexible connection of the data stream in the temporal and spatial dimensions, providing the necessary foundation for deeper trend solidification in the next step.

[0040] The dynamic rhythm control module performs alternating actions of mirror injection, phase reversal and pre-unloading during the data aggregation stage according to the time-series scheduling table, forming an adaptive rhythm control cycle, eliminating trend reversal superposition and solidifying the true trend direction from a dynamic dimension; After the aforementioned time-series scheduling table has clearly planned the pace of each data stream's progression within the absolute timeframe, in order to ensure that the data achieves stable directional unification and structural connection when entering the data convergence stage, it is also necessary to organize alternating actions of mirror injection, phase reversal, and pre-unloading during the data convergence phase according to the rhythm arrangement recorded in the time-series scheduling table. This allows the data to continuously receive dynamic support for directional correction as it enters the convergence node, thereby gradually forming a stable directional presentation pattern within the overall trend structure. The specific implementation steps are as follows:

[0041] After the timing schedule has fully presented the rhythm of micro-delay, short silence, and light replay, it is necessary to confirm the advancement position of each data stream at the beginning of the data convergence stage according to the timing schedule, so that the data stream has a clear rhythm positioning before entering the convergence position. Subsequently, in the process of advancing the data stream according to the rhythm positioning, a mirror injection action needs to be introduced when each data stream approaches its key node specified in the timing schedule. This allows the data stream to mirror the fluctuation segment that represents the direction adjustment in the direction correction index back to the preceding position of the original trajectory before reaching the key node. Through this mirror injection method, the data stream can integrate the structural changes caused by direction correction into its extension path in the absolute time context in advance, so that the position of potential direction deviation can be compensated for earlier. As the mirror injection action unfolds in multiple data streams one by one, the overall trend structure begins to show a flexible replenishment state before convergence, thus providing the necessary foundation for subsequent deeper direction control.

[0042] After mirror backinjection completes the directional supplementation of the preceding segments of each data stream, it is necessary to further adjust the trend of each data stream after entering the key position according to the timing schedule table, so that these trends can form a harmonious extension relationship with other data streams in the absolute time context. To this end, a phase reversal action needs to be introduced after mirror backinjection, so that some data streams in the directional convergence interval can turn back to the direction of the phase starting point during the advancement of the fluctuation segment, so that the directional acceleration segment and the directional deceleration segment can form a more natural transition in the absolute time context. Since the phase reversal action needs to determine the reversal amplitude based on the timing schedule table and the direction correction index, each data stream can maintain the directional consistency with other data streams during phase reversal, so that the direction conversion process will not produce new directional deviations. As the phase reversal action is gradually unfolded, the phase structure inside each data stream is gradually stretched or compressed into a form that is more suitable for presentation under a unified time reference, so that the data has more coordinated continuity in the convergence stage.

[0043] After phase reversal has sufficiently adjusted the direction conversion within each data stream, a pre-unloading action needs to be introduced into the data stream's propulsion path. This allows some fluctuating segments that have entered the pre-convergence stage but may have directional conflicts to be unloaded from the propulsion path in advance, preventing these segments from causing directional shocks just before entering the convergence node. To achieve this, based on the time window provided by the timing schedule table, segments marked as potentially having a risk of directional deviation in the direction correction index are moved forward out of the main propulsion path. This allows these segments to temporarily leave the propulsion sequence in the absolute time context, thus making room for subsequent fully synchronized segments. Through this pre-unloading method, the data stream no longer bears the additional pressure from directional conflicting segments when entering the convergence node, allowing the convergence operation to proceed under more stable conditions. With the completion of the pre-unloading action, the structure of each data stream near the convergence node shows a simpler and more continuous trend pattern, building a more stable foundation for the final solidification of the true trend direction.

[0044] After mirror injection, phase reversal, and pre-unloading have been sequentially performed, these actions need to be organized into a continuous loop according to the time sequence in the timing schedule, ensuring that the data stream is constantly under dynamic control throughout the convergence process. To this end, the pre-compensation formed by mirror injection, the directional transition formed by phase reversal, and the clearing space constructed by pre-unloading need to be arranged sequentially, so that the three actions repeat and alternate according to time windows within the absolute timeframe. With the continuous loop of these alternating actions, each data stream receives consistent directional support during its advancement, gradually strengthening the directional extension indicated in the directional correction index during the convergence process. With the long-term operation of this loop structure, directional deviations between data streams are continuously absorbed, and the risk of trend reversal and superposition is gradually mitigated, ensuring that the final trend pattern maintains continuous direction across cycles, sources, and structural characteristics. This solidifies the true trend direction in a dynamic dimension, maintaining a stable and extensible directionality throughout the entire data fusion process.

[0045] This invention achieves a unified temporal reference at the structural level by seamlessly integrating absolute timeline reconstruction, phase conflict identification, and trend direction correction. This ensures consistent representation of directional changes within a global framework. By placing peak and trough scales, phase boundaries, inflection point flow directions, and boundary information of anomalous segments within the same temporal context, directional shifts in the data are promptly captured and rearranged under the guidance of a direction correction index, allowing the trend to be restored before convergence. The resulting data foundation not only maintains structural coherence but also ensures sustained trend stability, providing a more reliable trajectory for subsequent scientific analysis.

[0046] This invention, through the integrated linkage of staggered entry, rhythmic arrangement, and cyclical regulation, enables data to possess flexible connectivity during the convergence phase, continuously consolidating directional consistency in a dynamic dimension. By executing alternating actions of mirror injection, phase reversal, and pre-unloading according to the time-series scheduling table, the data flow gradually overcomes the directional superposition caused by phase misalignment during its advancement, allowing the trend structure to continuously converge towards the true direction over time. This process endows the data flow with adaptive adjustment capabilities during the fusion phase, ensuring the stability of the trend direction unaffected by periodic differences, providing solid support for the reliability of research results and the accuracy of trend inference.

[0047] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Unless otherwise specified, an element defined by the phrase "comprising..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element.

[0048] The use of terms such as "an embodiment," "an embodiment," "an exemplary embodiment," and "some embodiments" in the specification indicates that the described embodiment may include a specific feature, structure, or characteristic, but not every embodiment necessarily includes that specific feature, structure, or characteristic. Furthermore, when a specific feature, structure, or characteristic is described in connection with an embodiment, implementing such a feature, structure, or characteristic in conjunction with other embodiments (whether explicitly described or not) should be within the knowledge of those skilled in the art.

[0049] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0050] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0051] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0052] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0053] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. An AI-driven auxiliary system for clinical research data integration and result transformation, characterized in that, It includes an absolute timeline reconstruction module, a phase conflict identification module, a trend direction correction module, a time series scheduling and planning module, and a dynamic rhythm control module: The absolute timeline reconstruction module constructs an absolute timeline based on the timestamp differences from multiple data sources. It extracts peak and valley scales within each observation period, calibrates the phase start and phase end points, and generates a complete phase anchor point band. The phase conflict identification module extends the flow trajectory of key inflection points in each observation cycle based on the phase anchor point band, identifies regions of sudden directional changes and generates forward and reverse conflict bands, forming a reversal suspected window; The trend direction correction module extracts the boundaries of abnormal segments based on the suspected reversal window, establishes continuous direction guide lines for each time series curve, and generates a direction correction index. The timing scheduling and planning module plans the staggered entry order of different data streams based on the direction correction index, and performs rhythm arrangement through micro-delay, short silence and light playback to generate a timing scheduling table; The dynamic rhythm control module performs alternating actions of mirror injection, phase reversal, and pre-unloading during the data aggregation stage according to the time-series scheduling table, forming an adaptive rhythm control cycle that eliminates trend reversal superposition and solidifies the true trend direction from a dynamic dimension.

2. The AI-driven clinical research data integration and achievement transformation assistance system according to claim 1, characterized in that, The steps for constructing the absolute time context and generating the phase anchor band are as follows: The raw time records from multiple data sources are organized so that time information from different carriers, formats and relative time systems can be integrated into a single time reference. Cross-source time differences are revealed by comparing time markers, thus forming a continuous timeline covering the entire observation period. Under the constraint of continuous time context, the peak and valley scales of each observation period are identified, so that the high and low value intervals can be confirmed under absolute time reference, and a structural basis is provided for the calibration of phase start and phase end. The phase start and phase end points of each observation period are determined by using the peak and valley scales as boundaries, so that the phase segments form a phase structure with a sequential relationship along the continuous time context; All phase segments are placed in a continuous time frame and connected laterally, so that each observation period forms a superimposed structural strip and generates a complete phase anchor band that can be used across periods.

3. The AI-driven clinical research data integration and achievement transformation assistance system according to claim 2, characterized in that, When generating the phase anchor band, the peak and valley scales of each observation period are arranged sequentially in the continuous time context to ensure that the peak-valley correspondence remains consistent across different periods. Furthermore, by continuously extending the phase start and phase end points, a tight connection is formed between phase segments, allowing the phase anchor band to exhibit a stable trajectory across the period structure, thereby enhancing the consistency of trend tracking.

4. The AI-driven clinical research data integration and achievement transformation assistance system according to claim 2, characterized in that, The process of forming a suspected inversion window based on the phase anchor point band includes the following steps: Given that the phase anchor point band presents peak scale, valley scale, and phase start and phase end points, the key inflection points of each observation cycle are identified and their corresponding positions in the absolute time context are completed, so that the key inflection points form a continuous flow trajectory along the phase anchor point band. By comparing the flow trajectory horizontally across the cycle range, the trajectories with separated trends will form directional offset characteristics, and each flow trajectory will obtain directional distribution information in the absolute time context. Based on the information on directional distribution, the relationship between the preceding and following directions is sorted out, so that the concentrated area of ​​directional deviation forms a concentrated zone of directional change, and the forward and reverse zones are formed by integrating the consistency of trends. Based on the forward and reverse bands, boundaries are drawn so that adjacent areas are organized as windows of potential reversal, used to present the risk of trend deflection.

5. The AI-driven clinical research data integration and achievement transformation assistance system according to claim 4, characterized in that, The formation of the reversal suspect window is achieved by defining the intersection edge of the forward and reverse bands at the concentration band of directional changes, and rearranging this intersection edge into a single continuous region in the continuous time context of the phase anchor band, so that the directional offset is concentrated in this region, thereby making the trend deflection position clearly expressed in the absolute time frame and used for subsequent directional adjustments.

6. The AI-driven clinical research data integration and achievement transformation assistance system according to claim 4, characterized in that, The steps for extracting abnormal fragment boundaries and generating a direction correction index based on the inverted suspect window are as follows: After revealing the location of the directional shift by reversing the suspect window, the time segment corresponding to the reversing suspect window is expanded segment by segment in each time series curve, so that the local changes are presented continuously in the absolute time context, and potential boundary clues of abnormal segments are identified accordingly. After identifying potential boundary clues, these clues are combined with the undulating structure of the curve to make the start and end points of the anomalous segments form continuous and comparable boundaries in the absolute time context, and to clearly separate the directional changes within the segments. After the boundary of the abnormal segment is clearly defined, the directional trend of the stable segment is extended from the outer extension of the cycle to the boundary of the abnormal segment, so that the position of directional offset is connected with the stable segment and a continuous directional line is formed. After the direction continuous lines are formed, the direction identifiers in the lines are matched with the original directions of the time-series curves, so that the direction offset segments can be adjusted under the mapping relationship, and a direction correction index covering the entire curve is generated.

7. The AI-driven clinical research data integration and achievement transformation assistance system according to claim 6, characterized in that, The generation of the direction correction index is limited to the node order of the continuous direction lines as the sole basis, so that the direction identifier of each node is connected and extended with the adjacent nodes, and the mapping relationship is continuously maintained in accordance with the absolute time context, so that the direction unification continues to advance continuously throughout the process.

8. The AI-driven clinical research data integration and achievement transformation assistance system according to claim 6, characterized in that, The steps for planning the staggered entry order of different data streams and generating a time-series scheduling table based on the direction correction index are as follows: After marking the direction adjustment nodes of each time-series curve with the direction correction index, the trend extension method of the data stream is classified according to the distribution of the direction adjustment nodes, so that the data stream forms an initial entry sequence in the absolute time context. Based on the initial entry sequence, a micro-delay is introduced before the data stream reaches the key nodes, and short silences are set at some entry nodes to create time intervals in the pace of the data stream's advancement within the absolute time context. After micro-delay and short silence form the basic rhythm, light playback is introduced to allow the data stream that has completed the directional adjustment to be appropriately pulled back in the absolute time context and to form a flexible connection with the preceding data stream. After the rhythm is choreographed, the specific moments of the data stream during micro-delay, short silence, and light playback are organized according to the absolute timeline, so that all rhythmic movements form a time schedule.

9. The AI-driven clinical research data integration and achievement transformation assistance system according to claim 8, characterized in that, After the light replay is completed, the entry rhythm forms a progressive sequence by continuing the micro-delay, so that each data stream maintains a stable distance before the key nodes of the absolute time context, and enters the convergence position in the order of the timing schedule after the short silence ends, so that all data streams maintain directional consistency and continuously follow the extended trajectory of the direction correction index during the rhythm connection process.

10. The AI-driven clinical research data integration and achievement transformation assistance system according to claim 8, characterized in that, The steps for performing alternating actions of mirror injection, phase foldback, and pre-unloading according to the timing schedule table to form an adaptive rhythm control cycle are as follows: After the timing schedule table presents a rhythm arrangement of micro-delay, short silence and light replay, the advancement position of each data stream is confirmed according to the timing schedule table, and mirror back-injection is performed before the data stream reaches its key node specified in the timing schedule table, so that the fluctuation segment after the direction adjustment is back-injected to the previous position in advance. After the mirror injection is completed, phase reversal is performed according to the timing schedule table, so that the data stream in the direction intersection interval rotates back along the direction of the phase starting point, so that the direction acceleration segment and the direction deceleration segment form a continuous transition in the absolute time context; After the phase reversal is completed, the pre-unloading is performed according to the timing schedule table, so that the fluctuation segment with the risk of directional deviation is removed from the propulsion path in advance, reserving space for the subsequent data flow. After the mirror injection, phase reversal and pre-unloading are completed in sequence, the above actions are alternately cycled according to the time window according to the timing schedule, so that the directional deviation is gradually absorbed and the true trend direction is solidified.

Images