Microfluidic-based oral drug precise delivery and efficacy evaluation system

Abstract

The present application relates to the technical field of microfluidic drug delivery, and discloses a microfluidic-based oral drug precise delivery and efficacy evaluation system.The system comprises four modules of signal acquisition and processing, instance recognition and feature extraction, path generation and efficacy evaluation.The signal acquisition and processing module continuously acquires multi-dimensional signals of oral drug delivery through a microfluidic sensing array, and integrates the signals into a unified data stream after real-time correction and structural standardization.The instance recognition module identifies the delivery instance, extracts the time identifier, spatial position descriptor and drug classification identifier.The path generation module divides intervals in combination with the time identifier, aligns the spatial information to a standard oral model, and generates a drug delivery path time sequence record.The efficacy evaluation module aggregates instances according to drug classification, calculates the target area drug distribution density, generates an efficacy evaluation data set, and realizes the precision of drug delivery monitoring and efficacy evaluation.

Description

Technical Field

[0001] This invention relates to the field of microfluidic drug delivery technology, specifically to a microfluidic-based system for precise delivery and efficacy evaluation of oral medications. Background Technology

[0002] In the treatment of oral diseases, the accuracy of drug delivery and the objectivity of efficacy evaluation remain core challenges in clinical practice. The oral cavity, as a complex physiological environment, is subject to multiple dynamic interference factors, including saliva flow, mucosal tissue heterogeneity, and masticatory movements. These factors directly lead to the difficulty of achieving precise drug concentration in the target lesion area using traditional oral drug delivery methods. Currently used oral drug formulations such as mouthwashes, patches, and gels generally suffer from uncontrollable drug release and dispersed action areas. Some drugs are rapidly washed away by saliva, which not only reduces drug utilization efficiency but may also cause adverse reactions such as local irritation due to drug residues in non-target areas.

[0003] Existing technologies also have significant limitations in monitoring and evaluating the efficacy of drug delivery. Traditional methods often rely on physicians' visual observations, patients' subjective feedback, or indirect methods such as in vitro sample testing to obtain efficacy information. These methods struggle to capture the dynamic changes during drug delivery, and the evaluation results are easily influenced by subjective factors and lack quantitative data support. Even when some studies have introduced sensing technology for oral signal acquisition, it is mostly single-dimensional sensing, failing to comprehensively reflect key information such as the time, space, and drug characteristics of drug delivery. Furthermore, the acquired signals are often distorted by factors such as electromagnetic interference and temperature fluctuations in the oral environment. The lack of effective correction and standardization methods during signal processing makes it difficult to integrate signals from different sources into a unified data stream, further hindering the systematic analysis of the drug delivery process.

[0004] Microfluidics technology, with its advantages of miniaturization, integration, and low consumption, has shown broad application prospects in the field of biomedical detection. Some studies have attempted to use it in the construction of drug delivery systems, but existing microfluidics technologies still have many shortcomings in oral drug applications. Most microfluidic systems focus only on the realization of drug delivery functions, without integrating delivery process monitoring and efficacy assessment into the design, resulting in a lack of a direct correlation between drug delivery effects and efficacy. Furthermore, the lack of adaptation to the special oral environment leads to poor stability of microfluidic sensing units within the oral cavity, making it difficult to guarantee the continuity and reliability of signal acquisition. At the data processing level, there is a lack of dedicated instance recognition and feature extraction algorithms for oral drug delivery scenarios, making it impossible to accurately extract key parameters of drug delivery from complex signals, thus failing to provide effective data input for efficacy assessment. These problems collectively result in a disconnect between drug delivery and efficacy assessment in oral drug therapy, failing to meet the clinical demand for precision treatment. Therefore, an integrated system capable of achieving precise drug delivery monitoring and objective efficacy assessment is needed. Summary of the Invention

[0005] The purpose of this invention is to provide a microfluidic-based system for precise delivery and efficacy evaluation of oral medications, in order to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides a microfluidic-based system for precise delivery and efficacy evaluation of oral medications, the system comprising:

[0007] The signal acquisition and processing module is used to continuously acquire multi-dimensional signals of the intraoral drug delivery process through a microfluidic sensor array, perform real-time correction and structural normalization of the signals, eliminate signal distortion, and integrate a unified data stream.

[0008] The instance identification and feature extraction module is used to identify drug delivery instances from a unified data stream and extract instance time identifiers, spatial location descriptors, and drug classification identifiers.

[0009] The path generation module is used to divide time intervals using instance time identifier sequences and align spatial location descriptors to a standard oral model to generate a drug delivery path time sequence record.

[0010] The efficacy evaluation module is used to aggregate delivery instances by category based on the drug delivery path time sequence record and drug classification identifier, calculate the drug distribution density in the target oral region, and generate an efficacy evaluation dataset.

[0011] Preferably, the step of continuously acquiring multi-dimensional signals of the intraoral drug delivery process via a microfluidic sensor array, performing real-time signal correction and structural normalization, eliminating signal distortion, and integrating a unified data stream includes:

[0012] The microfluidic sensor array generates multi-source signal streams, which are converted into digital format after signal conversion and the amplitude normalization algorithm is applied to adjust the signal level to make the signal dimensions consistent.

[0013] The system integrates dimensionally consistent signal streams, identifies format differences between different signal sources, converts them into standard data templates, and uses filtering algorithms to remove noise components, generating a clean signal set.

[0014] Based on a time reference point, the purification signal set is time-synchronized to compensate for time delay, multi-source data is fused, and a unified data stream is output.

[0015] Preferably, the application of the amplitude normalization algorithm to adjust the signal level to make the signal dimensions consistent includes:

[0016] Detect the peak and valley values ​​of multi-source signal streams, calculate the dynamic range, and use a scaling function to map the signals to a common amplitude range to achieve signal standardization;

[0017] Baseline correction is applied to the standardized signal to remove DC offset and ensure signal zero alignment.

[0018] Preferably, the step of structurally integrating the dimensionless signal stream, identifying format differences between different signal sources, converting them into standard data templates, and using filtering algorithms to remove noise components to generate a cleaned signal set includes:

[0019] Parse the header information of the dimensionally consistent signal stream, extract the format descriptor, match the field definition of the standard data template, and perform format conversion operations;

[0020] A low-pass filter is applied to suppress high-frequency noise, and an adaptive filter is used to eliminate environmental interference, resulting in a purified signal set.

[0021] Preferably, the step of identifying drug delivery instances from a unified data stream and extracting instance time identifiers, spatial location descriptors, and drug classification identifiers includes:

[0022] Scan the event feature patterns in the unified data stream to detect the instantaneous point of drug delivery and record the time stamp;

[0023] Location information is retrieved from data segments associated with event feature patterns, and three-dimensional coordinates are parsed using a spatial decoding algorithm to generate a spatial location descriptor;

[0024] The drug identification code is queried, the classification dictionary is consulted, the drug type is determined, and a drug classification identifier is assigned.

[0025] Preferably, the step of scanning the event feature patterns in the unified data stream, detecting the instantaneous point of drug delivery, and recording the time marker includes:

[0026] A pattern matching algorithm is used to identify mutation points in the signal, verify the consistency between mutation points and drug delivery logic, and mark timestamps; the timestamps are then sorted and deduplicated.

[0027] Preferably, the step of retrieving location information from the data segment associated with the event feature pattern, parsing the three-dimensional coordinates through a spatial decoding algorithm, and generating a spatial location descriptor includes:

[0028] Extract the raw readings from the position sensor and apply a coordinate solving algorithm to convert the readings into three-dimensional spatial points;

[0029] Cluster analysis is performed on points in three-dimensional space to remove outliers and generate robust spatial location descriptors.

[0030] Preferably, the step of dividing time intervals using instance time identifier sequences and aligning spatial location descriptors to a standard oral cavity model to generate a drug delivery path time sequence record includes:

[0031] Based on the distribution of instance time markers, the data is divided into equal-length blocks to form continuous time intervals;

[0032] For each time interval, the spatial location descriptor is transformed from the local coordinate system to the global coordinate system of the standard oral model, and a registration algorithm is used to achieve point-to-point mapping.

[0033] Organize the transformed location data in chronological order, add drug delivery attributes, and construct a time-series record of the drug delivery path.

[0034] Preferably, the step of transforming the spatial location descriptor from the local coordinate system to the global coordinate system of the standard oral cavity model, and using a registration algorithm to achieve point-to-point mapping, includes:

[0035] Load the coordinate system definition of the standard oral cavity model and calculate the transformation parameters between the local coordinate system and the global coordinate system;

[0036] The transformation parameters are applied to reproject each spatial location descriptor onto the global coordinate system, the projection accuracy is verified, and the error is adjusted.

[0037] Preferably, the step of aggregating delivery instances by category based on drug delivery path time-series records and drug classification identifiers, calculating drug distribution density within the target oral cavity region, and generating a efficacy evaluation dataset includes:

[0038] Based on drug classification identifiers, the time-series records of drug delivery routes are grouped into multiple category subsets;

[0039] For each category subset, the delivery frequency and drug amount within the target oral cavity region are statistically analyzed, and the density value is calculated by combining the region area.

[0040] The change in aggregation density over time generates a dataset for evaluating therapeutic efficacy.

[0041] Compared with the prior art, the beneficial effects of the present invention are:

[0042] The signal acquisition and processing module employs a microfluidic sensor array to construct a multi-dimensional acquisition system. Compared to traditional single-sensor methods, it can simultaneously capture multiple signals such as drug concentration, temperature, pH value, and drug spatial location within the oral cavity, achieving comprehensive perception of the drug delivery process. The module's built-in real-time correction mechanism dynamically corrects signal distortions caused by interference factors such as saliva flow and temperature fluctuations in the oral environment. Combined with structural standardization processing, it ensures consistency in format and accuracy for various signals. The resulting unified data stream provides a high-quality data foundation for subsequent analysis, avoiding the problem of integrating traditional multi-source signals.

[0043] The instance recognition and feature extraction module has developed a dedicated recognition algorithm for unified data streams, capable of accurately locating the start and end points of drug delivery instances. It extracts valuable information from complex data, providing core parameters for the quantitative analysis of the drug delivery process, including time signatures, spatial location descriptors, and drug classification identifiers. The time signature sequence clearly records the duration of drug action in the oral cavity, the spatial location descriptors accurately reflect the drug's distribution area, and the drug classification identifiers provide a basis for subsequent efficacy analysis based on drug characteristics. The extraction of these feature parameters transforms the drug delivery process from vague, empirical judgments into quantifiable, objective data records, solving the problem of missing key information in traditional methods.

[0044] The path generation module combines time stamps with spatial location information, decomposing the drug delivery process temporally through time interval division. Simultaneously, it aligns and calibrates the spatial location descriptors with a standard oral cavity model. The generated drug delivery path timeline record visually presents the dynamic migration process of the drug within the oral cavity. This visualized path record clearly shows key information such as whether the drug has reached the target lesion area and the time it spends in the target area, providing direct evidence for judging the accuracy of drug delivery. Compared to the shortcomings of traditional methods that cannot track the dynamic distribution of drugs, the timeline record generated by this module makes the drug delivery process traceable and analyzable, providing concrete directions for optimizing delivery strategies.

[0045] The efficacy assessment module aggregates delivery instances by drug category based on path time-series records and drug classification identifiers, and quantifies efficacy by calculating the drug distribution density in the target area. This module abandons traditional assessment methods that rely on subjective feelings and visual observation, using the actual distribution of drugs in the target area as the core indicator for efficacy judgment. The resulting efficacy evaluation dataset includes quantitative information such as drug delivery accuracy and drug enrichment in the target area. This assessment method objectively reflects the therapeutic effect of drugs and avoids assessment bias caused by subjective factors. Furthermore, the module's design ensures the comparability of efficacy data from different drugs and different patients, providing an objective reference for clinical drug selection. Attached Figure Description

[0046] Figure 1 This is a timing diagram of the microfluidic-based oral drug precision delivery and efficacy evaluation system described in this invention.

[0047] Figure 2 This is a flowchart of the signal acquisition and processing module;

[0048] Figure 3 A flowchart for signal structure integration and noise filtering;

[0049] Figure 4 A comparative analysis of the energy attenuation characteristics of adaptive filtering and the original signal;

[0050] Figure 5 This is a bar chart showing the total dose distribution of oral medications. Detailed Implementation

[0051] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] Please see Figure 1This invention provides a microfluidic-based system for precise oral drug delivery and efficacy evaluation. The system includes a signal acquisition and processing module, an instance identification and feature extraction module, a path generation module, and an efficacy evaluation module. The signal acquisition and processing module continuously acquires multi-dimensional signals from the intraoral drug delivery process using a microfluidic sensor array, performs real-time signal correction and structural normalization to eliminate signal distortion, and integrates a unified data stream. The instance identification and feature extraction module identifies drug delivery instances from the unified data stream and extracts instance time signatures, spatial location descriptors, and drug classification identifiers. The path generation module divides time intervals using instance time signature sequences and aligns the spatial location descriptors to a standard oral model to generate a time-series record of the drug delivery path. The efficacy evaluation module aggregates delivery instances by category based on the drug delivery path time-series record and drug classification identifiers, calculates the drug distribution density within the target oral region, and generates an efficacy evaluation dataset.

[0053] Example 1: See Figure 2 The microfluidic sensing array generates multi-source signal streams originating from miniature sensor nodes positioned at different anatomical locations within the oral cavity. These nodes are distributed in an array on the occlusal surfaces of teeth, the gingival margin sulcus, and the dorsum of the tongue mucosa. The microfluidic sensing array utilizes a flexible circuit board to achieve conformal fit with the oral cavity morphology. The multi-source signal streams include electrochemical impedance signals, optical reflection signals, and micropiezoresistive signals. The electrochemical impedance signal reflects changes in drug ion concentration, the optical reflection signal characterizes changes in mucosal color and wettability, and the micropiezoresistive signal monitors the mechanical forces involved in drug release. The signal conversion and processing stage employs a multi-channel analog-to-digital converter (ADC) to convert analog signals into digital format. The ADC has 16-bit resolution and a sampling rate of 1000 times per second. The digital signals are temporarily stored in a first-in-first-out (FIFO) buffer as fixed-length data packets. An amplitude normalization algorithm adjusts the amplitude of the multi-source signal streams. The algorithm detects the peak and trough values ​​of each signal channel in real time. Peak detection uses a sliding window local maximum identification method, and trough detection uses a local minimum scanning algorithm. The dynamic range calculation is based on the difference between the maximum and minimum signal values ​​within the window. The scaling function uses a linear mapping relationship to transform the original signal amplitude to a standardized range of zero to one. The slope and intercept of the linear mapping are calculated in real time based on the dynamic range to achieve consistency of signal dimensions. The standardized signal then enters the baseline correction stage, which is achieved through a digital high-pass filter. The cutoff frequency of the digital high-pass filter is set to 0.5 Hz, effectively removing the DC offset component in the signal and aligning the signal waveform to zero.

[0054] The dimensionally consistent signal stream enters the structure integration stage. The structure integration module parses the header information of each data packet, which includes the sensor node number, sampling timestamp, and signal quality identifier. The format descriptor extracts signal source features from the header information and includes a byte order marker, data precision identifier, and checksum field. The standard data template uses a unified field definition, including an absolute timestamp, signal channel number, three-dimensional signal value, and status flag. The format conversion operation performs field mapping and data reconstruction. Field mapping matches the field order of the standard data template according to the format descriptor, and data reconstruction handles signal interpolation alignment for different sampling rates. The filtering algorithm performs noise reduction on the structure-integrated signal. The low-pass filter uses a fourth-order Butterworth filter structure, and the cutoff frequency is dynamically configured according to the signal type: 10 Hz for electrochemical signals, 50 Hz for optical signals, and 100 Hz for pressure signals. The adaptive filter uses a multi-input single-output structure. The main input signal comes from the microfluidic sensor array, and the reference input signal comes from the environmental temperature and humidity sensor. The coefficient update of the adaptive filter uses a normalized least mean square algorithm with a step size parameter set to 0.01. The purified signal set undergoes time synchronization processing, based on a global clock source. This global clock source uses a temperature-compensated crystal oscillator to provide the time reference. Time delay compensation calculates the transmission delay of each signal channel, determined by measuring the propagation time of the signal from the sensor to the processing unit. Cubic spline interpolation is used to achieve time alignment of the multi-channel signals. Multi-source data fusion employs a weighted average algorithm, with weight coefficients dynamically allocated based on the signal-to-noise ratio. The output unified data stream is stored in a structured binary format.

[0055] The specific implementation of the amplitude normalization algorithm includes signal preprocessing steps. Signal preprocessing divides the continuous signal into fixed-length analysis frames, with each frame length set to 256 sampling points and a frame overlap rate of 50%. A Hanning window is applied for windowing, with the Hanning window coefficient multiplied by the signal samples within each analysis frame. Dynamic range calculation is performed within the windowed signal frame. Peak detection uses an amplitude threshold comparison method, with the amplitude threshold set to three times the root mean square value within the frame. Valley detection uses a negative threshold comparison, with the threshold setting symmetrical to that used for peak detection. The linear transformation formula of the scaling function uses floating-point arithmetic, and the transformation coefficients are updated in real-time based on the dynamic range within the frame. The DC offset estimation for baseline correction uses a moving average method, with the moving average window length set to 100 sampling points, and the DC offset subtracted from the original signal. After zero-alignment, the signal undergoes amplitude verification, checking whether the signal exceeds the normalized range. Signals exceeding the range are subject to soft clipping.

[0056] The data header parsing in the structural integration phase employs a layered parsing strategy. Layered parsing first identifies the packet start marker, then parses the metadata area, and finally extracts the payload. Format descriptor matching uses a pre-loaded configuration mapping table, which stores the data format specifications for different sensor models. Standard data template field definitions use byte alignment; fixed-length fields occupy consecutive bytes, while variable-length fields use length prefix encoding. Format conversion handles byte order conversion, selecting big-endian or little-endian mode based on the target platform characteristics. The low-pass filter in the filtering algorithm uses a direct type II digital filter structure, with filter coefficients calculated offline and stored in a lookup table. Environmental interference elimination by the adaptive filter is achieved through multi-stage filtering, with the main signal path and reference signal path undergoing the same preprocessing steps. Time synchronization of the purified signal set uses hardware timestamps generated by a dedicated timing circuit. The timestamp alignment algorithm uses a nearest-neighbor matching strategy with a matching tolerance of 1 millisecond. Data association for multi-source data fusion is based on the spatial location relationship of the sensors, prioritizing the fusion of data from adjacent sensors. The unified data stream output interface uses a high-speed serial bus, achieving a data transmission rate of 100 megabytes per second.

[0057] The microfluidic sensor array is physically implemented using a multilayer flexible circuit fabrication process. The circuit layers include signal acquisition electrodes, optical waveguides, and micro-piezoresistive films. Electrochemical impedance signal measurement employs a four-electrode method, which eliminates the influence of electrode polarization. The excitation signal frequency sweep range is from 100 Hz to 10 kHz. Optical reflection signal measurement uses a combination of dual-wavelength LEDs and a photodetector, with wavelengths selected at 660 nm and 880 nm, corresponding to the absorption peaks of oxyhemoglobin and deoxyhemoglobin, respectively. Micro-piezoresistive signal measurement utilizes a Wheatstone bridge structure, with the temperature compensation circuit integrated within the sensor node. The analog-to-digital converter for signal conversion and processing features a programmable gain amplifier with an automatically adjustable gain range from 1x to 1000x. The dynamic range calculation of the amplitude normalization algorithm incorporates a hysteresis mechanism to avoid frequent fluctuations in peak and valley values. The linear mapping of the scaling function supports a nonlinear piecewise mapping mode, which is used to handle saturated signals. The baseline-corrected digital high-pass filter adopts an infinite impulse response filter structure, and the infinite impulse response filter coefficients are designed using the bilinear transform method.

[0058] The hardware implementation of the structural integration module employs a Field-Programmable Gate Array (FPGA), which processes multiple signal streams in parallel. Format descriptor matching operations utilize Content-Addressable Memory (CORE), which stores format feature templates. Standard data templates are stored using Dual-Port Random Access Memory (DRAM), allowing simultaneous read and write access. The low-pass filter coefficients of the filtering algorithm are calculated using a frequency sampling method based on ideal filter response design. The reference signal input for the adaptive filter includes multiple environmental parameters, including temperature, humidity, and motion acceleration. Timestamp synchronization of the purified signal set employs a precision time protocol, achieving microsecond-level accuracy through network clock synchronization. Weighting coefficient calculation for multi-source data fusion is based on signal quality metrics, including signal-to-noise ratio (SNR) and signal stability. The unified data stream is stored using a time-series database structure, which supports fast time range lookups.

[0059] The microfluidic sensor array is powered by a miniature coin cell battery, with battery life extended through dynamic power management. The analog-to-digital converter (ADC) for signal conversion processing features automatic calibration, set to once per hour. Peak detection in the amplitude normalization algorithm incorporates digital filtering preprocessing to eliminate impulse noise interference. The scaling function's mapping interval is configurable and adjusts based on drug type. The moving average window length for baseline correction is adaptively variable, based on signal frequency component analysis. The structured header parsing supports multiple protocol formats, including custom binary protocols and standard medical device protocols. Data verification for format conversion operations employs cyclic redundancy check (CRC) to verify data integrity. The filtering algorithm's computational optimization utilizes a pipelined architecture, improving data processing throughput. The adaptive filter's coefficients are initialized using preset initial values ​​based on typical environmental conditions. The time synchronization clock source has a backup battery to maintain clock operation in case of mains power failure. The multi-source data fusion algorithm employs a distributed computing framework, executing in parallel on multi-core processors. The output of the unified data stream uses compression encoding, which reduces storage space usage.

[0060] Example 2: See Figure 3The structure integration operation of the dimensionally consistent signal stream initiates the header information parsing process. The header information, located at the beginning of each signal data packet, includes the signal source device identifier, data acquisition timestamp, signal sampling frequency parameters, and data format version number. The format descriptor extraction module scans specific fields in the header information. The format descriptor explicitly identifies the structural characteristics of the signal stream, including data encoding method, byte order, and checksum algorithm type. The standard data template uses predefined field definition specifications, which include fixed-length data fields and optional extended field areas. The format conversion operation performs a data mapping and reassembly process. Data mapping establishes a mapping table based on the correspondence between format descriptors and the standard data template, storing field names, data types, and byte offset information. The structured signal stream then enters the filtering stage. The low-pass filter is designed as a digital filter, with its cutoff frequency parameters dynamically configured according to the signal's frequency band characteristics. The adaptive filter employs a multi-channel filtering architecture. The main signal channel of the multi-channel filtering architecture receives the raw signal from the microfluidic sensor array, while the reference signal channel connects to the environmental monitoring sensor network. The purified signal set undergoes time-stamping processing, with the time stamps synchronized with a global clock source. The output signal is stored in a standardized data format. The process of identifying drug delivery instances from the unified data stream initiates an event feature pattern scanning mechanism. Event feature patterns are constructed based on multi-dimensional signal features, including signal amplitude abrupt change patterns, frequency spectrum characteristic changes, and waveform distortion features. The pattern matching algorithm employs a dynamic time warping calculation method, which compensates for scaling distortions on the signal's time axis. The abrupt change detection module monitors the first-order difference sequence of the signal. Threshold determination for the first-order difference sequence uses an adaptive threshold algorithm, adjusting the threshold size based on the signal's historical statistical characteristics. The consistency verification step between the abrupt change point and the drug delivery logic introduces multi-condition judgment logic, checking the signal amplitude change rate, event duration, and spatial location correlation. The timestamp recording function uses a high-precision clock chip, with its time synchronization signal sourced from a satellite navigation system. Timestamp sorting is implemented using a fast sorting algorithm. Timestamp deduplication is based on precise comparison of timestamp values, and the merging of duplicate timestamps uses a time window matching method.

[0061] The spatial location descriptor generation module retrieves location information data from data segments associated with event feature patterns. This location information data originates from the positioning sensor unit built into the microfluidic sensor array, which employs a combination of an ultrasonic ranging module and an inertial measurement unit. The spatial decoding algorithm performs a 3D coordinate calculation process based on a multi-sensor data fusion algorithm using an extended Kalman filter. 3D spatial point clustering analysis uses a density clustering algorithm, which can identify dense regions in the spatial point cloud. The outlier removal module applies statistical outlier detection technology, calculating the distance distribution characteristics between each data point and its neighbors. The robust spatial location descriptor outputs a spatial coordinate sequence with a confidence index derived from the residual analysis results of the coordinate calculation. The drug classification identifier determination process initiates a drug feature code lookup mechanism. Drug feature codes are derived from signal spectrum analysis feature vectors, which are obtained through Fast Fourier Transform. The classification dictionary stores the mapping relationship between drug types and signal features. The updating and maintenance of the classification dictionary are achieved through a machine learning model using a support vector machine classification algorithm. The drug classification identifier allocation process uses a unique identifier generation algorithm, which ensures that each drug delivery instance has a unique classification label. The instance record storage module packages the time identifier, spatial location descriptor, and drug classification identifier into a complete data structure, which is stored in binary format to optimize storage space.

[0062] The detailed implementation of event feature pattern scanning includes a signal buffer management mechanism, which adopts a circular buffer structure. The size of the circular buffer is dynamically adjusted according to the signal sampling rate. In the dynamic time warping implementation of the pattern matching algorithm, the warping path constraint limits the maximum ratio of time scaling, which is set to ±20%. The Sobel operator convolution calculation for edge detection uses a separable convolution optimization method to reduce computational complexity. The multi-condition check for consistency verification is integrated into the rule engine, which loads a configurable set of verification rules, including time constraints, spatial constraints, and signal feature constraints. The coordinate calculation in the spatial location descriptor generation uses sensor calibration parameters, which are obtained through a precise calibration process performed on a reference point with known coordinates. The density clustering algorithm parameters for cluster analysis are set based on oral anatomical features, and the minimum neighborhood radius parameter is determined based on the tooth spacing. The statistical method for outlier removal uses multi-dimensional outlier detection, which considers both spatial coordinate values ​​and signal strength values. The matching algorithm for drug feature code lookup calculates similarity using the cosine similarity metric. The mapping rule update mechanism for the classification dictionary incorporates incremental learning, allowing the system to progressively optimize classification accuracy during operation.

[0063] The positioning sensor unit of the microfluidic sensor array employs a miniaturized design, integrating the ultrasonic transducer and inertial measurement unit within a millimeter-scale package. The extended Kalman filter model of the spatial decoding algorithm includes state prediction equations and measurement update equations. The state prediction equations describe the sensor motion model, while the measurement update equations fuse multi-source observation data. Three-dimensional spatial point smoothing filtering utilizes a Kalman filter framework, where the state variables include position coordinates and velocity. The clustering analysis algorithm is selected based on point cloud density characteristics, employing grid clustering for high-density regions and hierarchical clustering for low-density regions. The outlier removal algorithm includes a noise point marking function, which marks suspicious data points as pending verification. The logic for assigning drug classification identifiers includes a conflict resolution mechanism to handle cases where feature code matching is uncertain. The database module for storing instance records uses a time-series database engine, which optimizes time-range query performance. The circular buffer implementation of the signal buffer management mechanism includes overflow protection, triggering a data transfer operation when the buffer is full. The pattern matching algorithm employs multi-threaded parallel processing for computational optimization, allocating signal segments to different processor cores. The threshold adaptive algorithm for mutation point detection uses a sliding window statistical method, which calculates the mean and variance of the signal within the window. The consistency verification rule engine supports dynamic rule loading, allowing for remote updates to the verification logic. Sensor data preprocessing for spatial coordinate calculation includes temperature compensation, correcting for the temperature effect on ultrasonic wave propagation speed. Cluster analysis outputs include cluster center coordinates and cluster radius parameters, used to define the drug delivery area. The signal processing for drug feature extraction includes wavelet transform analysis, providing time-frequency localization features. The classification dictionary is stored using a distributed database architecture, ensuring high availability for data access. The instance record data structure design includes data compression, reducing storage space usage and improving transmission efficiency.

[0064] See Figure 4 In the comparative analysis of the energy attenuation characteristics of adaptive filtering and the original signal, the plotting of the energy attenuation curve is based on multi-channel signal processing in a logarithmic coordinate system. Specifically, the energy curve of the original signal in the low-frequency band (10...) 0The low-pass filter curve exhibits a normalized energy value close to 1.0 (-10¹Hz), showing a fluctuating attenuation trend with increasing frequency, reflecting the fundamental spectral characteristics of the signal. An energy valley of 0.4 appears near 50Hz, indicating the design limitation of its fixed cutoff frequency. The adaptive filter curve shows a steeper attenuation slope in the high-frequency range (>30Hz) and multiple energy oscillation nodes in the 20-40Hz interval, characterizing the dynamic suppression mechanism of the adaptive algorithm against non-stationary noise, achieving noise suppression through real-time adjustment of filter parameters. During parameter configuration, the horizontal axis frequency uses a logarithmic scale (10¹⁰). 0 The normalized energy range on the vertical axis is set to 0 to 1.0, and the grid spacing is 0.2 units to enhance readability. This figure, calculated using the area under the curve integral, verifies that adaptive filtering improves high-frequency noise suppression efficiency by 15%-20% compared to traditional low-pass filtering while preserving effective signal bandwidth.

[0065] Example 3: Scanning event feature patterns in the unified data stream is implemented using a pattern matching algorithm. This algorithm is based on dynamic time warping, which can handle nonlinear deformations of signals along the time axis. The event feature patterns are defined by typical signal characteristics generated during drug delivery, including step changes in electrochemical signals, abrupt changes in reflectivity of optical signals, and pulse characteristics of pressure signals. The implementation of the pattern matching algorithm includes a signal preprocessing step, which divides the unified data stream into fixed-length analysis windows, with the window length set to 256 sampling points. Dynamic time warping calculates the minimum cumulative distance between the current window signal and the reference template signal, using a dynamic programming algorithm. A mutation point detection module monitors local feature changes in the signal, which are enhanced by calculating the first and second differences of the signal. Consistency verification between mutation points and drug delivery logic employs a multi-level verification strategy, including time correlation checks, spatial consistency verification, and signal feature matching degree evaluation.

[0066] The dynamic time-warped distance calculation in pattern matching algorithms uses the following mathematical expression:

[0067] in: Indicates the current cumulative distance. This represents the Euclidean distance between the current pair of points. and These represent the time indices of the test signal and the reference template signal, respectively. This recursive formula uses dynamic programming to find the optimal normalized path, and the cumulative distance value of the optimal normalized path is used for similarity judgment. When the cumulative distance is lower than a set threshold, an event feature pattern is considered to have been detected.

[0068] The mutation point detection employs a multi-scale analysis method, using the difference-of-Gaussian pyramid to detect local extrema of the signal. First-order difference calculations utilize the central difference formula, which provides accurate derivative estimates. Second-order difference calculations highlight changes in signal curvature, with curvature variation regions often corresponding to drug delivery events. An adaptive threshold algorithm adjusts the detection sensitivity based on the signal's background noise level, estimated using historical signal statistics. A temporal correlation check for consistency verification ensures mutation points occur within the expected drug delivery time window, while spatial consistency verification confirms the geometric plausibility of detection results from multiple sensors. The process of retrieving location information from data segments associated with event feature patterns initiates the spatial decoding process, which processes multi-source positioning data collected by the microfluidic sensor array. Raw readings from the position sensors include ultrasonic time-of-flight data, acceleration and angular velocity measurements from the inertial measurement unit, and orientation data from the magnetometer. The spatial decoding algorithm employs a sensor fusion framework, integrating multi-source observation data through an extended Kalman filter. Three-dimensional coordinate calculation is based on geometric positioning principles, utilizing the intersection positioning method of ultrasonic ranging. Cluster analysis employs density-based clustering algorithms, which can discover natural groupings in spatial data. Outlier removal utilizes statistical distribution-based outlier detection methods, which calculate the Mahalanobis distance between each data point and the cluster center.

[0069] The spatial decoding algorithm includes a coordinate transformation step, which converts measurements from the sensor's local coordinate system to the global coordinate system. The extended Kalman filter's state vector contains position, velocity, and attitude parameters; state vector prediction is based on numerical integration of inertial measurement data. The observation update phase fuses ultrasonic ranging data and magnetometer orientation data, correcting the state estimate by minimizing the difference between prediction and observation. Geometric positioning in 3D coordinate solution is optimized using the least squares method, which reduces the impact of measurement errors. The density clustering algorithm for cluster analysis employs a grid-based fast implementation, dividing the space into a uniform voxel grid. The Mahalanobis distance calculation for outlier removal considers the data's covariance structure, reflecting the correlation between dimensions. Preprocessing of the position sensor data includes a signal quality assessment step, checking the data's completeness and rationality. Ultrasonic time-of-flight data undergoes temperature compensation correction, adjusting the sound velocity parameters according to the ambient temperature. Integral errors in the inertial measurement unit data are corrected through zero-velocity updates, which reset velocity errors during the stationary phase. The calibration of magnetometer data eliminates both hard and soft iron interference; hard iron interference compensation is achieved through offset correction. Real-time implementation of the spatial decoding algorithm employs incremental computation, which reduces computational complexity and improves processing speed. Cluster analysis results are optimized using hierarchical clustering refinement, which merges over-segmented small clusters. The threshold for outlier removal is selected based on statistical significance testing, which uses a chi-square distribution to determine the critical value.

[0070] Drug feature extraction employs time-frequency analysis, which simultaneously captures the signal's time and frequency domain features. Fast Fourier Transform (FFT) calculates the signal's power spectral density, revealing its frequency distribution characteristics. Wavelet transform analysis provides a multi-resolution signal representation, suitable for analyzing non-stationary signals. Feature vector construction combines time-domain statistics and frequency-domain descriptors. Time-domain statistics include mean, variance, and skewness, while frequency-domain descriptors include spectral centroid and bandwidth parameters. The classification dictionary is constructed using supervised learning, which establishes a mapping between features and categories using labeled training data. A Support Vector Machine (SVM) classifier finds the optimal classification hyperplane, maximizing the margin between categories. The signal processing flow for drug feature extraction includes dimensionality reduction, achieved using Principal Component Analysis (PCA) to reduce feature dimensionality. A feature selection algorithm evaluates the contribution of each feature to classification based on the information gain criterion. The classification dictionary update mechanism supports online learning, allowing the system to progressively improve classification performance. The SVM classifier's kernel function is a radial basis function (RBF), which handles non-linearly separable problems. The confidence score of the classification result is calculated based on the distance from the point to the hyperplane, which reflects the degree of certainty in the classification.

[0071] The event detection sensitivity of the microfluidic sensing array is configurable, achieved by adjusting the detection threshold. The reference template library for the pattern matching algorithm supports dynamic updates, optimizing templates based on new observation data. Multi-scale analysis for mutation point detection employs a pyramid structure, providing signal representations at different resolutions. The rule set for consistency verification is customizable, adapting to various drug delivery scenarios. The coordinate reference system for the spatial decoding algorithm is defined based on a standard oral cavity model, providing an anatomical benchmark. Cluster analysis parameters are adaptively adjusted, automatically optimized based on point cloud density. The robustness of the outlier removal algorithm is enhanced through robust statistical methods. Standardization of drug feature codes eliminates the influence of dimensions, making different features comparable. Version management of the classification dictionary ensures consistency, recording the dictionary's modification history. Computational optimization for event feature pattern scanning utilizes parallel processing techniques, distributing signal segments to multiple processing units. The pattern matching algorithm is accelerated using an early termination strategy, terminating the search prematurely when the accumulated distance exceeds a threshold. The real-time performance of mutation point detection is ensured through a sliding window mechanism, which avoids redundant calculations of already processed data. Parallel checks for consistency verification reduce processing latency, as multiple conditions are verified simultaneously. The accuracy of the spatial decoding algorithm is improved through sensor calibration, which compensates for individual differences and installation errors. The computational efficiency of cluster analysis is enhanced through spatial indexing, which accelerates neighbor-to-neighbor search. Batch processing for outlier removal optimizes memory usage, handling multiple data points at once. The algorithm selection for drug feature extraction is based on signal characteristics, considering both computational complexity and feature discriminative power. Compressed storage of the classification dictionary reduces memory consumption, employing dictionary encoding techniques.

[0072] Example 4: Referring to Table 1, the process of dividing time intervals using the instance time stamp sequence begins with receiving the output data from the instance identification and feature extraction module. The instance time stamp sequence is a set of timestamps arranged chronologically, with each timestamp marking the precise moment a drug delivery instance was detected. Time interval division employs a fixed-duration equal-segment strategy, dividing the entire observation period into continuous and non-overlapping time windows. The duration of each time window is pre-set based on the drug delivery kinetics characteristics. The boundaries of the time windows are aligned to the integer second scale of the absolute time coordinates. This alignment ensures the regularity of the time intervals and avoids boundary ambiguity. The process of aligning the spatial location descriptor to the standard oral model involves coordinate system transformation. The standard oral model is a digital model containing three-dimensional geometric information of the teeth, gums, tongue, and oral mucosa. The local coordinate system is a servo coordinate system attached to the microfluidic sensor array, with its origin typically defined at the geometric center of the array. The global coordinate system is a world coordinate system fixed on the standard oral model, with its origin defined at the anatomical center of the oral cavity. Calculating transformation parameters requires solving for the rotation matrix and translation vector from the local coordinate system to the global coordinate system.

[0073] Table 1: Correspondence of Feature Points Used for Calculating Coordinate System Transformation Parameters

[0074]

[0075] The point cloud registration algorithm employs an iterative nearest-point method to achieve precise alignment between the local and global coordinate systems. This method iteratively calculates and finds the optimal rigid body transformation that minimizes the average distance between the two point sets. For each spatial location descriptor within a time interval, the calculated transformation parameters are applied to reproject its coordinate values ​​from the local coordinate system to the global coordinate system of the standard oral model. Projection accuracy is verified by calculating the reprojection error, which is the Euclidean distance between the projected point and the nearest point on the standard oral model. The drug delivery path time series records organize these coordinate-transformed spatial location points chronologically, with each record point appended with corresponding drug delivery instance attribute information. Preprocessing of the instance time stamp sequence includes timestamp normalization, which converts time information from different sources into milliseconds calculated from the start of the experiment. An equal-length block segmentation algorithm traverses the entire normalized timestamp sequence, dividing it into continuous time intervals at fixed intervals. Time interval boundary processing employs an up-rounding strategy to ensure that each timestamp uniquely belongs to a specific time interval. The spatial location descriptor's data structure includes three-dimensional coordinate values ​​and a direction vector, with the direction vector describing the local pose information during drug delivery. The global coordinate system definition of the standard oral cavity model follows medical imaging standards, with the coordinate system axes aligned with the anatomical orientation of the human body.

[0076] The specific steps for calculating transformation parameters begin with selecting corresponding point pairs, which are anatomical landmarks that can be clearly identified in both the local coordinate system and the global coordinate system of the standard oral model. Table 1 lists the correspondences of four typical feature points used to calculate transformation parameters. Based on these corresponding point pairs, the least squares method is used to solve for the optimal rotation matrix R and translation vector T, minimizing the objective function. The iterative process of the iterative nearest point method includes four steps: finding corresponding points, calculating the transformation, applying the transformation, and evaluating the error. This process is repeated until the average registration error is less than a set threshold or the maximum number of iterations is reached. Reprojection error evaluation calculates the distance between each projection point and the model surface. The distance calculation is accelerated using a spatial search algorithm. The construction of the drug delivery route time-series records adopts a time-series database model, which efficiently stores and retrieves route point data using time as an index. Each route point data record includes a timestamp field, X, Y, and Z coordinate fields in the global coordinate system, a drug classification identifier field, a delivery dose estimate field, and a confidence level field. The storage format of the time-series records adopts columnar storage to optimize query performance, facilitating time range queries and spatial region queries. The flexibility of time interval division is reflected in the configurable time window length, which allows adjustments based on different drug delivery rates. Shorter time windows are used for rapidly released drugs to capture details, while longer time windows are used for sustained-release drugs to observe overall trends. The alignment accuracy of the spatial location descriptor is affected by the installation stability of the microfluidic sensor array, which is ensured through the flexible fit design of the array substrate and the fixing device. Personalized adaptation of the standard oral cavity model involves acquiring the actual geometry of the user's oral cavity using 3D scanning technology, and then non-rigidly deforming the standard model to match individual characteristics. The robustness of the point cloud registration algorithm is enhanced by introducing a random sampling consensus algorithm, which effectively eliminates the influence of erroneous matching point pairs.

[0077] When the projection accuracy verification process detects errors exceeding the tolerance, an adjustment process is initiated. This process improves registration results by locally optimizing transformation parameters or introducing more corresponding point pairs. The data structure for the drug delivery path time-series records supports dynamic updates, allowing for the correction of historical paths after obtaining more accurate positioning data. The visualization of the time-series records overlays path points onto a 3D oral cavity model, using different colors to represent different drug types or time information. Local coordinate system calibration of the microfluidic sensor array is performed before each use. This calibration process requires the user to perform a series of standard actions to collect feature point data with known relative positions. The global coordinate system of the standard oral cavity model maintains a transformation parameter lookup table, which stores the transformation parameters corresponding to different individual users for rapid switching. The computational efficiency of the iterative nearest point method is optimized through a spatial partitioning tree data structure, which accelerates the nearest point search process. The compressed storage of the drug delivery path time-series records employs a lossy compression algorithm, which reduces storage space usage while preserving the path shape characteristics. The output of the entire route generation module is a structured drug delivery route time-series record file. This file is stored in an open standard format for easy data exchange with other healthcare information systems. Route data analysis functions include route length calculation, speed analysis, and regional coverage statistics.

[0078] Example 5: Category aggregation based on drug delivery path time-series records and drug classification identifiers begins with data grouping. The drug delivery path time-series records contain spatiotemporal information for numerous individual delivery instances, each with a unique drug classification identifier. The system reads the path time-series record database and hashes the drug classification identifiers, grouping instances with the same identifier into the same set. For example, identifier "A001" represents sodium fluoride anti-caries gel, and identifier "B002" represents chlorhexidine antibacterial mouthwash. The system creates two independent grouping containers to store all delivery instances of these two drug categories respectively. After grouping, multiple category subsets are formed, each containing all delivery locations and dosage data for a drug within a specific time period. The definition of the target oral cavity region is based on the anatomical divisions of a standard oral cavity model. The standard oral cavity model divides the oral cavity into multiple quantifiable regions, such as the maxillary anterior teeth region, the mandibular left molar region, and the dorsum of the tongue region. Each region has a clear geometric boundary in the model, defined by a set of vertices in a three-dimensional mesh. Calculating drug distribution density requires determining the surface area of ​​the target region. The system obtains an accurate surface area value by calculating the sum of the areas of all triangles that make up the grid of that region. For example, the surface area calculation for the left mandibular molar region will accumulate the areas of all the tiny triangles covering that region, thus obtaining an area value in square millimeters.

[0079] For each category subset, the system counts the delivery frequency within the target oral cavity region. Delivery frequency refers to the number of drug delivery events occurring per unit time. The system iterates through each instance in the category subset, checking if its spatial coordinates fall within the geometric boundaries of the target region. If an instance's coordinates are determined to be within the target region using ray casting, that instance is counted as a valid delivery. For example, if 15 valid deliveries of sodium fluoride anti-caries gel are detected in the dorsum of the tongue region during a ten-minute observation period, the delivery frequency is 1.5 times / minute. Drug quantity estimation is based on the dose information recorded for each delivery instance, derived from the quantification of drug concentration by a microfluidic sensor array. The system aggregates the dose values ​​of all valid delivery instances within the target region to obtain the total drug quantity. For example, if the dose values ​​of 15 delivery instances are 0.2 μL, 0.25 μL, ..., the sum is 3.75 μL. Density calculation combines the delivery frequency or drug quantity with the area of ​​the region; distribution density can be expressed as frequency density or mass density. The system performs a division operation, dividing the statistically obtained total delivery frequency or total drug quantity by the surface area of ​​the target region. For example, if the surface area of ​​the tongue dorsum is 500 square millimeters and the total amount of sodium fluoride anti-caries gel is 3.75 microliters, then the mass density is 0.0075 microliters / square millimeter. The change in density over time is captured using a sliding time window. The system divides the entire observation time axis into continuous, potentially overlapping time periods, and independently repeats the above statistical and calculation process for each time period. For example, using a one-minute window length and a 30-second sliding step, the density value within each minute window is calculated, thus forming a sequence of density changes over time.

[0080] Generating the efficacy evaluation dataset requires aggregating density values ​​for all categories, regions, and time points. The dataset uses a multidimensional table structure to store the data. Rows represent different time points, and columns represent different combinations of drug categories and oral regions. Each cell is filled with the corresponding density value. For example, at time point T1, the frequency density of sodium fluoride anti-caries gel in the maxillary anterior region is 0.002 times / minute / mm², and the mass density of chlorhexidine antibacterial mouthwash in the dorsum of the tongue region is 0.005 μL / mm². The efficacy evaluation dataset also includes metadata describing the dataset's generation parameters, time range, region definitions, and drug classification criteria. The hash grouping algorithm for category aggregation uses linear probing to resolve hash collisions, ensuring efficient grouping performance even with a large number of instances. Geometric boundary detection of target oral regions uses a point-in-polyhedron testing algorithm, which accurately determines the boundary by calculating the relationship between the point and the polygon face. Surface area calculation uses the Gaussian divergence theorem to transform surface integrals into linear integrals; the Gaussian divergence theorem improves computational efficiency, especially for complex surfaces. Dosage aggregation uses high-precision floating-point number accumulation, employing the Kahan summation algorithm to reduce rounding errors. Density calculation introduces normalization to prevent numerical overflow, scaling density values ​​to a reasonable range. Time series aggregation utilizes the database's aggregation query function, which leverages indexes to optimize query speed.

[0081] The query interface for drug delivery route time-series records supports filtering by time and spatial range, allowing the efficacy assessment module to quickly retrieve relevant instances. Partition information for the standard oral model is stored in a configuration file, which supports dynamic loading and modification to adapt to different clinically relevant areas. The density visualization module maps density values ​​to color codes, which are then rendered on the 3D oral model to form a heatmap. The efficacy evaluation dataset is exported in CSV and JSON formats, facilitating further processing by other data analysis tools. The system records detailed operation logs during processing, tracking each step of data transformation and potential sources of error. Category subsets are stored in memory using a linked list structure, facilitating dynamic insertion and deletion of instance data. Pre-calculation of bounding boxes for target regions accelerates spatial queries, enabling rapid initial screening through spatial indexing. Smoothing of the density time series employs a moving average filter, eliminating random fluctuations and highlighting trend changes. Version management of the efficacy evaluation dataset records parameters and results for each generation, supporting reproducibility and comparative analysis. The final dataset is transmitted to a central storage server via an encrypted channel, ensuring the security and privacy of patient data.

[0082] See Figure 5The cumulative drug dose data output by the efficacy assessment module is visualized through a stacked bar chart. Specifically, the horizontal axis of the chart labels six standard oral anatomical regions: the maxillary anterior region, maxillary posterior region, mandibular anterior region, mandibular posterior region, dorsum of the tongue region, and buccal mucosa region. The vertical axis represents the total dose (µL). Each region's bars are constructed by stacking the dose values ​​of four drug types in sequence: the bottom layer is compound chlorhexidine mouthwash, followed by sodium fluoride anti-caries gel, chlorhexidine antibacterial mouthwash, and the top layer is metronidazole oral ulcer gel. The color contrast clearly indicates the distribution differences corresponding to the drug classification identifiers. After categorical aggregation, the drug delivery path time sequence records are quantified, and the total dose reflects the cumulative drug amount of each delivery instance within each region. The data shows that the total dose in the buccal mucosa region and maxillary anterior region both exceed 100 µL, indicating that these regions are high-frequency targets for drug delivery, while the total dose in the dorsum of the tongue region is the lowest (50 µL), revealing spatial heterogeneity in distribution.

[0083] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A microfluidic-based system for precise delivery and efficacy evaluation of oral medications, characterized in that, The system includes: The signal acquisition and processing module is used to continuously acquire multi-dimensional signals of the intraoral drug delivery process through a microfluidic sensor array, perform real-time correction and structural normalization of the signals, eliminate signal distortion, and integrate a unified data stream. The instance identification and feature extraction module is used to identify drug delivery instances from a unified data stream, extracting instance time identifiers, spatial location descriptors, and drug classification identifiers, including: Scan event feature patterns in the unified data stream to detect the instantaneous point of drug delivery and record the time stamp; Location information is retrieved from data segments associated with event feature patterns, and three-dimensional coordinates are parsed using a spatial decoding algorithm to generate a spatial location descriptor; The system queries the drug feature code, compares it with the classification dictionary, determines the drug type, and assigns a drug classification identifier; the path generation module is used to divide the time interval using the instance time identifier sequence and align the spatial location descriptor to the standard oral model to generate a time sequence record of the drug delivery path. The efficacy evaluation module is used to aggregate delivery instances by category based on the drug delivery path time sequence record and drug classification identifier, calculate the drug distribution density in the target oral region, and generate an efficacy evaluation dataset. The event feature patterns in the unified data stream are scanned to detect the instantaneous point of drug delivery and record the time stamp, including: A pattern matching algorithm is used to identify mutation points in the signal, verify the consistency between mutation points and drug delivery logic, and mark timestamps; the timestamps are then sorted and deduplicated. The step of retrieving location information from data segments associated with event feature patterns, parsing three-dimensional coordinates using a spatial decoding algorithm, and generating a spatial location descriptor includes: Extract the raw readings from the position sensor and apply a coordinate calculation algorithm to convert the readings into three-dimensional spatial points; Cluster analysis is performed on points in three-dimensional space to remove outliers and generate robust spatial location descriptors.

2. The microfluidic-based oral medication precision delivery and efficacy evaluation system according to claim 1, characterized in that, The process involves continuously acquiring multi-dimensional signals from the intraoral drug delivery process using a microfluidic sensor array, performing real-time signal correction and structural normalization to eliminate signal distortion, and integrating a unified data stream, including: The microfluidic sensor array generates multi-source signal streams, which are converted into digital format after signal conversion and the amplitude normalization algorithm is applied to adjust the signal level to make the signal dimensions consistent. The system integrates dimensionally consistent signal streams, identifies format differences between different signal sources, converts them into standard data templates, and uses filtering algorithms to remove noise components, generating a clean signal set. Based on a time reference point, the purification signal set is time-synchronized to compensate for time delay, multi-source data is fused, and a unified data stream is output.

3. The microfluidic-based oral medication precision delivery and efficacy evaluation system according to claim 2, characterized in that, The application of the amplitude normalization algorithm to adjust the signal level to make the signal dimensions consistent includes: Detect the peak and valley values ​​of multi-source signal streams, calculate the dynamic range, and use a scaling function to map the signals to a common amplitude range to achieve signal standardization; Baseline correction is applied to the standardized signal to remove DC offset and ensure signal zero alignment.

4. The microfluidic-based oral drug precision delivery and efficacy evaluation system according to claim 3, characterized in that, The process of structurally integrating dimensionless signal streams, identifying format differences between different signal sources, converting them into standard data templates, and using filtering algorithms to remove noise components to generate a cleaned signal set includes: Parse the header information of the dimensionally consistent signal stream, extract the format descriptor, match the field definition of the standard data template, and perform format conversion operations; A low-pass filter is applied to suppress high-frequency noise, and an adaptive filter is used to eliminate environmental interference, resulting in a purified signal set.

5. The microfluidic-based oral medication precision delivery and efficacy evaluation system according to claim 4, characterized in that, The process of dividing time intervals using instance time identifier sequences and aligning spatial location descriptors to a standard oral cavity model to generate a drug delivery path time sequence record includes: Based on the distribution of instance time markers, the data is divided into equal-length blocks to form continuous time intervals; For each time interval, the spatial location descriptor is transformed from the local coordinate system to the global coordinate system of the standard oral model, and a registration algorithm is used to achieve point-to-point mapping. Organize the transformed location data in chronological order, add drug delivery attributes, and construct a time-series record of the drug delivery path.

6. The microfluidic-based oral medication precision delivery and efficacy evaluation system according to claim 5, characterized in that, The process of transforming the spatial location descriptor from the local coordinate system to the global coordinate system of the standard oral cavity model, using a registration algorithm to achieve point-to-point mapping, includes: Load the coordinate system definition of the standard oral cavity model and calculate the transformation parameters between the local coordinate system and the global coordinate system; The transformation parameters are applied to reproject each spatial location descriptor onto the global coordinate system, the projection accuracy is verified, and the error is adjusted.

7. The microfluidic-based oral medication precision delivery and efficacy evaluation system according to claim 6, characterized in that, The process involves aggregating delivery instances by category based on drug delivery path time-series records and drug classification identifiers, calculating the drug distribution density within the target oral cavity region, and generating a efficacy evaluation dataset, including: Based on drug classification identifiers, the time-series records of drug delivery routes are grouped into multiple category subsets; For each category subset, the delivery frequency and drug amount within the target oral cavity region are statistically analyzed, and the density value is calculated by combining the region area. The change in aggregation density over time generates a dataset for evaluating therapeutic efficacy.

Images