A method for determining the extent of effervescence from effervescent formulations using artificial intelligence

Abstract

ABSTRACT: A method for determining the extent of effervescence from effervescent formulations using artificial intelligence The present invention relates to a method for measuring the extent of effervescence produced by an effervescent formulation upon contact with an aqueous medium. The present invention provides a novel method and system for determining the extent of effervescence of pharmaceutical formulations using Artificial Intelligence (AI). The method utilizes a 3D Convolutional Neural Network (3D CNN) regression model trained on images of effervescence recorded at various time intervals. Images of the effervescence are captured using a digital camera in a controlled environment. These images are then processed by the trained 3D CNN model to predict the extent of effervescence, which can be correlated with the actual concentration of the formulation. This method provides a rapid, accurate, and cost-effective alternative to traditional methods for evaluating effervescent formulations.

Specification

Description:DESCRIPTION:
Field of the invention:
[0001] The present disclosure generally relates to the technical field of pharmaceutical technology, specifically to methods for evaluating the extent of effervescence produced by effervescent formulations. More particularly, it involves a method for measurement using a three-dimensional convolutional neural network (3D CNN) regression model that processes time-sequenced images of the effervescence reaction to predict the extent of effervescence.
Background of the invention:
[0002] Effervescent formulations are widely known and are generally solid preparations containing an acid-base couple that, when brought into contact with water, reacts with the production of carbon dioxide, resulting in bubbles and effervescence.

[0003] Current methods for measuring effervescence rely on chemical techniques such as gravimetry, volumetry, and gasometry. These methods typically involve the quantification of carbon dioxide evolved from effervescent formulations, which can be complex, costly, and challenging to perform with minimal equipment.

[0004] In the Chittick apparatus the carbon dioxide causes the displacement of water when acid and effervescent granules react together. This method support acid sources to carry out the reaction. But our invention is completely different from this invention as there is no use of acid for evolution of carbon dioxide, and we use distilled water which makes the process cheaper. Also, there is no measurement of displacement of water in the present invention (Dreimanis A. Quantitative gasometric determination of calcite and dolomite by using Chittick apparatus. J Sedimentary Res 1962; 32(3): 520-529, which is incorporated by reference herein in its entirety).

[0005] The modified Chittick apparatus also works on the same principle as the Chittick apparatus, uses the level of water displaced by carbon dioxide produced when the acid and effervescent granules react together (Arshad et al. Quantification of carbon dioxide released from effervescent granules as a predictor of formulation quality using modified Chittick apparatus. Trop J Pharm Res. 2021;18(3):449-58, which is incorporated by reference herein in its entirety).

[0006] Most of the current equipment measures the total volume of carbon dioxide gas liberated based on the chemical reactions using acidic solutions, but the present invention requires an aqueous dispersion of kaolin only, and the change in viscosity which can be determined easily by using a rotational viscometer and there is no requirement of devices such as camera.

[0007] For addressing all the above-mentioned problems, there is a need for a method which is simple and requires minimum investment for measuring the extent of effervescence produced by an effervescent formulation when come in contact with water.

[0008] A method employs the measurement of displacement of liquid by the evolved carbon dioxide, present invention measures the extent of effervescence by observing the change in viscosity (Arshad et al., Quantification of carbon dioxide released from effervescent granules as a predictor of formulation quality using modified Chittick apparatus, Tropical Journal of Pharmaceutical Research, 2019; 18 (3): 449-458, which is incorporated by reference herein in its entirety). This method is uses water level burette and acid‐dispensing burette.

[0009] Another study disclosed comparison of gravimetric, manometric, volumetric, gasometric and colorimetric methods for the determination of carbon dioxide (Amela et al., Methods for the determination of the carbon dioxide evolved from effervescent systems, Drug Development and Industrial Pharmacy, 19(9), 1019-1036 (1993), which is incorporated by reference herein in its entirety). The gravimetric, manometric, volumetric and colorimetric methods are entirely different methods compared to the teachings of the present invention. In the case of gasometric method, displacement of liquid is taken for the measurement of carbon dioxide determination.

[0010] A research work disclosed a method to monitor carbon dioxide pressure generation during the effervescent reaction in a plastic pressure vessel fitted with a pressure gauge (Anderson et al: Quantitative evaluation of pharmaceutical effervescent systems I: Design of Testing Apparatus", J Pharm Sci. 1982 Jan;71(1):3-6. doi: 10.1002/jps.2600710103, which is incorporated by reference herein in its entirety). Wherein, the quantification of an effervescent reaction is accomplished by measuring the dissolution time of the effervescent system and the pressure generated. It discloses two methods for the determination of liberated carbon dioxide: first by use of a pressure gauge and second by gravimetric method attributed to weight loss due to carbon dioxide loss.

[0011] A published review described the formulation and evaluation of effervescent tablets. "Patel Salim G et al: Formulation and evaluation of effervescent tablets: a review", Journal of Drug Delivery & Therapeutics. 2018; 8(6):296-303, which is incorporated by reference herein in its entirety). It described the formulation and evaluation of effervescent tablets. But it is not even a research work and is just a review on formulation and evaluation of effervescent tablets. Further, it just mentions measurement of carbon dioxide liberated by gravimetric method attributed to weight loss due to carbon dioxide loss.

[0012] An Indian patent application (No. 202441036191 dated 07/05/2024- An apparatus for evaluation of effervescent formulations and method of measurement thereof; which is incorporated by reference herein in its entirety) describes the measure the extent of effervescence by observing the movement of the ball inside the glass tube.

[0013] An Indian patent application (No. 202441068420 dated 10/09/2024- A method for the evaluation of effervescent formulations by measurement of viscosity changes in a kaolin dispersion; which is incorporated by reference herein in its entirety) describes the measure the extent of effervescence by measurement of viscosity changes in a kaolin dispersion.

[0014] However, NONE of the above-mentioned prior arts discloses the measurement of extent of effervescence of formulations by measurement using a three-dimensional convolutional neural network (3D CNN) regression model.

[0015] The present invention offers a valuable tool for pharmaceutical manufacturers to improve quality control, standardize production processes, and enhance the evaluation of effervescent formulations. The method contributes to better product quality, consistency, and patient safety.

Objectives of the invention:
[0016] The objective of the present invention is to provide a standard method for determining the extent of effervescence in pharmaceutical formulations using a 3D Convolutional Neural Network (3D CNN) regression model trained on images of effervescence.

Summary of the invention:
[0017] The present invention discloses a method for measuring the extent of effervescence produced by effervescent formulations using a 3D Convolutional Neural Network (3D CNN) regression model trained on images of effervescence. The following presents a simplified summary in order to provide a basic understanding of some aspects of the claimed subject matter. This summary is not an extensive overview. It is not intended to identify key/critical elements or to delineate the scope of the claimed subject matter. Its sole purpose is to present some concepts in a simplified form as a prelude to the more detailed description that is presented later.

[0018] In an aspect, the present disclosure relates to the use of artificial intelligence method for measuring the extent of effervescence produced by an effervescent formulation upon contact with an aqueous medium. Specifically, a 3D CNN model, to analyze images of the effervescence process. The 3D CNN is trained on images captured from a controlled experiment where effervescence is generated from standard effervescent samples.

[0019] In an aspect, the system includes a setup for capturing the effervescence, using a dark box with a light source pointed toward the container to enhance the visibility of the bubbles. A digital camera is used to capture images of the effervescence at different time intervals.

[0020] In an aspect, the captured images are processed using the trained 3D CNN regression model. The model analyzes features of the effervescence, such as the volume, frequency, or duration of the bubbles, to predict the extent of effervescence, which can be correlated with the concentration of the effervescence forming components of the formulation.

[0021] In an aspect, the method is validated by comparing the predicted data with actual data to generate a calibration curve. The high R² value of 0.987 obtained from this curve demonstrates the accuracy and reliability of the method.

[0022] In an aspect, the effervescent formulation containing a mixture of citric acid monohydrate and sodium bicarbonate is tested as standard.

[0023] In an aspect, the effervescent formulation comprises a combination of citric acid monohydrate and sodium bicarbonate, and is in the form of powders, granules, or tablets.

[0024] Further, objects and advantages of the present invention will be apparent from a study of the following portion of the specification, the claims, and the attached drawings.
Detailed description of drawings:
[0025] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate an embodiment of the invention, and, together with the description, explain the principles of the invention.

[0026] FIG. 1 illustrates a calibration curve plotted using the actual concentration values and the predicted concentration values obtained from the 3D CNN model. The R² value, a statistical measure of how well the predicted values fit the actual values, is displayed on the curve. A high R² value (e.g., 0.987) indicates a strong correlation between the predicted and actual values, further validating the model's accuracy and reliability.

[0027] FIG. 2 illustrates a visual representation of the experimental setup used to capture images of effervescence. The setup consists of a boiling tube containing water (200) placed inside a dark box (206). A 40-mesh basket may be used in the case of testing a tablet sample (202) A light source (204) is positioned to illuminate the effervescence, allowing for clear visualization and image capture. A digital camera (208) with a support stand (210) is used to record the effervescence at various time intervals. This figure illustrates the controlled environment and the tools used to acquire images for training and evaluating the 3D CNN model.

[0028] FIG. 3 presents a series of images extracted from video frames of a standard effervescent sample. These images showcase the varying degrees of effervescence over time. The images are arranged chronologically to illustrate the progression of effervescence from the initial stage to the final stage. This figure highlights the visual data used to train the 3D CNN model to recognize and quantify the extent of effervescence.

[0029] FIG. 4 provides a schematic representation of the different layers within the 3D CNN model architecture. Each layer is visually depicted with its corresponding type (e.g., convolutional, pooling, dense) and parameters (e.g., kernel size, number of filters). This figure helps visualize the flow of data through the model and the operations performed at each stage, aiding in understanding the complexity and functionality of the 3D CNN.

[0030] FIG. 5 illustrates the process of layer formation within the 3D CNN. It can show how input data is transformed as it passes through each layer, highlighting the feature extraction and abstraction capabilities of the network.

[0031] FIG. 6 presents a representation of the various parameters generated by the 3D CNN model during training and prediction. This figure helps illustrate the internal workings of the model and the quantitative aspects of its operation.

Detailed invention disclosure:
[0032] Various embodiments of the present invention will be described in reference to the accompanying drawings. Wherever possible, same or similar reference numerals are used in the drawings and the description to refer to the same or like parts or steps.

[0033] Embodiment of the present disclosure generally relates to the technical field of pharmaceutical technology, specifically to methods for evaluating the extent of effervescence produced by effervescent formulations. More particularly, it involves a method for measurement using a three-dimensional convolutional neural network (3D CNN) regression model that processes time-sequenced images of the effervescence reaction to predict the extent of effervescence.

[0034] Embodiment of the present disclosure has been made with a view towards solving the problem with the prior art described above, and it is an object of the present invention to provide an artificial intelligence-based method for measuring the extent of effervescence produced by effervescent formulations when come in contact with water.

[0035] In an embodiment, the present disclosure is about a method for measuring the extent of effervescence produced by effervescent formulations utilizing a 3D CNN model trained on images of effervescence, enabling it to analyze complex patterns and predict the extent of effervescence with high accuracy.
[0036] In an embodiment, the present disclosure relates to a method for determining the extent of effervescence of a standard effervescent formulation using a 3D Convolutional Neural Network (3D CNN) regression model trained on images of effervescence. This method involves preparing an effervescent formulation with a known concentration of ingredients like citric acid monohydrate and sodium bicarbonate. 115 mL of water is added to a boiling tube, which is placed in a dark box with a light source positioned to illuminate the effervescence produced when the formulation is added to the water. Images of the effervescence are captured at different time intervals using a digital camera. These images are then pre-processed by resizing them to a uniform dimension (216 pixels x 290 pixels) before being input into the trained 3D CNN model. The model predicts the extent of effervescence, which can then be compared to the known concentration of the formulation.
[0037] In an embodiment, a method for evaluating the efficiency of the 3D CNN model using effervescent formulation such as powders, granules, or tablets is provided. This method involves preparing effervescent formulation containing known amounts of ingredients like citric acid monohydrate and sodium bicarbonate. The formulation is dropped into a glass tube containing water. Tablet formulations are introduced after placing in a 40-mesh basket to avoid possible floating. A video of the effervescence is recorded, and images are extracted from the video at different time intervals. These images are pre-processed and analyzed using the trained 3D CNN model. The predicted extent of effervescence is then compared with actual data to determine the extent of effervescence.
[0038] In an embodiment, the 3D CNN regression model architecture and parameters are provided. The model includes convolutional layers, pooling layers, a flatten layer, and dense layers. The convolutional layers apply 3D convolutions to the input, preserving spatial dimensions. MaxPooling3D layers reduce the spatial dimensions of the feature maps. The flatten layer converts the 3D feature maps into a 1D vector for the fully connected layers. The dense layers perform regression to predict the concentration.
[0039] In an embodiment, the trained 3D CNN model is used for real-time prediction on new samples. Images of effervescence from a new sample are captured using the experimental setup described above. The images are pre-processed (resizing and normalization), and then input into the trained 3D CNN model. The model predicts the extent of effervescence for the new sample.
[0040] In an embodiment, the images are pre-processed by resizing them to a uniform dimension of 216 pixels x 290 pixels before being input into the trained 3D CNN regression model.
[0041] In an embodiment, the images are converted to grayscale prior to analysis to reduce computational complexity and enhance feature extraction relevant to effervescence detection.
[0042] In an embodiment, the 3D CNN regression model has an input layer with a shape of (6, 216, 290, 1), corresponding to a sequence of six grayscale images with a resolution of 216 by 290 pixels.
[0043] According to another exemplary embodiment of the invention, FIG. 1 refers to a calibration curve plotted using the actual concentration values and the predicted concentration values obtained from the 3D CNN model. The R² value, a statistical measure of how well the predicted values fit the actual values, is displayed on the curve. A high R² value (0.987) indicates a strong correlation between the predicted and actual values, further validating the model's accuracy and reliability.

[0044] According to another exemplary embodiment of the invention, FIG. 2 refers to an illustration of the experimental setup used to capture images of effervescence. The setup consists of a boiling tube containing water placed inside a dark box. A light source is positioned to illuminate the effervescence, allowing for clear visualization and image capture. A digital camera is used to record the effervescence at various time intervals. This figure illustrates the controlled environment and the tools used to acquire images for training and evaluating the 3D CNN model.

[0045] According to another exemplary embodiment of the invention, FIG. 3 refers to a series of images extracted from video frames of a standard effervescent sample. These images showcase the varying degrees of effervescence over time. The images are arranged chronologically to illustrate the progression of effervescence from the initial stage to the final stage. This figure highlights the visual data used to train the 3D CNN model to recognize and quantify the extent of effervescence.

[0046] According to another exemplary embodiment of the invention, FIG. 4 refers to a schematic representation of the different layers within the 3D CNN model architecture. Each layer is visually depicted with its corresponding type (e.g., convolutional, pooling, dense) and parameters (e.g., kernel size, number of filters). This figure helps visualize the flow of data through the model and the operations performed at each stage, aiding in understanding the complexity and functionality of the 3D CNN.

[0047] According to another exemplary embodiment of the invention, FIG. 5 refers to an illustration the process of layer formation within the 3D CNN. It can show how input data is transformed as it passes through each layer, highlighting the feature extraction and abstraction capabilities of the network.

[0048] According to another exemplary embodiment of the invention, FIG. 6 refers to a representation of the various parameters generated by the 3D CNN model during training and prediction. This figure helps illustrate the internal workings of the model and the quantitative aspects of its operation.
[0049] According to another exemplary embodiment of the invention, FIG. 7 refers to a series of images extracted from video frames of a sample tablet formulation, in triplicate. These images showcase the varying degrees of effervescence over time. The images are arranged chronologically to illustrate the progression of effervescence from the initial stage to the final stage. This figure highlights the visual data used to predict the extent of effervescence from the tablet formulation.

[0050] In an embodiment, the effervescent formulation contains a mixture of citric acid monohydrate and sodium bicarbonate.

[0051] In an embodiment, the effervescent formulation comprises a combination of citric acid monohydrate and sodium bicarbonate, and is in the form of powders, granules, or tablets.

[0052] Numerous advantages of the present disclosure may be apparent from the discussion above.

[0053] The invention is particularly useful for effervescent tablets, where the extent of effervescence is a critical quality attribute. The AI-powered method provides a rapid and objective assessment of the effervescence process, leading to improved product quality and consistency.

[0054] From the present disclosures, the invention of using a 3D CNN for determining the extent of effervescence in pharmaceutical formulations presents a non-obvious and inventive solution to a previously unaddressed problem in the field. The novelty lies in the application of 3D CNN technology to a new domain, the transformation of a qualitative observation into a quantitative measurement, and the achievement of high accuracy and efficiency. These factors contribute to the inventive step and the potential patentability of the invention.

[0055] From the present disclosures, it may be apparent that the method for measuring the extent of effervescence can serve as a quality control or evaluation parameter for pharmaceutical effervescent formulations, demonstrating industrial applicability.

[0056] It will readily be apparent that numerous modifications and alterations can be made to the method and system described in the foregoing examples without departing from the principles underlying the invention, and all such modifications and alterations are intended to be embraced by this application.

[0057] The following will illustrate in detail specific embodiments of the present invention:
Example 1: The extent of effervescence of a standard effervescent formulation can be determined. An effervescent formulation containing a known concentration of citric acid monohydrate and sodium bicarbonate is prepared. 115 mL of water is added to a boiling tube and placed within a dark box. A light source is positioned to illuminate the effervescence. The effervescent formulation is added to the water. Images of the effervescence are captured at different time intervals using a digital camera. The images are pre-processed by resizing them to a uniform dimension (216 pixels x 290 pixels). The pre-processed images are then input into the trained 3D CNN model. The 3D CNN model predicts the extent of effervescence, which is then compared to the known concentration of the formulation.

Example 2: The trained 3D CNN model can be used on new samples for real-time prediction. Images of effervescence from a new, unknown sample are captured using the established experimental setup. The images are pre-processed (resizing and normalization). The pre-processed images are fed into the trained 3D CNN model, which then predicts the extent of effervescence for the new sample. This example highlights the practical application of the invention for analyzing and evaluating new effervescent formulations.

Example 3: The efficiency of the method was determined using effervescent tablets. Effervescent tablets containing 1.1722 g of a mixture of citric acid monohydrate and sodium bicarbonate in a weight ratio of 1.3:1 were prepared. The tablet was placed in a 40-mesh basket (to avoid floating) and dropped into a glass tube containing water. A video of the effervescence process was recorded, and images were extracted from the video at different time intervals. The experiment was conducted in triplicate. These images (FIG. 7) were pre-processed and analyzed using the trained 3D CNN regression model. The extent of effervescence using the 3D CNN regression model was predicted and found to be 103.75%, compared to the theoretical effervescence produced by the same quantities of concentration of citric acid monohydrate and sodium bicarbonate.
, Claims:CLAIMS:
We Claim:
1. A method for determining the extent of effervescence of a pharmaceutical formulation, comprising:
providing a pharmaceutical formulation capable of effervescence;
placing the formulation in a container with a liquid to initiate effervescence;
placing the container in a dark box with a light source directed towards the container;
capturing images of the effervescence at different time intervals using a digital camera;
pre-processing the captured images to resize and convert them into a suitable format for analysis; and
processing the pre-processed images using a trained 3D Convolutional Neural Network (3D CNN) regression model to predict the extent of effervescence.
2. The method of claim 1, wherein the pharmaceutical formulation is effervescent powders, granules, or tablets.
3. The method of claim 1, wherein the pre-processed images are provided in either PNG or JPEG format.
4. The method of claim 1, wherein the 3D CNN model is trained on images of effervescence from standard samples.
5. The method of claim 1, wherein the images are pre-processed by resizing them to a uniform dimension of 216 pixels x 290 pixels before being input into the trained 3D CNN regression model.
6. The method of claim 1, wherein the 3D CNN model comprises:
An input layer configured to receive a sequence of pre-processed images representing the effervescence at different time intervals;
Multiple 3D convolutional layers with padding to extract spatial-temporal features from the input images;
Max pooling layers to reduce the dimensionality of the extracted features;
A flatten layer to convert the multi-dimensional feature maps into a one-dimensional vector;
Multiple dense layers for performing regression to predict the extent of effervescence.
7. The method of claim 6, wherein the images are converted to grayscale prior to analysis.
8. The method of claim 6, wherein the 3D CNN regression model has an input layer with a shape of (6, 216, 290, 1), corresponding to a sequence of six grayscale images with a resolution of 216 by 290 pixels.
9. The method of claim 6, wherein the 3D convolutional layers utilize the 'ReLU' activation function and the 'same' padding scheme.
10. The method of claim 6, wherein the dense layers utilize the 'ReLU' activation function except for the output layer which uses a linear activation function.

Documents