A SYSTEM FOR DETECTION AND CLASSIFICATION OF HEART DISEASES

Abstract

ABSTRACT A SYSTEM FOR DETECTION AND CLASSIFICATION OF HEART DISEASES The system(100) for detection and classification of heart diseases comprises data collection module(110), pre-processing module(130), cardiac analysis module (140), and report generation module (150). The data collection module(110) captures infant health data from data capturing sources(110a) and stores the data in a repository(120). The pre-processing module(130) applies standardization and normalization techniques to this data and generates pre-processed infant health data. The cardiac analysis module(140) employs a hybrid neural network model, consisting of first neural network module(140A), second neural network module(140B), and classification module(140C). The first neural network module(140A) processes the pre-processed data to generate a feature map. The second neural network module(140B) takes this feature map to compute a first-order derivative. The classification module(140C) featuring fully connected layers, outputs probability values across heart disease classes: no disease, low risk, mild disease, and advanced disease. The report generation module(150) creates detailed cardiac-health-report, selecting the class with highest probability as output.

Specification

Description:FIELD OF DISCLOSURE
The disclosure relates to diagnostics and healthcare technology, particularly to the detection of cardiac diseases in infants.
BACKGROUND
The background information herein below relates to the disclosure but is not necessarily prior art.
Early diagnosis of cardiac diseases in newborns is essential for reducing mortality rates and improving health outcomes, yet current diagnostic methods often lack the precision needed to interpret complex ECG signals. This issue is further exacerbated in rural areas, where access to advanced technologies is limited, leading to significant health disparities. Existing systems frequently overlook subtle indicators of cardiac distress, particularly in vulnerable populations.
Additionally, rural healthcare systems frequently lack access to advanced diagnostic technologies, exacerbating health disparities and increasing the risk of complications from undetected cardiac issues. The existing technologies often overlook subtle signs of distress, which can lead to severe consequences for infants who require timely intervention. This underscores a significant gap in pediatric cardiac care.
Therefore, there is a need for a system for detection and classification of heart diseases that alleviates the aforementioned drawbacks.
OBJECTS
Some of the objects of the disclosure, which at least one embodiment herein satisfies, are as follows:
It is an object of the disclosure to ameliorate one or more problems of the prior art or to at least provide a useful alternative.
An object of the disclosure is to develop a system that can enable early identification of cardiac conditions in infants, thereby allowing timely medical intervention and improving health outcomes.
Another object of the disclosure is to provide a system for detection and classification of heart diseases.
Still, another object of the disclosure is to enable precision and continuous monitoring of the ECG for a patient particularly newborns and infants in order to detect subtle changes in the ECG waveform over time.
Another object of the disclosure is to surpass the performance of traditional ECG analysis methods by reducing false positives and negatives, thus increasing the reliability of cardiac disease detection.
Still, another object of the disclosure is to provide advanced and affordable cardiac diagnostic facilities in rural healthcare settings, addressing health disparities by making advanced diagnostic tools accessible to underserved populations.
Yet another object of the disclosure is to provide solutions and develop strategies for mitigating the impact of limited data availability and variability in ECG quality.
Still, another object of the disclosure is to create a centralized database of diverse ECG datasets, ensuring the training of the system on a wide range of cases to enhance its accuracy and efficacy.
Yet another object of the disclosure is to develop a system to uncover insights into cardiac disease progression in infants and as an educational resource for training medical professionals in AI-based diagnostics.
Still, another object of the disclosure is to enable integration with telemedicine platforms, allowing for real-time ECG monitoring of infants, thereby facilitating timely alerts and consultations between healthcare providers and families.
Other objects and advantages of the disclosure will be more apparent from the following description, which is not intended to limit the scope of the disclosure.
SUMMARY
The present disclosure envisages a system for detection and classification of heart diseases. The system is configured to comprise a data collection module, a pre-processing module, a cardiac analysis module, a report generation module, and a weight optimization module.
The data collection module is configured to capture infant health data through one or more data capturing sources including clinical electrocardiogram (ECG) devices, wearable devices, telemedicine platforms, and electronic health records, and is further configured to store said infant health data in a repository.
The pre-processing module is configured to receive said infant health data from said data collection module or said repository, and further configured to implement a set of pre-processing techniques on said infant health data to standardize and normalize said infant health data so as to generate pre-processed infant health data.
The cardiac analysis module is configured to implement a hybrid neural network model having a first neural network model and a second neural network model.
The first neural network module is configured to receive said preprocessed infant health data from the pre-processing module, and further configured to implement the first neural network model on said preprocessed infant health data to generate a feature map.
The second neural network module is configured to receive the feature map generated by the first neural network model as input data, and is further configured to implement the second neural network model on the input data to calculate a first-order derivative of the input data.
The cardiac analysis module further comprises a classification module. The classification module comprises fully connected layers of the hybrid neural network model to output probability values distributed across four predefined heart disease classes, including no disease, low risk, mild disease, and advanced disease.
The report generation module is configured to cooperate with said cardiac analysis module to generate a detailed cardiac health report of the infant, in which a class having the highest probability value being selected as a final output.
In an embodiment of the disclosure, the Infant health data comprises data indicating cardiac health status of an infant including heart rate, electrocardiogram waveform, heart rhythm, blood pressure, oxygen saturation, respiratory rate, weight and growth parameters, family history, and physical examination findings.
In an embodiment of the disclosure, the system further comprises a weight optimization module configured to optimize the weights of the hybrid neural network model by means of a modified Bald Eagle Optimizer (MBEO) to avoid local optima.
In an embodiment of the disclosure, the weight optimization module optimizes the weights by:
P_(new,i )= P_best+ α*w_i* (P_mean-P_i)………………..[1]
w_i=exp⁡(P_i /P_best )………………………….[2]
〖 P〗_new = P_best+ α*exp⁡(P_i /P_best )*(P_mean-P_i)………………..[3]
where Pbest is the best solution, Pi is the current position, Pmean is the mean of all previous positions, and α is a parameter for a position with a value ranging from 1.5 to 2, ensuring that the modified Bald Eagle Optimizer (MBEO) escapes the local optima.
In an embodiment of the disclosure, the first neural network model is a convolution neural network (CNN) comprising:
a first convolutional layer with 32 filters applied to the pre-processed data;
a first max pooling layer applied to an output data of the first convolutional layer;
a second convolutional layer with 64 filters applied to an output of the first max pooling layer;
a second max pooling layer applied to an output of the second convolutional layer; and
a 32-bit embedding layer applied to an output of the second max pooling layer so as to generate the feature map.
The system (100) as claimed in claim 1, wherein the second neural network model is a Long Short Term Memory network (LSTM) comprising:
an input layer to receive the feature map generated by the CNN as the input data;
a first dense layer consisting of 64 neurons to apply a first-order derivative operation on the input data;
a second dense layer consisting of 4 neurons to process an output of the first dense layer; and
fully connected layers to calculate probability values distributed across the four predefined heart disease classes based on the output of the second dense layer, the four predefined heart disease classes including no disease, low risk, mild disease, and advanced disease;
wherein the class with the highest probability is selected as the final output.
In an embodiment of the disclosure, the set of pre-processing techniques comprises the following steps:
receiving the infant health data from the repository;
removing high-frequency noise from the infant health data by using a low pass filter with a cutoff frequency of approximately 50 Hz;
eliminating low-frequency baseline drift in the infant health data by using a high pass filter with a cutoff frequency of 0.5 Hz;
generating filtered infant health data and standardizing said infant health data by z-score normalization;
calculating the mean and standard deviation of said filtered infant health data;
subtracting the mean from each data point in said filtered infant health data;
dividing the result by the standard deviation and standardizing the data values, resulting in data with a mean of 0 and a standard deviation of 1; and
generating pre-processed infant health data and feeding said pre-processed infant health data to the cardiac analysis module.
In an embodiment of the disclosure, the hybrid neural network model is configured to implement a combination of computational neural networks (CNN) and Long Short Term Memory networks (LSTM).
In an embodiment of the disclosure, the weight optimization module is configured to update network weights in the hybrid neural network model implemented by said cardiac analysis module by means of metaheuristic optimization techniques.
In an embodiment of the disclosure, the hybrid neural network model implemented by said cardiac analysis module is further configured to detect and classify heart diseases for individuals with diverse physiological and health characteristics including but not limited to varying age-groups, heterogenous populations, and individuals with specific health conditions.
In an embodiment of the disclosure, the hybrid neural network model implemented by said cardiac analysis module is further configured to be able to detect and classify other diseases including but not limited to respiratory diseases, renal diseases, metabolic disorders, dermatological conditions, and psychological disorders.
In an embodiment of the disclosure, the system is further configured to operate as a cloud-based system and as a local system.
The present disclosure also envisages a method for detection and classification of heart diseases, the method comprises the following steps:
capturing, by a data collection module, infant health data through one or more data capturing sources, including clinical electrocardiogram (ECG) devices, wearable devices, telemedicine platforms, and electronic health records, and storing the infant health data in a repository;
receiving, by a pre-processing module, said infant health data from the data capturing sources or the repository;
implementing, by the pre-processing module, a set of pre-processing techniques on the infant health data to standardize and normalize the infant health data, thereby generating pre-processed infant health data;
implementing, by a cardiac analysis module, a hybrid neural network model for cardiac analysis, wherein the hybrid neural network model includes a first neural network model and a second neural network model;
receiving, by the cardiac analysis module, the pre-processed infant health data and applying the first neural network model to generate a feature map;
receiving, by the cardiac analysis module, the feature map as input data and applying the second neural network model to calculate a first-order derivative of the input data;
outputting, by fully connected layers of the hybrid neural network model implemented by the cardiac analysis module, probability values distributed across four predefined heart disease classes, including no disease, low risk, mild disease, and advanced disease;
selecting, by the cardiac analysis module, the class with the highest probability value as a final classification output; and
generating, by a report generation module, a detailed cardiac health report of the infant based on the analysis and classification results, wherein the final output is included in the report.
BRIEF DESCRIPTION OF THE ACCOMPANYING DRAWING
A system detection and classification of heart diseases of the disclosure will now be described with the help of the accompanying drawings, in which:
Figure 1 illustrates a block diagram representing a system for detection and classification of heart diseases in accordance with one embodiment of the disclosure;
Figure 2 illustrates a flowchart for a system for detection and classification of heart diseases in accordance with an embodiment of the disclosure; and
Figures 3A-3B illustrate a method for detection and classification of heart diseases in accordance with one embodiment of the disclosure.
LIST OF REFERENCE NUMERALS
100 System for detection and classification of heart diseases
110 Data collection module
110a Data capturing sources
120 Repository
130 Preprocessing module
140 Cardiac analysis module
140A First neural network module
140B Second neural network module
140C Classification module
150 Report generation module
160 Weight optimization module
DETAILED DESCRIPTION
Embodiments, of the disclosure, will now be described with reference to the accompanying drawing.
Embodiments are provided so as to thoroughly and fully convey the scope of the disclosure to the person skilled in the art. Numerous details are set forth, relating to specific components, and methods, to provide a complete understanding of embodiments of the disclosure. It will be apparent to the person skilled in the art that the details provided in the embodiments should not be construed to limit the scope of the disclosure. In some embodiments, well-known processes, well-known apparatus structures, and well-known techniques are not described in detail.
The terminology used, in the disclosure, is only for the purpose of explaining a particular embodiment and such terminology shall not be considered to limit the scope of the disclosure. As used in the disclosure, the forms "a," "an," and "the" may be intended to include the plural forms as well, unless the context clearly suggests otherwise. The terms "including," and "having," are open-ended transitional phrases and therefore specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not forbid the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof. The particular order of steps disclosed in the method and process of the disclosure is not to be construed as necessarily requiring their performance as described or illustrated. It is also to be understood that additional or alternative steps may be employed.
When an element is referred to as being "engaged to," "connected to," or "coupled to" another element, it may be directly engaged, connected, or coupled to the other element. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed elements.
Timely detection of cardiac diseases in newborns is crucial for lowering mortality rates and enhancing health outcomes; however, current diagnostic methods often lack the necessary precision to effectively analyze complex ECG signals. This problem is especially pronounced in rural areas, where limited access to advanced diagnostic technologies contributes to significant health disparities. Existing systems frequently miss subtle indicators of cardiac distress, particularly in vulnerable populations, increasing the risk of complications from undiagnosed conditions
To address the issues of the existing systems and methods, the present disclosure envisages a system (hereinafter referred to as "system 100") for detection and classification of heart diseases and a method for detection and classification of heart diseases (hereinafter referred to as "method 300"). The system (100) will now be described with reference to Figure 1 and Figure 2 and the method (300) will be described with reference to Figure 3A and Figure 3B.
Figure 1 illustrates a block diagram representing a system for detection and classification of heart diseases in accordance with one embodiment of the disclosure. The system (100) includes a data collection module (110) that gathers data from various sources such as clinical electrocardiogram (ECG) devices, wearable devices, telemedicine platforms, and electronic health records, subsequently storing the information in a repository (120). The pre-processing module (130) receives this infant health data and applies a series of standardization and normalization techniques to produce pre-processed infant health data. This data is then fed into the cardiac analysis module (140), which employs a hybrid neural network model consisting of a first neural network module (140A) that generates a feature map from the pre-processed data, and a second neural network module (140B) that computes the first-order derivative of the feature map. The classification module (140C) of this hybrid model outputs probability values across four predefined heart disease classes: no disease, low risk, mild disease, and advanced disease. Additionally, a report generation module (150) collaborates with the cardiac analysis module (140) to create a comprehensive cardiac health report, selecting the class with the highest probability value as the final output. To enhance the accuracy of the neural network, the system also features a weight optimization module (160), which utilizes a modified Bald Eagle Optimizer (MBEO) to refine the weights and mitigate the risk of local optima.
In an embodiment of the disclosure, the weight optimization module (160) optimizes the weights by:
P_(new,i )= P_best+ α*w_i* (P_mean-P_i)………………..[1]
w_i=exp⁡(P_i /P_best )………………………….[2]
〖 P〗_new = P_best+ α*exp⁡(P_i /P_best )*(P_mean-P_i)………………..[3]
where Pbest is the best solution, Pi is the current position, Pmean is the mean of all previous positions, and α is a parameter for a position with a value ranging from 1.5 to 2, ensuring that the modified Bald Eagle Optimizer (MBEO) escapes the local optima.
In a preferred embodiment of the disclosure, the system (100) can detect infant cardiac issues that might be missed by human doctors by analyzing a few key factors to identify patterns in time series data and highlight temporal dependencies. The first-order derivative of the infant health data is combined with LSTM networks to capture dynamic features and fluctuations in the ECG waveform instead of the traditional LSTM. This derivative represents the rate of change at each point in the time series signal, providing insights into the signal's changing tendencies. By adding the derivative as an additional input, the LSTM can effectively identify percentage changes in the ECG waveform over time. This approach allows the CNN to extract critical spatial features, while the LSTM captures temporal dependencies, enhancing the early detection of cardiac issues in infants.
In another embodiment of the disclosure, the hybrid neural network implemented by the cardiac analysis module (140) employs a sequential process starting with the computation of first-order derivatives from the infant health data, which are then integrated into a CNN-LSTM model. This combination enables the effective capture of both spatial and temporal information inherent in the analysis of infant health data. The pre-processed infant health data and its derivatives are inputted into a CNN, which extracts spatial features through convolutional layers, pooling layers, and non-linear activations. The resultant high-level features depicting specific patterns in the ECG signal are subsequently fed into an LSTM with first order derivate network. The LSTM with first order derivate utilizes its memory cell and gating mechanisms to discern short- and long-term associations, thereby learning patterns and correlations over time. The output from the LSTM with first order derivatives is then directed to one or more fully connected layers for classification or predictive tasks, offering insights into various cardiac conditions. Following the initial model setup, the weight optimization module (160) is configured to refine the network's weights. These methods strategically adjust weight positions based on optimization procedures and are evaluated using a fitness function to select the most accurate configuration. The optimized network ensures minimal errors in predictions.
Figure 2 illustrates a flowchart for a system for detection and classification of heart diseases (100) in accordance with an embodiment of the disclosure. The data collection module (110), captures health information from diverse sources such as clinical electrocardiogram (ECG) devices, wearable technology, telemedicine platforms, and electronic health records (110a), subsequently storing this data in a centralized repository (120). This data is then received by the pre-processing module (130), where various techniques are employed to standardize and normalize the data, including the removal of high-frequency noise and low-frequency drift, leading to the generation of pre-processed infant health data. The core of the system lies in the cardiac analysis module (140), which implements a hybrid neural network model combining a convolutional neural network (CNN) and a Long Short Term Memory (LSTM) network. The CNN extracts features from the pre-processed data, producing a feature map, which is then analyzed by the LSTM to calculate a first-order derivative. This analysis outputs probability values across four predefined heart disease categories: no disease, low risk, mild disease, and advanced disease, with the class of highest probability being selected as the final classification. Additionally, the weight optimization module (160) enhances the model's accuracy by optimizing network weights using a Modified Bald Eagle Optimizer.
Figures 3A and 3B depict the steps involved in method (300) for detection and classification of heart diseases. The order in which method 300 is described is not intended to be construed as a limitation, and any number of the described method steps may be combined in any order to implement method 300, or an alternative method. Furthermore, method 300 may be implemented by processing resource or computing device(s) through any suitable hardware, non-transitory machine-readable medium/instructions, or a combination thereof. The method 300 comprises the following steps:
At step 302, the method (300) includes capturing, by a data collection module (110), infant health data through one or more data capturing sources, including clinical electrocardiogram (ECG) devices, wearable devices, telemedicine platforms, and electronic health records, and storing the infant health data in a repository (120);
At step 304, the method (300) includes receiving, by a pre-processing module (130), said infant health data from the data capturing sources or the repository (120);
At step 306, the method (300) includes implementing, by the pre-processing module (130), a set of pre-processing techniques on the infant health data to standardize and normalize the infant health data, thereby generating pre-processed infant health data;
At step 308, the method (300) includes implementing, by a cardiac analysis module (140), a hybrid neural network model for cardiac analysis, wherein the hybrid neural network model includes a first neural network model and a second neural network model;
At step 310, the method (300) includes receiving, by the cardiac analysis module (140), the pre-processed infant health data and applying the first neural network model to generate a feature map;
At step 312, the method (300) includes receiving, by the cardiac analysis module (140), the feature map as input data and applying the second neural network model to calculate a first-order derivative of the input data;
At step 314, the method (300) includes outputting, by fully connected layers of the hybrid neural network model implemented by the cardiac analysis module (140), probability values distributed across four predefined heart disease classes, including no disease, low risk, mild disease, and advanced disease;
At step 316, the method (300) includes selecting, by the cardiac analysis module (140), the class with the highest probability value as a final classification output; and
At step 302, the method (300) includes generating, by a report generation module (150), a detailed cardiac health report of the infant based on the analysis and classification results, wherein the final output is included in the report.
The functions described herein may be implemented in hardware, software executed by a processor, firmware, or any combination thereof. If implemented in software executed by a processor, the functions may be stored on or transmitted over as one or more instructions or codes on a computer-readable medium. Other examples and implementations are within the scope and spirit of the disclosure and appended claims. For example, due to the nature of software, functions described above can be implemented using software executed by a processor, hardware, firmware, hardwiring, or combinations of any of these. Features implementing functions may also be physically located at various positions, including being distributed such that portions of functions are implemented at different physical locations.
The foregoing description of the embodiments has been provided for purposes of illustration and is not intended to limit the scope of the disclosure. Individual components of a particular embodiment are generally not limited to that particular embodiment but are interchangeable. Such variations are not to be regarded as a departure from the disclosure, and all such modifications are considered to be within the scope of the disclosure.
TECHNICAL ADVANCEMENTS
The disclosure described herein above has several technical advantages including, but not limited to, the realization of a system for detection and classification of heart diseases, that:
enables increased sensitivity in identifying subtle cardiac indicators through combining spatial and temporal analysis of ECG signals effectively;
demonstrates superior accuracy (96;5%) and robustness in detecting various cardiac conditions through experimental results, significantly outperforming traditional methods;
enables real-time analysis of ECG data thereby facilitating timely medical interventions;
reduces healthcare costs through early detection and intervention, significantly lowering long-term expenses associated with advanced cardiac diseases for families and healthcare systems;
provides accurate preliminary diagnoses to improve resource allocation in healthcare settings;
ensures equitable access to essential medical services and enhances public health outcomes by enabling easy deployment in rural areas and areas with limited medical infrastructure;
adapts to different populations or conditions, ensuring effectiveness across diverse settings and datasets;
promotes economic growth through healthier populations by improving infant health outcomes;
detects dynamic changes in ECG waveforms over time, thereby enhancing the model's ability to identify conditions not evident in static analyses;
enables cost-effective healthcare solutions by facilitating accurate early detection of cardiac diseases, reducing the need for expensive diagnostic procedures and interventions;
contributes to healthier populations by facilitating early diagnosis and intervention to significantly lower mortality rates among infants with cardiac conditions;
prevents escalation of heart diseases through early detection and timely treatment, reducing the incidence of chronic conditions requiring ongoing management;
facilitates integration with telemedicine platforms to enhance remote healthcare services, offering continuous monitoring and support for families, which can lead to improved health management and reduced hospital visits;
improves convergence speed and overall performance compared to conventional methods;
facilitates advancement in medical research by utilizing the invention to analyze large datasets, revealing insights into the early manifestations and progression of cardiac diseases in infants, which can inform new treatment protocols and enhance understanding of various diseases; and
enables advanced training of medical professionals by demonstrating AI diagnostic strengths and providing hands-on experience with ECG data interpretation.
The embodiments herein and the various features and advantageous details thereof are explained with reference to the non-limiting embodiments in the following description. Descriptions of well-known components and processing techniques are omitted so as to not unnecessarily obscure the embodiments herein. The examples used herein are intended merely to facilitate an understanding of ways in which the embodiments herein may be practiced and to further enable those of skill in the art to practice the embodiments herein. Accordingly, the examples should not be construed as limiting the scope of the embodiments herein.
The foregoing description of the specific embodiments so fully reveals the general nature of the embodiments herein that others can, by applying current knowledge, readily modify and/or adapt for various applications such specific embodiments without departing from the generic concept, and, therefore, such adaptations and modifications should and are intended to be comprehended within the meaning and range of equivalents of the disclosed embodiments. It is to be understood that the phraseology or terminology employed herein is for the purpose of description and not of limitation. Therefore, while the embodiments herein have been described in terms of preferred embodiments, those skilled in the art will recognize that the embodiments herein can be practiced with modification within the spirit and scope of the embodiments as described herein.
The use of the expression "at least" or "at least one" suggests the use of one or more elements or ingredients or quantities, as the use may be in the embodiment of the disclosure to achieve one or more of the desired objects or results.
While considerable emphasis has been placed herein on the components and component parts of the preferred embodiments, it will be appreciated that many embodiments can be made and that many changes can be made in the preferred embodiments without departing from the principles of the disclosure. These and other changes in the preferred embodiment as well as other embodiments of the disclosure will be apparent to those skilled in the art from the disclosure herein, whereby it is to be distinctly understood that the foregoing descriptive matter is to be interpreted merely as illustrative of the disclosure and not as a limitation. , Claims:WE CLAIM
A system (100) for detection and classification of heart diseases, said system (100) comprising:
a data collection module (110) configured to capture infant health data through one or more data capturing sources (110a) including clinical electrocardiogram (ECG) devices, wearable devices, telemedicine platforms, and electronic health records, and further configured to store said infant health data in a repository (120);
a pre-processing module (130) configured to receive said infant health data from said data collection module (110) or said repository (120), and further configured to implement a set of pre-processing techniques on said infant health data to standardize and normalize said infant health data so as to generate pre-processed infant health data;
a cardiac analysis module (140) configured to implement a hybrid neural network model having a first neural network model and a second neural network model, said cardiac analysis module (140) comprises:
a first neural network module (140A) configured to receive said preprocessed infant health data from the pre-processing module (130), and further configured to implement the first neural network model on said preprocessed infant health data to generate a feature map;
a second neural network module (140B) configured to receive the feature map generated by the first neural network model as input data, and further configured to implement the second neural network model on the input data to calculate a first-order derivative of the input data;
a classification module (140C) comprising fully connected layers of the hybrid neural network model to output probability values distributed across four predefined heart disease classes, including no disease, low risk, mild disease, and advanced disease; and
a report generation module (150) configured to cooperate with said cardiac analysis module (140) to generate a detailed cardiac health report of the infant, in which a class having the highest probability value being selected as a final output.
The system (100) as claimed in claim 1, wherein said Infant health data comprises data indicating cardiac health status of an infant including heart rate, electrocardiogram waveform, heart rhythm, blood pressure, oxygen saturation, respiratory rate, weight and growth parameters, family history, and physical examination findings.
The system (100) as claimed in claim 1, further comprises a weight optimization module (160) configured to optimize the weights of the hybrid neural network model by means of a modified Bald Eagle Optimizer (MBEO) to avoid local optima.
The system (100) as claimed in claim 2, wherein the weight optimization module (160) optimizes the weights by:
P_(new,i )= P_best+ α*w_i* (P_mean-P_i)………………..[1]
w_i=exp⁡(P_i /P_best )………………………….[2]
〖 P〗_new = P_best+ α*exp⁡(P_i /P_best )*(P_mean-P_i)………………..[3]
where Pbest is the best solution, Pi is the current position, Pmean is the mean of all previous positions, and α is a parameter for a position with a value ranging from 1.5 to 2, ensuring that the modified Bald Eagle Optimizer (MBEO) escapes the local optima.
The system (100) as claimed in claim 1, wherein the first neural network model is a convolution neural network (CNN) comprising:
a first convolutional layer with 32 filters applied to the pre-processed data;
a first max pooling layer applied to an output data of the first convolutional layer;
a second convolutional layer with 64 filters applied to an output of the first max pooling layer;
a second max pooling layer applied to an output of the second convolutional layer; and
a 32-bit embedding layer applied to an output of the second max pooling layer so as to generate the feature map.
The system (100) as claimed in claim 1, wherein the second neural network model is a Long Short Term Memory network (LSTM) comprising:
an input layer to receive the feature map generated by the CNN as the input data;
a first dense layer consisting of 64 neurons to apply a first-order derivative operation on the input data;
a second dense layer consisting of 4 neurons to process an output of the first dense layer; and
fully connected layers to calculate probability values distributed across the four predefined heart disease classes based on the output of the second dense layer, the four predefined heart disease classes including no disease, low risk, mild disease, and advanced disease;
wherein the class with the highest probability is selected as the final output.
The system (100) as claimed in claim 1, wherein said set of pre-processing techniques comprises the following steps:
receiving the infant health data from the repository (120);
removing high-frequency noise from the infant health data by using a low pass filter with a cutoff frequency of approximately 50 Hz;
eliminating low-frequency baseline drift in the infant health data by using a high pass filter with a cutoff frequency of 0.5 Hz;
generating filtered infant health data and standardizing said infant health data by z-score normalization;
calculating the mean and standard deviation of said filtered infant health data;
subtracting the mean from each data point in said filtered infant health data;
dividing the result by the standard deviation and standardizing the data values, resulting in data with a mean of 0 and a standard deviation of 1; and
generating pre-processed infant health data and feeding said pre-processed infant health data to the cardiac analysis module (140).
The system (100) as claimed in claim 1, wherein said hybrid neural network model is configured to implement a combination of computational neural networks (CNN) and Long short term memory networks (LSTM).
The system (100) as claimed in claim 3, wherein said weight optimization module (160) is configured to update network weights in the hybrid neural network model implemented by said cardiac analysis module (140) by means of metaheuristic optimization techniques.
The system (100) as claimed in claim 1, wherein the hybrid neural network model implemented by said cardiac analysis module (140) is further configured to detect and classify heart diseases for individuals with diverse physiological and health characteristics including but not limited to varying age-groups, heterogenous populations, and individuals with specific health conditions.
The system (100) as claimed in claim 1, wherein the hybrid neural network model implemented by said cardiac analysis module (140) is further configured to be able to detect and classify other diseases including but not limited to respiratory diseases, renal diseases, metabolic disorders, dermatological conditions, and psychological disorders.
The system (100) as claimed in claim 1, wherein said system (100) is further configured to operate as a cloud-based system and as a local system.
A method (300) for detection and classification of heart diseases, said method (300) comprises the following steps:
capturing, by a data collection module (110), infant health data through one or more data capturing sources, including clinical electrocardiogram (ECG) devices, wearable devices, telemedicine platforms, and electronic health records, and storing the infant health data in a repository (120);
receiving, by a pre-processing module (130), said infant health data from the data capturing sources or the repository (120);
implementing, by the pre-processing module (130), a set of pre-processing techniques on the infant health data to standardize and normalize the infant health data, thereby generating pre-processed infant health data;
implementing, by a cardiac analysis module (140), a hybrid neural network model for cardiac analysis, wherein the hybrid neural network model includes a first neural network model and a second neural network model;
receiving, by the cardiac analysis module (140), the pre-processed infant health data and applying the first neural network model to generate a feature map;
receiving, by the cardiac analysis module (140), the feature map as input data and applying the second neural network model to calculate a first-order derivative of the input data;
outputting, by fully connected layers of the hybrid neural network model implemented by the cardiac analysis module (140), probability values distributed across four predefined heart disease classes, including no disease, low risk, mild disease, and advanced disease;
selecting, by the cardiac analysis module (140), the class with the highest probability value as a final classification output; and
generating, by a report generation module (150), a detailed cardiac health report of the infant based on the analysis and classification results, wherein the final output is included in the report.
Dated this 08th Day of November, 2024

_______________________________
MOHAN RAJKUMAR DEWAN, IN/PA - 25
OF R. K. DEWAN & CO.
AUTHORIZED AGENT OF APPLICANT

Documents