A PRIVACY PRESERVING HEALTH MONITORING SYSTEM

Abstract

The present invention is relating to a privacy preserving health monitoring system (100) for real time health monitoring while maintaining confidentiality and privacy of user information, even in absence of internet connectivity, includes biometric parameter generator (102) for converting sensed sensitive user health data into bio-electrical signals (D_i ) transmitted over primary network (104), primary processing unit (106) for processing and analyzing bio-electrical signals to predict health condition based on trained primary prediction model and generate primary weights (w_i^(t+1) ) of primary prediction model transmitted over secondary network (108) eliminating need of actually sharing of original sensitive user health data and maintaining confidentiality and privacy of user, and secondary processing unit (110) processes and analyzes primary weights of multiple primary processing unit (106) for modifying global weights (w^(t+1) ) of global prediction model to generate updated new primary prediction model transmitted to multiple primary processing unit (106) for replacing the primary prediction model.

Specification

Description:FIELD OF THE INVENTION
The present invention relates to a privacy preserving health monitoring system. More particularly it relates to a privacy preserving health monitoring system for the protection of sensitive user health data.

BACKGROUND OF THE INVENTION
Advances in wearable technologies, cloud computing and data analytics have facilitated remote tracking of health metrics for example diabetes is common lifestyle disease that requires continuous monitoring. Diabetes management is a critical aspect of healthcare, necessitating constant monitoring and precise control of blood glucose levels to prevent serious complications. Continuous Glucose Monitoring (CGM) systems have emerged as a vital technology for individuals with diabetes, allowing real-time tracking of glucose levels through a small sensor placed under the skin. This sensor measures glucose levels in the interstitial fluid, transmitting data wirelessly to a receiver, typically an app or device running on a fog computing server. By continuously collecting data, CGM systems enable users and healthcare professionals to analyse trends and fluctuations in glucose levels, facilitating more informed decisions for better glycemic control.

Despite the advantages offered by CGM technology, one significant challenge remains: the protection of sensitive health data. The effectiveness of Machine Learning (ML) models in predicting glucose levels depends heavily on the availability of diverse datasets collected from numerous users. However, many individuals are understandably hesitant to share their personal health information due to privacy concerns. This apprehension is particularly pronounced in the context of diabetes management, where data privacy is paramount.

Thus, there is an essential need of a health monitoring system for privacy preserving and the protection of sensitive data.

OBJECTIVES OF THE INVENTION
The key objective of the present invention is to provide a privacy preserving health monitoring system for maintaining the confidentiality of user information thereby preserving privacy of user data.

Another objective of the present invention is to implement federated learning to train machine learning models without transferring user data, maintaining the confidentiality of user information for ensuring that user data remains within the local fog servers.

Another objective of the present invention is to provide the privacy preserving system that uses fog computing to process data locally and deliver health predictions in real-time with reduction in latency and enhancement the timeliness of interventions.

Another objective of the present invention is to provide the privacy preserving system that integrates with wearable Continuous Glucose Monitoring (CGM) Devices to continuously monitor blood glucose levels for wirelessly transmitting CGM data to the fog servers, where CGM data is used to predict future glucose levels.

Another objective of the present invention is to provide minimizes data transfer and improves data security by processing data from multiple user on local fog servers by performing machine learning tasks and transmitting only the necessary prediction model updates to the central server.

Another objective of the present invention is to provide the privacy preserving system designed to be scalable, with multiple fog servers able to handle data from various users allowing a flexible setup that can be deployed in different environments, from individual homes to healthcare facilities.

SUMMARY OF THE INVENTION
A summary is provided to facilitate an understanding of the innovative characteristics unique to the disclosed embodiments and is not intended for the full description of the invention. In accordance, a full appreciation of the various aspects of the preferred embodiments disclosed herein can be understood in depth by taking the entire specification, claims, drawings, and abstract as a whole.
The present disclosure relates to a privacy preserving health monitoring system configured for real time health monitoring while maintaining the confidentiality and privacy of user information, even in the absence of internet connectivity. In an embodiment, the privacy preserving health monitoring system includes a biometric parameter generator, a primary network, a primary processing unit, secondary network and a secondary processing unit. The biometric parameter generator converts the sensed sensitive user health data into corresponding bio-electrical signals (D_i ) and transmits over primary network. The primary processing unit processes and analyses the bio-electrical signals (D_i ) of sensitive user health data, received from the biometric parameter generator over the primary network for predicting the health condition based on the trained primary prediction model and generating the primary weights (w_i^(t+1) ) of the primary prediction model transmitted over a secondary network to eliminate the need of actually sharing of the original sensitive user health data and maintaining the confidentiality and privacy of the user. The secondary processing unit processes and analyses the primary weights (w_i^(t+1) ) received from the multiple primary processing unit over corresponding secondary network providing the internet for modifying the global weights (w^(t+1) ) of the global prediction model to generate the updated new primary prediction model transmitted to the multiple primary processing unit for replacing the primary prediction model.
These and other features, aspects and advantages of the present subject matter will become better understood with reference to the following description. This summary is provided to introduce a selection of concepts in a simplified form. This summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.
BRIEF DESCRIPTION OF THE DRAWINGS
The present invention will be described in more detail herein with the detailed description that relates well with the preferred embodiments of the invention as explained with reference to the following accompanying schematic drawings:

Fig. 1 depicts a privacy preserving health monitoring system 100, according to one or more embodiments of the present disclosure;

Fig. 2 depicts a biometric parameter generator 102, according to one or more embodiments of the present disclosure;

Fig. 3 depicts a primary processing unit 106, according to one or more embodiments of the present disclosure; and

Fig. 4 depicts a secondary processing unit 110, according to one or more embodiments of the present disclosure.

DETAILED DESCRIPTION OF THE INVENTION
The preferred embodiments are provided so that the disclosure will be thorough and will fully convey the scope to those who are skilled in the art. Various specific details are set as specific components to provide an overall understanding of the preferred embodiments of the present disclosure. It will be apparent to those skilled in the art that the specific details need not be employed, and the embodiments may be embodied in many different forms and the steps followed do not limit the scope of the disclosure. It is also to be understood that the terminology used herein is for the purpose of describing only the particular embodiments of the invention and is not intended to limit the scope of the invention in any manner. It must also be noted that as used herein and in the appended claims, the singular forms "a," "an," and "the" include plural references unless the context clearly dictates otherwise.

The present disclosure discloses a privacy preserving health monitoring system for real time health monitoring of the user even in the absence of internet connectivity while maintaining the confidentiality of user information.

The privacy preserving health monitoring system preserves the privacy of user health data without the use of heavyweight encryption or differential privacy techniques that add noise to the user health data such as protected health information (PHI) encompassing any health-related data that can be linked to an individual for example, blood pressure, heart rate, blood glucose levels, respiratory rate, body mass index (BMI), oxygen saturation, cholesterol levels, electrocardiogram (ECG), skinfold thickness, etc.

The privacy preserving health monitoring system is designed to protect individuals' personal health information while allowing for data processing and analysis for predicting the real time health conditions such as hypertension, hypotension, cardiovascular disease, arrhythmias, cardiovascular fitness, stress levels, diabetes, hypoglycemia, infection, fever, hypothermia, respiratory issues, lung conditions, overall fitness, overweight, obesity, underweight, respiratory disorders, heart conditions, sleep apnea, risk of heart disease, metabolic syndrome, heart disease, arrhythmias, heart attacks, body fat percentage, overall fitness level etc. The privacy preserving health monitoring system ensures that sensitive user health data is not exposed or misused and maintains the confidentiality of the sensitive user health data.

Fig. 1 depicts a privacy preserving health monitoring system 100, according to one or more embodiments of the present disclosure.
The privacy preserving health monitoring system 100 (hereafter known as system 100) is configured to provide the protection of the sensitive user health data (denoted as the? D?_i ) from breach or misuse. The system 100 is configured to maintain the confidentiality and the privacy. The system 100 is configured to be a cloud computing system, an edge computing system, a local computing system, a hybrid computing system, a distributed computing system, a blockchain computing system, a serverless computing system, a fog computing system etc. In an embodiment, the system 100 is configured to be fog computing system. The system 100 is configured to use a privacy preserving machine learning, differential privacy, homomorphic encryption, secure multi-party computation (SMPC), federated learning (with a focus on local training), anonymization techniques, data masking, access control and governance policies, blockchain for data integrity, synthetic data generation, zero-knowledge proofs etc. for investigating the effective and predictions of the real-time health condition based on data processing and analysis. In an embodiment, the system 100 is configured to use a privacy preserving federated learning (with a focus on local training) for investigating the effective and real-time glucose level predictions in diabetes patients based on continuous glucose monitoring (CGM) by data processing and analysis. The system 100 is configured to ensure that sensitive data is not exposed or misused and to provide the continuous health monitoring the user in real-time even in the absence of internet connectivity.

The system 100 includes a biometric parameter generator 102, a primary network 104, a primary processing unit 106, a secondary network 108, and a secondary processing unit 110.

The biometric parameter generator 102 is configured to sense the sensitive user health data for example, blood pressure, heart rate, blood glucose levels, respiratory rate, body mass index (BMI), oxygen saturation, cholesterol levels, electrocardiogram (ECG), skinfold thickness, etc. In an embodiment, the biometric parameter generator 102 is configured to sense the sensitive user health data of blood glucose levels by using the continuous glucose monitoring (CGM) device. The biometric parameter generator 102 is configured to convert the sensed sensitive user health data into the corresponding bio-electrical signals? ( D?_i ). The biometric parameter generator 102 is configured to transmit the generated bio-electrical signals ? ( D?_i ) of the sensitive user health data over the primary network 104 even in the absence of internet connectivity.

The primary network 104 is configured to be a Wi-Fi networks, cellular networks, Bluetooth networks, zigbee, LoRaWAN (Long Range Wide Area Network), satellite networks, mesh networks, WiMAX (Worldwide Interoperability for Microwave Access), near field communication (NFC), etc. In an embodiment, the primary network 104 is configured to be Wi-Fi networks may or may not be providing an internet. The primary network 104 is configured to be connected to the biometric parameter generator 102 even in absence the internet connectivity. The primary network 104 is configured to receive the bio-electrical signals of the sensitive user health data from the biometric parameter generator 102. In an embodiment, the primary network 104 is configured to receive the bio-electrical signals of the sensitive user health data of the glucose level from the biometric parameter generator 102.

The primary processing unit 106 is configured to be connected to biometric parameter generator 102 over the primary network 104 even in absence of the internet connectivity. The primary processing unit 106 is configured to be computing device or personal electronic for example, a smart phone, a laptop, a desktop, a tablet etc. The primary processing unit 106 is configured to receive the bio-electrical signals (D_i) of the sensitive user health data from the biometric parameter generator 102 over the primary network 104. In an embodiment, the primary processing unit 106 is configured to receive the bio-electrical signals (D_i) of the sensitive user health data of the glucose level from the biometric parameter generator 102 over the primary network 104. The primary processing unit 106 is configured to process and analyse the received bio-electrical signals (D_i) sensitive user health data for predicting the health condition based on the primary prediction model for example, long short-term memory (LSTM) model, Recurrent Neural Network (RNN), Gated Recurrent unit (GRU), Autoregressive integrated moving average (ARIMA) etc. trained by using the privacy preserving machine learning, differential privacy, homomorphic encryption, secure multi-party computation (SMPC), federated learning (with a focus on local training), anonymization techniques, data masking, access control and governance policies, blockchain for data integrity, synthetic data generation, zero-knowledge proofs etc. In an embodiment, the primary processing unit 106 is configured to process and analyse the received bio-electrical signals (D_i) sensitive data of the glucose level of the user for predicting the diabetic health condition based on long short-term memory (LSTM) model as the primary prediction model trained by using the federated learning (with a focus on local training). The primary processing unit 106 is configured to store the received bio-electrical signals of the sensitive user health data and the corresponding predictions for multiple days. In an embodiment, the primary processing unit 106 is configured to store the received bio-electrical signals of the sensitive user health data and the corresponding predictions for at least 15 days. The primary processing unit 106 is configured to predict biometric parameters of the health condition even in absence of internet connectivity based on the trends, patterns, fluctuations in biometric parameters of the user over time, risk factors and recommend with reference to multiple users without the need to actually share the person's original sensitive data of health condition elsewhere. For example, the primary processing unit 106 is configured to forecast potential hypoglycemic or hyperglycemic events, enabling timely interventions or adjustments to insulin dosages without risk of privacy leaks in real time. The primary processing unit 106 is configured to generate the primary weights (w_i^(t+1)) of the primary prediction model based on the received bio-electrical signals (D_i) of the sensitive user health data transmitted over the secondary network 108 for eliminating the need of actually sharing of the original sensitive user health data and maintaining the confidentiality and privacy of the user.

The secondary network 108 is configured to be a Wi-Fi networks, cellular networks, LoRaWAN (Long Range Wide Area Network), satellite networks, mesh networks, WiMAX (Worldwide Interoperability for Microwave Access), etc. providing the internet connectivity. In an embodiment, the secondary network 108 is configured to be Wi-Fi networks providing internet connectivity. The secondary network 108 is configured to be connected to the primary processing unit 106 via internet connectivity. The secondary network 108 is configured to receive the primary weights w_i^(t+1) generated by the primary processing unit 106 of multiple days instead of the original sensitive user health data for maintaining the privacy and confidentiality of the user. In an embodiment, the secondary network 108 is configured to receive the primary weights w_i^(t+1) of the bio-electrical signals of the of the glucose level generated by the primary processing unit 106 of multiple days instead of the original sensitive user health data of glucose levels and the corresponding predictions of diabetic health condition for maintaining the privacy and confidentiality of the user.

The secondary processing unit 110 is configured to be connected to the multiple primary processing unit 106 over their corresponding secondary network 108 providing the internet. The secondary processing unit 110 is configured to receive the primary weights w_i^(t+1) from the multiple primary processing unit 106 over their corresponding secondary network 108. In an embodiment, the secondary processing unit 110 is configured to receive the primary weights w_i^(t+1) of bio-electrical signals of the sensitive user health data of the glucose level from the multiple primary processing unit 106 over their corresponding secondary network 108. The secondary processing unit 110 is configured to process and analyse the received the primary weights w_i^(t+1) for modifying the global weights w^(t+1) of the global prediction model for example long short-term memory (LSTM) model, Recurrent Neural Network (RNN), Gated Recurrent unit (GRU), Autoregressive integrated moving average (ARIMA) etc. In an embodiment, the secondary processing unit 110 is configured to process and analyse the primary weights w_i^(t+1) of glucose level of the user for modifying the global weights w^(t+1) of the long short-term memory (LSTM) model as the global prediction model. The secondary processing unit 110 is configured to update the global weights w^(t+1) of global prediction model based on the primary weights w_i^(t+1) for multiple days for generating the updated new primary prediction model. The secondary processing unit 110 is configured to transmit the updated new prediction model to the multiple primary processing unit 106 over their corresponding secondary network 108 for replacing the primary prediction model by the updated new prediction model. The secondary processing unit 110 may be a cloud server, on premises servers, dedicated servers, virtual private servers (VPS), hybrid, hosting, colocation services, edge computing, managed hosting, local network storage (NAS/SAN), etc. In an embodiment, secondary processing unit 110 may be a cloud server.

Fig. 2 depicts the biometric parameter generator 102, according to one or more embodiments of the present disclosure.
The biometric parameter generator 102 is configured to sense the sensitive user health data. The biometric parameter generator 102 is configured to convert sense the sensitive user health data into corresponding bio-electrical signals. The biometric parameter generator 102 is configured to transmit the bio-electrical signals to the primary processing unit 106 over the primary network 104. The biometric parameter generator 102 includes a plurality of sensors 202 and a first communication unit 204.

The plurality of sensors 202 is configured to sense the sensitive user health data such blood pressure, heart rate, blood glucose levels, respiratory rate, body mass index (BMI), oxygen saturation, cholesterol levels, electrocardiogram (ECG), skinfold thickness, etc. In an embodiment, the plurality of sensors 202 is configured to sense the sensitive health data of blood glucose levels. The plurality of sensors 202 is configured to be a sphygmomanometer (manual or digital), heart rate monitor (wristband, chest strap, or smartwatch), glucometer, digital thermometer, pulse oximeter (also measures oxygen saturation), weighing scale and height measuring device, pulse oximeter, lipid panel (usually done in a lab), ECG machine or portable ECG monitor, caliper etc. In an embodiment, the plurality of sensors 202 is configured to be a continuous glucose monitoring (CGM) device, used to monitor blood glucose levels in real-time throughout the day and night by measuring glucose levels in the interstitial fluid, which is a thin layer of fluid that surrounds the cells, with a small sensor placed under the skin, on the abdomen or the back of the upper arm. The plurality of sensors 202 is configured to convert the sensed sensitive user health data into corresponding bio-electrical signals?(D?_i).

The first communication unit 204 is configured to establish the communication between the biometric parameter generator 102 and the primary processing unit 106 over the primary network 104.

The first communication unit 204 is configured to transmit the bio-electrical signals generated by the plurality of sensor 202 to primary processing unit 106 over the primary network 104 after the communication link is established.

Fig. 3 depicts the primary processing unit 106, according to one or more embodiments of the present disclosure.
The primary processing unit 106 is configured to receive bio-electrical signals of the sensitive user health data from the biometric parameter generator 102 over the primary network 104 for predicting the corresponding health conditions and generating the primary weights (w_i^(t+1)). In an embodiment, the primary processing unit 106 is configured to receive the bio-electrical signals of the sensitive user health data of glucose levels from the biometric parameter generator 102 over the primary network 104 for predicting the corresponding health conditions and generating the primary weights (w_i^(t+1)).

The primary processing unit 106 includes a second communication unit 302, a prediction module 304, a processor 306, a memory 308, a database 310 and a weight generator module 312.

The second communication unit 302 is configured to establish the communication between the primary processing unit 106 and the biometric parameter generator 102 over the primary network 104 even in absence of internet connectivity. The second communication unit 302 is configured to establish the communication between the primary processing unit 106 and the secondary processing unit 110 over the secondary network 108 via internet connectivity. The second communication unit 302 is configured to receive the bio-electrical signals from the biometric parameter generator 102 over the primary network 104. The second communication unit 302 is configured to transmit the primary weights (w_i^(t+1)) to the secondary processing unit 110 over the secondary network 108.

The prediction module 304 is configured to receive the bio-electrical signals of the sensitive user health data from the second communication unit 302. The prediction module 304 is configured to process the received bio-electrical signals of the sensitive user health data by using the processor 306 based on the primary prediction model stored in the memory 308 for generating the prediction of the user's health conditions. The prediction module 304 is configured to store the received bio-electrical signals of the sensitive health data of the user and the corresponding prediction of the user's health in the database 310. The prediction module 206 is configured to store the bio-electrical signals of the sensitive health data of the user and the corresponding prediction of the user's health of multiple days in the database 310.

The weight generator module 312 is configured to access the stored bio-electrical signals of the sensitive user health data and the corresponding prediction of the user's health from the database 310. The weight generator module 312 is configured to generate the primary weights w_i^(t+1) by using the processor 306 based on the bio-electrical signals D_i of the sensitive user health data of multiple days and the instructions stored in the memory 308, for example:

w_i^(t+1)= w_i^t- ? ? L_i (w_i^t ,D_i )
Where:
w_i^(t+1) : The updated primary weights at i^th primary processing unit 106 after local training.
w_i^t : Primary weights before the update.
? : Learning rate.
? L_i (w_i^t ,D_i) : Gradient of the loss function L at i^th primary processing unit 106 based on D_i.

The weight generator module 312 is configured to transmit the generated primary weights w_i^(t+1) to the secondary processing unit 110 through the communication link established communication link between the primary processing unit 106 and secondary processing unit 110 over the secondary network 108 via the internet.

Fig. 4 depicts the secondary processing unit 110, according to one or more embodiments of the present disclosure.
The secondary processing unit 110 is configured to receive primary weights w_i^(t+1) of the bio-electrical signals (D_i) of the sensitive user health data, generated by the multiple weight generator module 312 and transmitted by multiple the second communication unit 302 of the multiple primary processing unit 106 over the corresponding secondary network 108 via internet of the for updating the global weights ?(w?^(t+1)) of the global prediction model. In an embodiment, the secondary processing unit 110 is configured to receive primary weights w_i^(t+1) of the bio-electrical signals (D_i) of the sensitive user health data of the glucose levels, generated by the multiple weight generator module 312 and transmitted by multiple the second communication unit 302 of the multiple primary processing unit 106 over the corresponding secondary network 108 via internet of the for updating the global weights ?(w?^(t+1)) of the global prediction model.

The secondary processing unit 110 includes a third communication unit 402, a modifying module 404, a processor 406, a memory 408, a database 410.

The third communication unit 402 is configured to establish the communication between the secondary processing unit 110 and the multiple primary processing unit 106 over the secondary network 108 via internet. The third communication unit 402 configured to transmit the updated new prediction model to the multiple primary processing unit 106 over the corresponding secondary network 108 via internet.

The modifying module 404 is configured to receive the primary weights w_i^(t+1) of the bio-electrical signals from the third communication unit 402. The modifying module 404 is configured to process the received primary weights w_i^(t+1) of the bio-electrical signals by using the processor 406 based on the instructions stored in the memory 408 for generating the global weights w^(t+1) for the global prediction model. The modifying module 404 is configured to store the generated the global weights w^(t+1) in the database 410. The modifying module 404 is configured to update the global prediction model based on the generated global weights w^(t+1) for generating the updated new primary prediction model. The modifying module 404 is configured to generate the updated new primary prediction module based on the updated global weights w^(t+1) based on the primary weights w_i^(t+1) received from the multiple primary processing unit 106 for example:
w^(t+1)= ?_(i=1)^N¦n_i/n w_i^(t+1)
Where:
w^(t+1) : Global weights after aggregation.
w_i^(t+1) : Primary weights from i^th primary processing unit 106 after local training.
n_i : Number of bio-electrical signals samples on i^th primary processing unit 106.
N: Total number of the primary processing unit 106.
n: Total number of bio-electrical signals samples across all primary processing unit 106, n=?_(i=1)^N¦n_i .

The modifying module 404 is configured to transmit the updated new primary prediction module to the multiple primary processing unit 106 for replacing the primary prediction model with the updated new primary prediction model.

Although a preferred embodiment of the invention has been illustrated and described, it will at once be apparent to those skilled in the art that the invention includes advantages and features over and beyond the specific illustrated construction. Accordingly, it is indented that the scope of the invention be limited solely by the scope of the hereinafter appended claims, and not by the forgoing specification, when interpreted in light of the relevant prior art. , Claims:1. A privacy preserving health monitoring system (100) configured for real time health monitoring while maintaining the confidentiality and privacy of user information, even in the absence of internet connectivity, comprising:
a biometric parameter generator (102) configured to convert the sensed sensitive user health data into the corresponding bio-electrical signals? ( D?_i ) and transmit over a primary network (104);
a primary processing unit (106) configured to process and analyze the bio-electrical signals (D_i) of sensitive user health data, received from the biometric parameter generator (102) over the primary network (104) for predicting the health condition based on the trained primary prediction model and generating the primary weights (w_i^(t+1)) of the primary prediction model transmitted over a secondary network (108) to eliminate the need of actually sharing of the original sensitive user health data and maintaining the confidentiality and privacy of the user; and
a secondary processing unit (110) configured to process and analyze the primary weights (w_i^(t+1)) received from the multiple primary processing unit (106) over corresponding secondary network (108) providing the internet for modifying the global weights ?(w?^(t+1)) of the global prediction model to generate the updated new primary prediction model transmitted to the multiple primary processing unit (106) for replacing the primary prediction model.

2. The privacy preserving health monitoring system (100) as claimed in claim 1, wherein the biometric parameter generator (102), comprising:
a plurality of sensors (202) configured to sense the sensitive user health data and convert the sensed the sensitive user data into corresponding bio-electrical signals ?(D?_i); and
a first communication unit (204) configured to establish the communication between the biometric parameter generator (102) and the primary processing unit (106) over the primary network (104) for transmitting the bio-electrical signals generated by the plurality of sensor (202) to primary processing unit (106).

3. The privacy preserving health monitoring system (100) as claimed in claim 1, wherein the primary processing unit (106), comprising:
a second communication unit (302) configured to establish the communication of the primary processing unit (106) with the biometric parameter generator 102 over the primary network (104) for receiving the bio-electrical signals and with the secondary processing unit (110) over the secondary network (108) for transmitting the primary weights (w_i^(t+1));
a prediction module (304) configured to predict the user health condition by processing the bio-electrical signals stored in a database (310), received from the second communication unit (302) using a processor (306) based on the primary prediction model stored in a memory (308); and
a weight generator module (312) configured to generate the primary weights (w_i^(t+1)) by using the processor (306) based on the bio-electrical signals (D_(i)) stored in database (310) of multiple days and the instructions stored in the memory (308) for transmitting to the secondary processing unit (110) over the secondary network (108).

4. The privacy preserving health monitoring system (100) as claimed in claim 1, wherein the secondary processing unit (110), comprising:
a third communication unit (402) configured to establish the communication between the secondary processing unit (110) and the multiple primary processing unit (106) over the secondary network (108) via internet and transmit the updated new prediction model to the multiple primary processing unit 106; and
a modifying module (404) is configured to process primary weights ?(w?_i^(t+1)) received from the third communication unit (402) by using a processor (406) based on the instructions stored in a memory (408) for generating the global weights (w^(t+1)) stored in a database (410) for updating the global prediction model based to generate the updated new primary prediction model and transmit the updated new primary prediction module to the multiple primary processing unit (106) for replacing the primary prediction model.

Documents