AU2020103057A4 - A technique to analyse glucose levels using bio sweat sensor - Google Patents

Abstract

A TECHNIQUE TO ANALYSE GLUCOSE LEVELS USING BIO-SWEAT SENSOR ABSTRACT The blood glucose is the main indicator for the patient's health, particularly for the symptoms of diabetes mellitus. In recent days, the diabetic patients have increased enormously, and this had led to approach the study of glucose measurement in order to enable the correct monitoring of glucose level. The electrochemical analysis is the simple method, and its technology has incorporated to the research level from strips to wearable devices and the implantable method. The non-invasive method will estimate the blood glucose levels with the help of sweat. The wearable sensors had a major role in continuous and non-invasive monitoring of the diabetes. The occurrence of diabetes is due to food or gene and that can be predicted with the help of machine learning to take some important precautions. The blood glucose monitoring by using the sweat will help the individual from painful method like blood-based monitoring. The new method is painless, cost-effective, and monitoring is easy for measuring the glucose. 11 P a g e A TECHNIQUE TO ANALYSE GLUCOSE LEVELS USING BIO-SWEAT SENSOR Drawings: atab ase - Data Test Data process-Ing SPlitting! - Machine! __EClssifictior Learning Prediction Traning _Agirth Rule Machine Learnrng Model Result Predicto Sw eat C cV Glucose Real-Tinme Input PM Les~ltveR . Monitoring Unit PowCud Figure 1: Block Diagram of proposed Method 1 P a g e

Description

A TECHNIQUE TO ANALYSE GLUCOSE LEVELS USING BIO-SWEAT SENSOR

Drawings:

atab ase - Data Test Data process-Ing SPlitting! - Machine! __EClssifictior Learning Prediction Traning _Agirth Rule

Machine Learnrng Model

Result Predicto

Sw eat C cV Glucose Real-Tinme Input PM Les~ltveR

. Monitoring Unit

PowCud

Figure 1: Block Diagram of proposed Method

1 Pag e

A TECHNIQUE TO ANALYSE GLUCOSE LEVELS USING BIO-SWEAT SENSOR

Description

Field of Invention:

This field of invention addresses the non-invasive method to analyze the glucose levels using the bio-sweat sensor. The occurrence of diabetes is predicted using the machine learning.

Background of invention:

Diabetes is the chronic disease, where the person will suffer from the elevated level of glucose present in the body. In recent days, most of the people are getting affected due to this disease. The blood should be checked daily for the diabetic patients and corresponding insulin is to be taken. Some of the complications like cardiovascular, kidney problems, blindness, stroke, nerve problem will be present, when the blood is taken daily from the body. The diabetes mellitus is one of the diseases which is incurable and results from the insufficient insulin in the body which causes the high blood-glucose levels or the reduction in glucose concentration which is known as hyperglycemia. The insulin is the hormone which will secret from the pancreas for the metabolic reactions which involves the glucose. The glucose will be initiated by types of cell in the body and hence it reduces the glucose concentration in the blood. Diabetes leads to many medical conditions like cystic fibrosis, TB, and the heart disease. These complications will result to the retinopathy and leads to blindness, renal failure, nerve damages, nephropathy, risk of ulcer, and the cancer. The artificial intelligence and wearable sensors are the two important fields to analyze the goals for the purpose of correct medical treatment for the patients. The integration of artificial intelligence and wearable sensors will enable the better accession of patient's data and improves the design of the wearable sensors in order to monitor the health and fitness. The artificial intelligence bio-sensors with proper technical features are facing the new approaches and challenges. Bio-sensing has entered the new stage for the implementation of concepts like IoT, big data etc. The Integration of artificial intelligence which includes the pattern analysis and classification algorithm along with the bio-sensors, that will bridge up the gap between the data accession and analysis for purpose of achieving the diagnostic and therapy. The wearable bio sensors will be available to the consumers and it is used as the indicators for the heartbeats and the

1| P a g e bio-physical activity. But they fail to detect the individual biophysical and biochemical marker. The Electrochemical analysis is simple in operation and it is most widely used in the glucose sensors. Different types of sensors are developed to monitor the sweat glucose while doing the exercise. For example: wearable wristband and patch-type sensors will help to monitor the presence of changes in the sweat glucose. The disposable strip model sensors which is analogous to the blood strip will be a simple monitoring of the sweat glucose. The patch model device will be modeled to analyze the sweat glucose which is shed for one day. The sweat glucose concentration of the daily profile from the patch model sensor will correspond from the commercial glucose.

Objects of the invention:

• The objective is to use non-invasive method in order to find the glucose level from the sweat without any break in the human body. • The Machine learning algorithms aims to predict the correct output with the help of sample dataset. • The Machine learning methods aims to train the system using algorithm, test the system how well the performance is done, and predicts the best accuracy output. • The sweat sensor aims to monitor the conductance of skin in order to obtain the voltage level. Based on this value, the glucose level of the person can be obtained. • The training part of machine learning will ensure about recognizing the patterns present in

the data. • The Testing part also make sures about the performance of machine regarding the

prediction of new output on the basis of training part. • The ADC is used to convert the analog output of sweat sensor to the digital form.

Summary of the invention:

The sweat glands will be distributed over the body and its response will be considered in rapid way in order to reflect the physiological symptoms in the body. Generally, the sweat will be secreted on the surface of the skin after completion of isotonic secretion from the sweat gland and reabsorption of sodium chloride via the reabsorptive duct. The monitoring sensor will measure the sweat glucose when the sweat is not absorbed. In body fluids, the sweat is most accessible fluid

21Page while comparing with others, and its main role is the thermoregulation. For sample use, the eccrine gland will excrete the sweat which is found all over the body and it is particularly concentrated in some of the locations. For example: hands, underarms, lower back and feet. The sensors will be closely placed to the skin in order to collect the samples of sweat and it can be processed in fast manner without any contamination. The two methods to monitor the blood glucose is that, invasive and non-invasive method. The invasive method will create the break in the skin in order to take blood. The non-invasive method does not create any break in the human body. The diabetic patient has to check the glucose regularly on daily basis. To check the glucose level, the patient has to take the drop of blood from the body. This will create the wounds in the human body. Some diabetic patient will result in taking insulin. This leads to painful task, so that, most of the patients will refuse to create the wounds in the human body. So instead, some patients will like to take medicines. The efforts are made to develop the wearable biosensors for non-invasive detection of the bio-markers in order to access the bio-fluids like sweat, tears, saliva, interstitial and wound fluid. Human sweat will carry the complex physiological information regarding the healthcare and disease. It also includes the constituents like water, ions, small molecules, small proteins and peptides. Bio-sweat sensor will capture the glucose, urea, chloride, lactate, alcohol, pH, creatinine and also the heavy metals. When comparing to other bio-fluid method like saliva, tears and interstitial fluid, the method of sweat has so many benefits in the wearable bio-sensors. The sweat consists of the water-soluble biomarker which is related to the human health. Then, the sweat will be monitored in continuous manner at the comfortable location because it is secreted in the body. The sweat can also be secreted in continuous way on-demanding via the exercise, iontophoresis and the stimulation of chemicals for the non-invasive wearable bio-sensing. The key challenges to be addressed such as lower sweat rate, evaporation, contamination from the skin, obtaining fresh sweat, its effects, sweat blood-correlation, lower concentration of the molecular detection, stability of sensor. The sweat has benefits comparing to other bio-fluids for the use of monitoring non invasive glucose. Th method is deployed in order to check the glucose level in the body by using non-invasive method. Sweat is considered as the non-invasive method for the purpose of monitoring the glucose level. The machine learning is the application and it helps to predict the diabetes based on the previous data. It is the mathematical model which is done on the basis of some sampling data. In this model, the training, algorithm, testing and prediction is done. The system should be trained first and later, the algorithms will be applied, and then, the testing will

31Page be done. At last, the prediction is done. The supervised learning is the method of machine learning, in that, the system will already know the expected output. The set of data will be trained to the system and it predicts the output on basis of trained data. In unsupervised learning, the trained machine will use the knowledge. They will vary as per the similarities, patterns and the differences. The reinforcement will take action according to the situation.

Detailed Description of the Invention:

The machine learning is used to predict the occurrence of diabetes. There are different classification methods to classify the occurrence of diabetes and the best method is to be chosen for the prediction. The database is needed and based on that; the predictions has to be done. There are many side effects in invasive method which will be harmful in monitoring blood glucose level. The Figure 1 shows the detailed method of analyzing glucose level using non-invasive method. The non-invasive method is helpful in monitoring the glucose level of the person without any harm to the patient. The sweat, saliva and breathe is useful in monitoring the blood glucose. In this method, the sweat is one of the concepts which is chosen for non-invasive method to monitor the glucose. The sensor id used to monitor the skin conductance which is used to reach the voltage level. On basis of obtained value, the glucose level of the person can be known. The skin response sensor is used for the monitoring in real-time. The creation of wound and using the blood drop is the painful method, where most of the patients are suffering. To avoid this method, the sweat sensor is used here to monitor the glucose level. The cloud is used to store the glucose level of the person. The details like date and time of values is also stored in the cloud. The database which is stored in the cloud will be helpful in future for the reference. The block diagram consists of the two categories, 1. Machine learning and 2. Real-time glucose and monitoring system. The database will be given as input to the machine learning. After pre-processing, the database will be considered as training and testing part. The training part will help to ensure that machine will identify the patterns which is present in the data. In order to ensure the accuracy and efficiency of the algorithm which is used for the purpose of training the machine, the cross-validation is done for data. Testing part will make sure of proper working process of the machine regarding the prediction performance of new results on the basis of training part. The machine learning methods are applied and accuracy is to be known. The technique with highest performance is used for the prediction purpose. In real time monitoring method, the sensor will make use to collect the reading

41Page from sweat. The Microcontroller is used to control the system. The power supply is also used to the controller. The sweat sensor will help to monitor the sweat. The output of sensor is the considered as skin conductance. The sensor will be in the form of analog. The Analog to Digital converter is used to convert the sensor output in the form of digital output. The salt level will be measured from the skin conductance whereas, the skin conductance is inversely proportional to the content of salt. The conductance is also converted to the voltage. The sweat sensor is inversely proportional to the glucose value level. Based on the obtained voltage value, the sugar level, stress level and the hydration level will help to monitor. The obtained voltage value will be converted to the sugar level. The glucose level which is collected from the real time monitoring will be later applied to the machine learning part for the reason of prediction. After prediction is done, the message will be obtained from the output and it will be indicated as the occurrence of diabetes. On basis of existing database, the prediction of diabetes is done. The best prediction is done from the output of glucose level which is calculated using sweat sensor. The algorithms like KNN, logistic regression, Random Forest, decision tree and Naive Bayes which is applied will help to classify the diabetes and makes the prediction and decision. In the real time part, the sensor will help to monitor the sweat. On basis of these sweat values, the glucose level can be found out.

51Page

Claims (5)

A TECHNIQUE TO ANALYSE GLUCOSE LEVELS USING BIO-SWEAT SENSOR Claim

1. The Non-invasive method is implemented to find the glucose level of the diabetic patient using the sweat sensor and this method is painless and cost-effective.

2. The Machine learning method is used to predict the occurrence of diabetes to take prior precaution. - The system will be trained first and later, the algorithms will be applied, and the output will be predicted based on the trained data. - The Test data is used to know the performance of system regarding prediction of the

output. - The data is pre-processed before splitting the data into training and test part.

3. The method claim, non-invasive method aims to monitor the glucose level whereas the invasive method for predicting glucose will create side effects.

4. The method claim2, the accuracy can be calculated and based on that, the prediction can be done.

5. The method claim2, the cloud storage is used to store the date and time of the values for the future reference.

1 Pag e

A TECHNIQUE TO ANALYSE GLUCOSE LEVELS USING BIO-SWEAT SENSOR 28 Oct 2020

Drawings: 2020103057

Figure 1: Block Diagram of proposed Method

1|Page