•

•

•

Published October 27, 2024 © MIT

Leveraging ECG for Advanced Personal Identification

This project explores the potential of the Hexabitz EXG Module for secure ECG-based identification.

AdvancedWork in progressOver 33 days198

Leveraging ECG for Advanced Personal Identification

Things used in this project

Hardware components

Hexabitz 4-Pin USB-Serial Prototype Cable

Hexabitz BitzClamp

Hexabitz STLINK-V3MODS Programmer (H40Rx)

Hexabitz Single-Lead, EXG Monitor (H2BR0x)

Hexabitz Sensor Cable - Electrode Pads (3 Connector)

Hexabitz 3.3V/1A DC-DC Power Supply Module (H03R00)

Software apps and online services

STMicroelectronics STM32CUBEPROG

MATLAB

Story

#️⃣ This project explores the potential of Hexabitz EXG Module in ECG acquisition and processing, which is unique to each individual, as a reliable and secure means of identification #️⃣

📝 Introduction:

In the modern era of biometric security, traditional methods such as fingerprints and facial recognition are being complemented by innovative approaches. One such promising method is the use of Electrocardiogram (ECG) signals for personal recognition. This project delves into the potential of ECG signals, which are inherently unique to each individual, as a reliable and secure means of identification. By harnessing the distinct patterns of ECG signals, we aim to explore a novel and robust approach to biometric security.

Research in biometrics has contributed with many approaches to implement ECG as biometrics. All these approaches can be classified into three groups: fiducial, non-fiducial, and hybrid (Odinaka et al., 2012). A fiducial approach extracts specific biometric features such as valley to peaks distances, amplitudes, and angles. In general, it detects a specific feature of the signal and performs a biometric match based on the correlation of these features. Non-fiducial approaches extract features based on the whole signal and not a specific characteristic. Wavelets or autocorrelation coefficients are common approaches in this group. Hybrid approaches use a combination of the previous two.

🔍 Project Des cription:

1. Objective: To develop a system that utilizes ECG signals for accurate and secure people recognition.

2. Methodology: The project involves collecting ECG data from participants, preprocessing the signals, and applying AI algorithms to identify unique patterns.

3. Expected Outcomes: The system aims to achieve high accuracy in recognizing individuals based on their ECG signals, demonstrating the feasibility of this biometric method in real-world applications. Additionally, due to the inherent uniqueness and complexity of ECG signals, this method offers a high level of security and is resistant to forgery.

ECG biometric modules

🛠️ Tools:

🌟 Single-Lead, EXG Monitor (H2BR0x):

Hexabitz Single-Lead, EXG Monitor Module (H2BR0) is one of a kind module that can record publication-grade biopotential signals from your body be it from the heart (ECG), brain (EEG), eyes (EOG), and muscles (EMG).

What makes it different?

Record publication-quality biopotential signals like ECG, EMG, EOG, or EEG.
Small size allows easy integration into mobile and space-constrained projects.
Notch filter (second order) to remove 50 Hz AC mains.
H2BR0 is based on STM32G0 MCU.
Program advanced C code with our easy-to-use APIs.
You can connect to external hardware or combine with other Hexabitz modules!
The module is equipped with an open source MATLAB interface.

1 / 4 • H2BR0 Module

🩺 Sensor Cable – Electrode Pads (3 Connector):

This is a simple three conductor sensor cable with electrode pad leads. The cable is 24˝ long and feature a 3.5mm audio jack connector on one end with snap style receptacles for biomedical sensor pads.

1 / 3 • Sensor Cable

🌟 STLINK-V3MODS Programmer (H40Rx):

H40Rx is a programmer module which contains STLINK-V3MODS stand-alone debugging and programming mini probe for STM32 micro-controllers (Hexabitz modules and other MCUs).

It supports the SWD (Serial Wire Debugging) interface for the communication with any STM32 micro-controller located on an application board.

It also provides bridge interfaces to several communication protocols, allowing for instance the programming of the target through boot-loader.

🫀 How We build it:

Step 1 🖤 Writing codes with STM32CubeIDE software:

Check out this article for writing code with STM32CubeIDE.
We Initialize the signal type to be measured.

EXG_Init API

Then, we used "PlotToTerminal" API, which sends ECG samples through port 3 and display both unfiltered and filtered samples on the Terminal or plot signal ECG,

PlotToTerminal API

Extract the heart rate from the ECG signal

Monitor the status of electrodes

Project Flowchart

Step 2 🖤 Experiment of Building a Model for Classifying Individuals Based on ECG Signals Using a Feedforward Neural Network and MATLAB:

#️⃣ A Feedforward Neural Network is a basic type of artificial neural network characterized by transmitting information in only one direction through successive layers.

Feedforward Neural Network

#️⃣ Advantages of Feedforward Network:

1. Ease of Learning: The Feedforward network is simple in structure and easy to train. It can be trained using an ECG dataset and analyze it to learn the distinctive features of each individual's ECG signals.

2. High Generalization Ability: A trained network can generalize to new data from the same individuals or similar others, meaning it can accurately predict the identity of a new person.

3. Ability to Handle Non-linear Data: The network uses non-linear activation functions, allowing it to learn complex and non-linear relationships between ECG features and identify the person.

4. Noise Resistance: The Feedforward network can reduce the impact of noise or distortions in ECG signals, improving the accuracy of person recognition in non-ideal environments.

5. Scalability: Additional hidden layers and nodes can be added to the network to improve its performance as data complexity increases.

6. Suitability for Temporal Data: The Feedforward network is suitable for handling ECG data, which is temporal data, enabling the network to learn changes over time and relationships between time points in the signal.

Different Activation Functions have various properties and effects on the learning process. The choice of activation function depends on the type of problem, its complexity, data distribution, and required performance.
The default activation function in MATLAB for this network and used in our current training is `tanh`. This function maps values between -1 and 1 and is centered around 0, which can make learning easier for the next layer.

#️⃣ Building the Database for the Person Classification and Recognition Project Using ECG Signals Requires Several Basic Steps:

1. Data Collection: Collect ECG data for several individuals.

2. Preprocessing the Data: Process the data to remove noise and improve signal quality (this step is included in the module itself).

3. Labeling: Each ECG signal needs to be labeled with the appropriate person's identifier. This step is crucial for the subsequent training of the model.

#️⃣ Mechanism of the Electrode Status Check Button:

When the button is pressed in the MATLAB interface, a message is sent to the module to check the status of the electrodes, and the module responds by sending one of these values, from which we know the status of the electrodes:

Lead P_CONNECTED_Lead N_CONNECTED = 0
Lead P_CONNECTED_Lead N_NOTCONNECTED = 1
Lead P_NOTCONNECTED_Lead N_CONNECTED = 2
Lead P_NOTCONNECTED_Lead N_NOTCONNECTED = 3

#️⃣ Code Analysis:

1. The code relies on serial communication with the EXG unit.

2. The code reads raw and filtered data from the EXG unit.

3. The data is plotted on a graph for display and analysis.

Step 3 🖤 Tests Conducted in MATLAB Environment:

1. Data Loading: We started by loading ECG data from three different Excel files, each representing a different individual. We used the `xlsread()` function to read the data from each file and store it in matrices.

1 / 3

2. Feature Extraction: These features help in identifying the unique characteristics of each person's ECG signals. The features include:

ECG Features from fiducial points

Feature Extraction

Peaks: Using the `findpeaks()` function to identify the major peaks in the ECG signal.

Peaks in the ECG signal

Mean
Standard Deviation

SD & Mean

Gradient: Calculating the absolute difference between successive R-wave peak values.
Frequency Analysis: Converting R-wave peak data to the frequency domain using Fast Fourier Transform (FFT) and identifying the highest frequency (`freq1`) in the frequency domain with fs=120Hz.

Fast Fourier Transform (FFT)

Heart Rate (HR): Calculating the heart rate by measuring the time interval between successive R-wave peaks.

1 / 2 • Heart Rate (HR)

3. Storing Feature Values: Storing the peak, mean, and standard deviation values for all individuals in a matrix called `Inputs`.

4. Creating Target Matrix: The target for the recognition process is represented in a matrix called `targets`, where each column represents a different individual.

5. Neural Network Setup:

Network Type: A feedforward neural network was used.
Network Structure: Consisting of three hidden layers with 40, 30, and 20 neurons respectively.
Training Algorithm: The Levenberg-Marquardt algorithm was used to supervise the network training.
Importance of Algorithm: This algorithm helps in determining the best set of weights that minimize prediction error.
Network Importance:The neural network is used to recognize individuals based on the extracted features.

6. Data Splitting: The ECG data was split into training, validation, and testing sets with proportions of 70%, 20%, and 20% respectively.

7. Learning Rate Adjustment: The learning rate was set to 0.05.

8. Network Training: The neural network was trained using the `train` function with input and target data.

9. Error Calculation: Using the `gsubtract` function to subtract the values of the target and output matrices.

10. Performance Evaluation: The network's performance was evaluated using the validation and testing data sets.

11. Matrix Conversion: Converting the target and output matrices from matrix format to a comparable indexed image format, facilitating the evaluation of the neural network's performance and comparison with target values.

12. Evaluation Results:

Confusion Matrix: Used to display the prediction accuracy of the neural network in recognizing individuals.
Training Performance: Displaying the network's performance improvement with each training epoch.
Overall Evaluation: Helps in understanding the efficiency of the person recognition system.
Target and Output Relationship: Displaying the network structure using `view(net)`.

1 / 4

13. Testing and Recognition: This phase involves testing and recognition, where a new signal is obtained, the same features applied during training are used, and it is determined whether the new input signal matches the training, leading to recognition. The result is displayed by comparing the resulting signal with the remaining signals to show the match and then displaying the name of the person associated with the signal.

The Project GUI

💻Test the System:

Examples of graph 📈🫀🔍

🧩Challenges Faced:

1. Building the Database: One of the main challenges was constructing the database, particularly in selecting features. The ECG signals of healthy individuals tend to be similar in terms of features and timing, making it difficult to distinguish between them. In contrast, the signals from patients are more distinct, allowing for easier feature extraction.

2. Noise Issues: Various types of noise posed significant challenges, including cable length, surrounding environment, and electrical fields.

📈 Future Prospects:

🔴 Image Conversion and Classification: ECG signals can be converted into images, allowing the application of image processing algorithms for classification.

🔴 Audio Signal Transformation: The signals can be transformed into audio files through frequency domains, enabling unique audio-based identification.

🔴 Real-Time Recognition Model: A model can be designed using Python on Colab, which can be embedded within an STM32 microcontroller. This allows for real-time, immediate person recognition without the need for extensive training and testing.

Please feel free to leave a comment here if you have any questions or concerns regarding this project 😃💡

💫 References:

[1] https://link.springer.com/chapter/10.1007/978-3-030-81982-8_4

[2] https://www.mathworks.com

[3] https://onlinelibrary.wiley.com/doi/10.1155/2022/1238864

[4] https://hexabitz.com/modules/

[5] https://hexabitz.com/wiki/

Code

/*
 BitzOS (BOS) V0.3.2 - Copyright (C) 2017-2024 Hexabitz
 All rights reserved

 File Name     : main.c
 Description   : Main program body.
 */
/* Includes ------------------------------------------------------------------*/
#include "BOS.h"

/* Private variables ---------------------------------------------------------*/

/* Private function prototypes -----------------------------------------------*/

/* Main function ------------------------------------------------------------*/

int main(void){

	Module_Init();		//Initialize Module &  BitzOS

	//Don't place your code here.
	for(;;){}
}

/*-----------------------------------------------------------*/

/* User Task */
void UserTask(void *argument){

EXG_Init(ECG);
	

	// put your code here, to run repeatedly.
	while(1){


		

PlotToTerminal(P3,ECG);
		


			// 	  }

	}
}

/*-----------------------------------------------------------*/

function varargout = EXG_Module(varargin)
% EXG_MODULE MATLAB code for EXG_Module.fig
%      EXG_MODULE, by itself, creates a new EXG_MODULE or raises the existing
%      singleton*.
%
%      H = EXG_MODULE returns the handle to a new EXG_MODULE or the handle to
%      the existing singleton*.
%
%      EXG_MODULE('CALLBACK',hObject,eventData,handles,...) calls the local
%      function named CALLBACK in EXG_MODULE.M with the given input arguments.
%
%      EXG_MODULE('Property','Value',...) creates a new EXG_MODULE or raises the
%      existing singleton*.  Starting from the left, property value pairs are
%      applied to the GUI before EXG_Module_OpeningFcn gets called.  An
%      unrecognized property name or invalid value makes property application
%      stop.  All inputs are passed to EXG_Module_OpeningFcn via varargin.
%
%      *See GUI Options on GUIDE's Tools menu.  Choose "GUI allows only one
%      instance to run (singleton)".
%
% See also: GUIDE, GUIDATA, GUIHANDLES

% Edit the above text to modify the response to help EXG_Module

% Last Modified by GUIDE v2.5 26-May-2024 22:36:32

% Begin initialization code - DO NOT EDIT
gui_Singleton = 1;
gui_State = struct('gui_Name',       mfilename, ...
                   'gui_Singleton',  gui_Singleton, ...
                   'gui_OpeningFcn', @EXG_Module_OpeningFcn, ...
                   'gui_OutputFcn',  @EXG_Module_OutputFcn, ...
                   'gui_LayoutFcn',  [] , ...
                   'gui_Callback',   []);
if nargin && ischar(varargin{1})
    gui_State.gui_Callback = str2func(varargin{1});
end

if nargout
    [varargout{1:nargout}] = gui_mainfcn(gui_State, varargin{:});
else
    gui_mainfcn(gui_State, varargin{:});
end
% End initialization code - DO NOT EDIT


% --- Executes just before EXG_Module is made visible.
function EXG_Module_OpeningFcn(hObject, eventdata, handles, varargin)
% This function has no output args, see OutputFcn.
% hObject    handle to figure
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)
% varargin   command line arguments to EXG_Module (see VARARGIN)

% Choose default command line output for EXG_Module
handles.output = hObject;

% Update handles structure
guidata(hObject, handles);
global Port index filteredECG
filteredECG=[];
index = 0;
channel='COM7';
Port=serial(channel,'BaudRate',115200,'Timeout',5); 


% UIWAIT makes EXG_Module wait for user response (see UIRESUME)
% uiwait(handles.figure1);


% --- Outputs from this function are returned to the command line.
function varargout = EXG_Module_OutputFcn(hObject, eventdata, handles) 
% varargout  cell array for returning output args (see VARARGOUT);
% hObject    handle to figure
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)

% Get default command line output from handles structure
varargout{1} = handles.output;


% --- Executes on button press in pushbutton1.
function pushbutton1_Callback(hObject, eventdata, handles)
global Port filteredECG index
fopen(Port);
fs=120;
RecordTime=4; %sec
k=fs*RecordTime; % k Number of samples
ECG=[];
filteredECG=[];
t=0;
fprintf(Port,'a00s');   % start
for i=1:k
    tic;
     x=fscanf(Port); % size of message 
    if( isempty(x) )
        fclose(Port);
    end
    
    if ( x(1)=='a' && x(end)==10) % '\n' = 10
        delimiter_b_index = find(x=='b');
        ECG(i)= str2double(x(2 : delimiter_b_index -1) );  % mesage (a...b...\n)
        filteredECG(i)= str2double(x(delimiter_b_index + 1 :end - 1) );
    end
    t=toc+t;   
end
fprintf(Port,'a00f');  %stop
fclose(Port);
axes(handles.axes1);

plot(ECG(5:end))
grid on
axes(handles.axes2);

plot(filteredECG(5:end))
grid on
index = index +1;
fopen(Port);
fprintf(Port,'a00d');   % ask for HR
x=fscanf(Port); % size of message 
if( isempty(x) )
   fclose(Port);
else
    if ( x(1)=='a' && x(end)==10) % '\n' = 10
    HR = x(2 :end - 1);
%     HR = str2num(EOG_state);
    end
    set(handles.edit2,'string',HR);
    fclose(Port); 
end
    
  
function edit1_Callback(hObject, eventdata, handles)
% hObject    handle to edit1 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)

% Hints: get(hObject,'String') returns contents of edit1 as text
%        str2double(get(hObject,'String')) returns contents of edit1 as a double


% --- Executes during object creation, after setting all properties.
function edit1_CreateFcn(hObject, eventdata, handles)
% hObject    handle to edit1 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    empty - handles not created until after all CreateFcns called

% Hint: edit controls usually have a white background on Windows.
%       See ISPC and COMPUTER.
if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor'))
    set(hObject,'BackgroundColor','white');
end



function edit2_Callback(hObject, eventdata, handles)
% hObject    handle to edit2 (see GCBO)

% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)

% Hints: get(hObject,'String') returns contents of edit2 as text
%        str2double(get(hObject,'String')) returns contents of edit2 as a double


% --- Executes during object creation, after setting all properties.
function edit2_CreateFcn(hObject, eventdata, handles)
% hObject    handle to edit2 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    empty - handles not created until after all CreateFcns called

% Hint: edit controls usually have a white background on Windows.
%       See ISPC and COMPUTER.
if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor'))
    set(hObject,'BackgroundColor','white');
end


% --- Executes on button press in pushbutton3.
function pushbutton3_Callback(hObject, eventdata, handles)
global Port
fopen(Port);
fprintf(Port,'a00w');   % ask for wires states
x=fscanf(Port); % size of message 
if( isempty(x) )
   fclose(Port);
else
    if ( x(1)=='a' && x(end)==10) % '\n' = 10
    wire_state = str2double(x(2 : end -1));  % mesage (a...\n)
    end
    fclose(Port); 
    if (wire_state==0)
        set(handles.edit1,'string','LeadP_CONNECTED_LeadN_CONNECTED');
    elseif (wire_state==1)
        set(handles.edit1,'string','LeadP_CONNECTED_LeadN_NOTCONNECTED');
    elseif (wire_state==2)
        set(handles.edit1,'string','LeadP_NOTCONNECTED_LeadN_CONNECTED');
    elseif (wire_state==3)
            set(handles.edit1,'string','LeadP_NOTCONNECTED_LeadN_NOTCONNECTED');
    end
end
    

        
% hObject    handle to pushbutton3 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)



function edit3_Callback(hObject, eventdata, handles)
% hObject    handle to edit3 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)

% Hints: get(hObject,'String') returns contents of edit3 as text
%        str2double(get(hObject,'String')) returns contents of edit3 as a double


% --- Executes during object creation, after setting all properties.
function edit3_CreateFcn(hObject, eventdata, handles)
% hObject    handle to edit3 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    empty - handles not created until after all CreateFcns called

% Hint: edit controls usually have a white background on Windows.
%       See ISPC and COMPUTER.
if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor'))
    set(hObject,'BackgroundColor','white');
end



function edit4_Callback(hObject, eventdata, handles)
% hObject    handle to edit4 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)

% Hints: get(hObject,'String') returns contents of edit4 as text
%        str2double(get(hObject,'String')) returns contents of edit4 as a double


% --- Executes during object creation, after setting all properties.
function edit4_CreateFcn(hObject, eventdata, handles)
% hObject    handle to edit4 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    empty - handles not created until after all CreateFcns called

% Hint: edit controls usually have a white background on Windows.
%       See ISPC and COMPUTER.
if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor'))
    set(hObject,'BackgroundColor','white');
end


% --- Executes when user attempts to close figure1.
function figure1_CloseRequestFcn(hObject, eventdata, handles)
% hObject    handle to figure1 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)
% Hint: delete(hObject) closes the figure
delete(hObject);


% --- Executes on button press in pushbutton4.
function pushbutton4_Callback(hObject, eventdata, handles)
% hObject    handle to pushbutton4 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)
global filteredECG index
ECGsignal=dataset(filteredECG');
export(ECGsignal,'XLSFile',['ECGsignal' num2str(index)]);


% --- Executes on button press in pushbutton5.
function pushbutton5_Callback(hObject, eventdata, handles)
% hObject    handle to pushbutton5 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)
data1 = xlsread('ECGsignal1.xls');
data2 = xlsread('ECGsignal2.xls');
data3 = xlsread('ECGsignal3.xls');
data4 = xlsread('ECGsignal4.xls');
data5 = xlsread('ECGsignal5.xls');
[peaks1,locs1]=findpeaks(data1,'MinpeakHeight',1,'MinpeakDistance',60);
[peaks2,locs2]=findpeaks(data2,'MinpeakHeight',1,'MinpeakDistance',60);
[peaks3,locs3]=findpeaks(data3,'MinpeakHeight',1,'MinpeakDistance',60);
[peaks4,locs4]=findpeaks(data4,'MinpeakHeight',1,'MinpeakDistance',60);
[peaks5,locs5]=findpeaks(data5,'MinpeakHeight',1,'MinpeakDistance',60);
Data=[peaks1; peaks2; peaks3; peaks4; peaks5];
Data=Data(1:25);
%Data=Data'
% --- Executes on selection change in delete.
mynet=handles.mynet;
INPUTS=[Data;Data;Data];
INPUTS=INPUTS';
out=sim(mynet,INPUTS);
out=out';
sum(out)

% --- Executes on button press in delete.
function delete_Callback(hObject, eventdata, handles)
i=0:20;
 filei='E:\k\ECGsignal1.xls';
delete(filei);



% hObject    handle to delete (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)


% --- Executes on key press with focus on pushbutton5 and none of its controls.
function pushbutton5_KeyPressFcn(hObject, eventdata, handles)
% hObject    handle to pushbutton5 (see GCBO)
% eventdata  structure with the following fields (see MATLAB.UI.CONTROL.UICONTROL)
%	Key: name of the key that was pressed, in lower case
%	Character: character interpretation of the key(s) that was pressed
%	Modifier: name(s) of the modifier key(s) (i.e., control, shift) pressed
% handles    structure with handles and user data (see GUIDATA)
data1 = xlsread('ECGsignal1.xls');
data2 = xlsread('ECGsignal2.xls');
data3 = xlsread('ECGsignal3.xls');
data4 = xlsread('ECGsignal4.xls');
data5 = xlsread('ECGsignal5.xls');
[peaks1,locs1]=findpeaks(data1,'MinpeakHeight',1,'MinpeakDistance',60);
[peaks2,locs2]=findpeaks(data2,'MinpeakHeight',1,'MinpeakDistance',60);
[peaks3,locs3]=findpeaks(data3,'MinpeakHeight',1,'MinpeakDistance',60);
[peaks4,locs4]=findpeaks(data4,'MinpeakHeight',1,'MinpeakDistance',60);
[peaks5,locs5]=findpeaks(data5,'MinpeakHeight',1,'MinpeakDistance',60);
Data=[peaks1; peaks2; peaks3; peaks4; peaks5];



% --- Executes on button press in pushbutton7.
function pushbutton7_Callback(hObject, eventdata, handles)
% hObject    handle to pushbutton7 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)
data1 = xlsread('newreem1.xlsx');
data2 = xlsread('newruba1.xlsx');
data3 = xlsread('newahmad1.xlsx');
inputs = [data1; data2; data3]';
targets = [repmat([1; 0; 0], 1, 25), repmat([0; 1; 0], 1, 25), repmat([0; 0; 1], 1, 25)];
hiddenLayerSize = [20 40 60];
net =feedforwardnet(hiddenLayerSize);
net.divideParam.trainRatio =0.6;
net.divideParam.valRatio = 0.2;
net.divideParam.testRatio = 0.2;
net.trainparam.lr=0.01;
net.trainFcn='trainlm';
[net,tr]=train(net,inputs,targets);
outputs = net(inputs);
errors = gsubtract(outputs,targets);
performance = perform(net,targets,outputs);
p = vec2ind(outputs);
p1 = vec2ind(targets);
%plotconfusion (targets,outputs)
figure, plotperform(tr)
view(net)
handles.mynet=net;
guidata(hObject,handles);



function edit5_Callback(hObject, eventdata, handles)
% hObject    handle to edit5 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)
data1 = xlsread('newreem1.xlsx');% peaks extracted from raw data
data2 = xlsread('newruba1.xlsx');% peaks extracted from raw data
data3 = xlsread('newahmad1.xlsx');% peaks extracted from raw data
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%standard diviation & mean value extractes also from raw data:
data11 = xlsread('reem1.xls');
data12 = xlsread('reem2.xls');
data13 = xlsread('reem3.xls');
data14 = xlsread('reem4.xls');
data15 = xlsread('reem5.xls');
ecgg=[data11;data12;data13;data14;data15];
ecgg1=(ecgg-min(ecgg));
ecgg1=ecgg1/max(ecgg1);
m=mean(ecgg1);
s = std(ecgg1);
data1(26)=m;
data1(27)=s;
%%%%%% ruba
data11 = xlsread('ruba1.xls');
data12 = xlsread('ruba2.xls');
data13 = xlsread('ruba3.xls');
data14 = xlsread('ruba4.xls');
data15 = xlsread('ruba5.xls');
ecgg=[data11;data12;data13;data14;data15];
ecgg1=(ecgg-min(ecgg));
ecgg1=ecgg1/max(ecgg1);
m=mean(ecgg1);
s = std(ecgg1);
data2(26)=m;
data2(27)=s;
%%%%%% Ahmad:
data11 = xlsread('ruba1.xls');
data12 = xlsread('ruba2.xls');
data13 = xlsread('ruba3.xls');
data14 = xlsread('ruba4.xls');
data15 = xlsread('ruba5.xls');
ecgg=[data11;data12;data13;data14;data15];
ecgg1=(ecgg-min(ecgg));
ecgg1=ecgg1/max(ecgg1);
m=mean(ecgg1);
s = std(ecgg1);
data3(26)=m;
data3(27)=s;
inputs = [data1; data2; data3]'; 
targets = [repmat([1; 0; 0], 1, 27), repmat([0; 1; 0], 1, 27), repmat([0; 0; 1], 1, 27)];
hiddenLayerSize = [20 40 60];
net =feedforwardnet(hiddenLayerSize);
net.divideParam.trainRatio =0.6;
net.divideParam.valRatio = 0.2;
net.divideParam.testRatio = 0.2;
net.trainparam.lr=0.01;
net.trainFcn='trainlm';
[net,tr]=train(net,inputs,targets);
outputs = net(inputs);
errors = gsubtract(outputs,targets);
performance = perform(net,targets,outputs);
p = vec2ind(outputs);
p1 = vec2ind(targets);
%plotconfusion (targets,outputs)
figure, plotperform(tr)
view(net)
handles.mynet=net;
guidata(hObject,handles);
% Hints: get(hObject,'String') returns contents of edit5 as text
%        str2double(get(hObject,'String')) returns contents of edit5 as a double


% --- Executes during object creation, after setting all properties.
function edit5_CreateFcn(hObject, eventdata, handles)
% hObject    handle to edit5 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    empty - handles not created until after all CreateFcns called

% Hint: edit controls usually have a white background on Windows.
%       See ISPC and COMPUTER.
if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor'))
    set(hObject,'BackgroundColor','white');
end


function edit6_CreateFcn(hObject, eventdata, handles)
% hObject    handle to edit6 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    empty - handles not created until after all CreateFcns called


% Hint: edit controls usually have a white background on Windows.
%       See ISPC and COMPUTER.
if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor'))
    set(hObject,'BackgroundColor','white');
end



function edit7_Callback(hObject, eventdata, handles)
% hObject    handle to edit7 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    structure with handles and user data (see GUIDATA)
data1 = xlsread('ECGsignal1.xls');
data2 = xlsread('ECGsignal2.xls');
data3 = xlsread('ECGsignal3.xls');
data4 = xlsread('ECGsignal4.xls');
data5 = xlsread('ECGsignal5.xls');
[peaks1,locs1]=findpeaks(data1,'MinpeakHeight',1,'MinpeakDistance',60);
[peaks2,locs2]=findpeaks(data2,'MinpeakHeight',1,'MinpeakDistance',60);
[peaks3,locs3]=findpeaks(data3,'MinpeakHeight',1,'MinpeakDistance',60);
[peaks4,locs4]=findpeaks(data4,'MinpeakHeight',1,'MinpeakDistance',60);
[peaks5,locs5]=findpeaks(data5,'MinpeakHeight',1,'MinpeakDistance',60);
Data=[peaks1; peaks2; peaks3; peaks4; peaks5];
Data=Data(1:25);
%standard diviation & mean value extractes also from raw data:
data11 = xlsread('ECGsignal1.xls');
data12 = xlsread('ECGsignal2.xls');
data13 = xlsread('ECGsignal3.xls');
data14 = xlsread('ECGsignal4.xls');
data15 = xlsread('ECGsignal5.xls');
ecgg=[data11;data12;data13;data14;data15];
ecgg1=(ecgg-min(ecgg));
ecgg1=ecgg1/max(ecgg1);
m=mean(ecgg1);
s = std(ecgg1);
Data(26)=m;
Data(27)=s;
%Data=Data'
% --- Executes on selection change in delete.
mynet=handles.mynet;
INPUTS=[Data;Data;Data];
INPUTS=INPUTS';
out=sim(mynet,INPUTS);
out=out';
sum(out)
% Hints: get(hObject,'String') returns contents of edit7 as text
%        str2double(get(hObject,'String')) returns contents of edit7 as a double


% --- Executes during object creation, after setting all properties.
function edit7_CreateFcn(hObject, eventdata, handles)
% hObject    handle to edit7 (see GCBO)
% eventdata  reserved - to be defined in a future version of MATLAB
% handles    empty - handles not created until after all CreateFcns called

% Hint: edit controls usually have a white background on Windows.
%       See ISPC and COMPUTER.
if ispc && isequal(get(hObject,'BackgroundColor'), get(0,'defaultUicontrolBackgroundColor'))
    set(hObject,'BackgroundColor','white');
end