PARAMETERIZATION HEART RATE SIGNAL (TIMESERIES HEART RATE) USING R-DFA

• Main Author: WIJAYA - NIM: 33211306, RIFKI
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/30471
• Institution: Institut Teknologi Bandung

<p align="justify">Heart disease has long been the leading cause of death in the world. In 2008, 17.3 million people died of heart disease. 80% of deaths in the world are caused by heart disease. Currently there are many technologies that can identify heart disease. The current technology is to use an echocardiogram. This technology makes it possible to identify the heart right away. This technology is not yet portable so it does not allow for continuous identification every day and every time. There is a tool that can be used daily in the chest that can detect the heart with a heart rate per minute output. This tool is not like an echocardiogram that has excellent accuracy. This tool has the accuracy of one data per second and can be used for daily activities. The data generated by this tool will be uploaded to the server using android phone with pre-prepared apps. <br /> <br /> Heart Rate is generally used by health workers to check heart condition. Heart rate has a non-stationary nature, so it cannot be predicted or processed before it is processed first. The non-stationary process used is Detrended Fluctuation Analysis (DFA). Peng can use heart disease by inserting heart rate into the DFA. The heart rate that changes over time convert into a scaling index that can show the difference between healthy and sick. <br /> <br /> Toru Yazawa perfects Peng’s Scaling index by creating a new data retrieval with new time box. Toru Yazawa have successfully differentiated people with specific heart disease. <br /> <br /> R-DFA is new algorithm proposed by this research. Peng and Toru do not create algorithms for a very long data (100 data retrieval for one subject) and do not have the ability to determine sick or healthy tendencies. A case study of 9 patients was sampled every 15 minutes for 100 times at different times. Nine patients were taken in a state of sitting still with the same environment. The data taken is a preprocessing by eliminating the data that is not rational and cannot be processed. The data process to form the timeseries of the DFA's Peng scaling index. The Timeseries process with substraction of the scalling index then regression. R-DFA can classify the three classes with 92% accuracy with significant ( p<0,05 ). This method produces a threshold used to determine the three classes. <br /> <br /> R-DFA has generated a threshold used to determine the three classes. Based on the data of 9 subjects, the R-DFA method has a classification tendency with 3 classes (Health, Mild Left Ventricle Diastolic Dysfunction, and Hypertension Heart Disease) tested significantly with p <0.005. Trend of healthy class with gradient (8 <Healthy <8 and healthy &#8805;-18), MLVDD class (-18 <MLVDD <-8) and HHD class (HHD&#8805; 8). <br /> <br /> The advantages of R-DFA compared to other is able to produce trend of HHD disease and MLVDD. With the trend is expected to change the pattern of life if there is a risk trend of HHD and MLVDD. Everyone can identify themselves before there is a greater risk of illness. Everyone can also come to the doctor early before the risk of disease increases. <p align="justify">