| Title |
Tekstiliniais elektrodais registruojamo EKG signalo kokybės vertinimas naudojant sprendimų medžius / |
| Translation of Title |
Signal quality determination of textile electrode based ECG using decision trees. |
| Authors |
Gasparovičius, Justinas |
| Full Text |
|
| Pages |
73 |
| Keywords [eng] |
Electrocardiogram quality ; decision trees ; machine learning ; smart textile |
| Abstract [eng] |
As the textile electrodes continue its integration into daily smart clothing market real time electrocardiographic signal quality assessment is becoming an increasingly relevant topic. Its relevance is also related with still unsolved problem of false alarms because of bad signal quality in clinical patient monitoring systems used today. In this research machine learning based decision tree algorithm is used to solve the problem of electrocardiogram quality assessment. Main benefits of this algorithm are: as small as possible computing resources compatible with low power embedded systems, efficiency to determine not qualitative episodes, good performance while detecting heart rate in episodes assessed as qualitative, compatibility with any existing monitoring hardware and ability to waive information provided by other sensors. Decision tree was trained and tested using recordings of six people subjects using textile electrodes, various simulated data and noisy records from “MIT-BIH“ database. Automatic signal quality annotation algorithm was proposed in this research which concluded in better performance for decision tree learning and a lot better efficiency compared to expert annotations. |
| Dissertation Institution |
Kauno technologijos universitetas. |
| Type |
Master thesis |
| Language |
Lithuanian |
| Publication date |
2016 |
