Scholarlink

<p>Trypanosomiasis is a parasitic disease caused by&#xa0;<i>Trypanosoma</i>&#xa0;species and can lead to severe cardiovascular damage if not diagnosed effectively. Traditional diagnosis relies on blood tests, often missing early-stage detection due to unmonitored cardiac patterns. The aim of this research is to classify Electrocardiogram (ECG) signals to determine the likelihood of cardiac involvement in patients with Trypanosomiasis. Although ECG signals are widely used to monitor cardiac activity, the presence of substantial noise complicates the extraction of clear and meaningful patterns. This research implements various filtering techniques, including Butterworth, Chebyshev Type I and II, and Moving Average filters, to reduce unwanted frequencies while preserving the essential information. Once the signals were filtered, different machine learning techniques were used to determine the presence of Trypanosomiasis and plot the feature importance. With a validation accuracy of 0.998, the Multilayer Perceptron (MLP) model utilizing a moving average filter with a window size of 30 was determined to be the optimal model, thus offering a data-driven and non-invasive way to support the early detection of cardiac involvement related to Trypanosomiasis.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

ECG-Based Disease Classification for Trypanosomiasis Using Signal Processing and Machine Learning Techniques

  • 期刊论文

  • 发表日期：

    2026年4月9日

查看全文
  • J. Judeson Antony Kovilpillai

摘要

Trypanosomiasis is a parasitic disease caused by Trypanosoma species and can lead to severe cardiovascular damage if not diagnosed effectively. Traditional diagnosis relies on blood tests, often missing early-stage detection due to unmonitored cardiac patterns. The aim of this research is to classify Electrocardiogram (ECG) signals to determine the likelihood of cardiac involvement in patients with Trypanosomiasis. Although ECG signals are widely used to monitor cardiac activity, the presence of substantial noise complicates the extraction of clear and meaningful patterns. This research implements various filtering techniques, including Butterworth, Chebyshev Type I and II, and Moving Average filters, to reduce unwanted frequencies while preserving the essential information. Once the signals were filtered, different machine learning techniques were used to determine the presence of Trypanosomiasis and plot the feature importance. With a validation accuracy of 0.998, the Multilayer Perceptron (MLP) model utilizing a moving average filter with a window size of 30 was determined to be the optimal model, thus offering a data-driven and non-invasive way to support the early detection of cardiac involvement related to Trypanosomiasis.