Deteksi Intensitas Suara Batuk pasien Infeksi Saluran Pernafasan Akut (ISPA) Menggunakan Edge Impulse Machine Learning berbasis Model Mel Frequency Cepstral Coefficients (MFCC)

Aris Widodo, Muhamad Azwar Annas

Sari


Batuk merupakan salah satu indikator kondisi pasien pengidap penyakit infeksi saluran pernafasan akut (ISPA). Teknologi terkini memiliki banyak metode untuk mendeteksi batuk diantaranya analisis gelombang suara batuk langsung, penggunaan Frequency-Modulated Continuous Wave radar (FMCW) atau jaringan saraf tepi konvolusi dan sebagainya sebagai acuan deteksi suara batuk namun masih belum pada tingkatan pengukuran beruntun dalam bentuk intensitas deteksi batuk tiap waktu. Pada penelitian ini telah dilakukan uji coba alternatif deteksi intensitas menggunakan Mel Frequency Cepstral Coefficients (MFCC) pada platform Edge Impulse untuk mengetahui nilai akurasi deteksi intensitas batuk ISPA. Penelitian ini dilakukan dengan membuat dataset batuk ISPA, membuat pemodelan MFCC pada design Impulse dan pengembangan library mikrokontroler. Library ini diunggah pada mikrokontroler untuk dilakukan uji langsung deteksi batuk beruntun dengan variasi tanpa jeda, jeda 5 detik dan 10 detik dari kompilasi 50 suara batuk. Hasil deteksi diakumulasi dengan nilai confidence level di atas 50% dianggap sebagai batuk dan dihitung nilai akurasi dari rasio jumlah batuk yang terukur. Pada penelitian ini dihasilkan akurasi pengukuran suara batuk tanpa jeda, jeda 5 detik dan 10 detik sebesar 18%, 34% dan 62%.

 

Cough is an indicator of the condition of patients with acute respiratory infections (ARI). Latest technology has many methods for detecting cough, such as analysis of direct cough sound waves, use of frequency-modulated continuous wave radar (FMCW), convolutional peripheral nerve networks, etc., as a reference for cough detection, but still not at the continuous measurement level in the form of cough detection intensity each time. In this study, an alternative intensity detection test will be tested using the Mel Frequency Cepstral Coefficients (MFCC) on the Edge Impulse platform to determine the accuracy of the intensity detection of ARI cough intensity. This research was carried out by creating an ISPA cough dataset, doing MFCC modeling on the Impulse design, and developing a microcontroller library. This library is uploaded to the microcontroller for a direct test of continuous cough detection with variations without pause of 5 seconds and 10 seconds from a compilation of 50 coughing sounds. The detection results accumulated a confidence level value above 50%, which was considered a cough, and the accuracy value was calculated from the ratio of the number of coughs measured. In this study, the accuracy of cough sound measurement without pauses, pause of 5 seconds, and 10 seconds was 18%, 34%, and 62%, respectively.


Kata Kunci


Akurasi; Batuk; Deteksi; Edge Impulse, MFCC, Machine Learning

Teks Lengkap:

PDF


Dilihat:
Sari 64 kali
PDF 11 kali

Referensi


P. Setiawan, “Risk factors causing Acute Respiratory Infection (ARI) in toddlers in Indonesia: A literature review,” World Journal of Adanced Research and Reviews, vol. 18, no. 3, pp. 1556–1559, 2023, doi: 10.30574/wjarr.2023.18.3.1266.

L. Lutpiatina, L. Sulistyorini, H. B. Notobroto, R. P. Raya, R. D. Utama, and A. Thuraidah, “Multilevel Analysis of Lifestyle and Household Environment for Toddlers with Symptoms of Acute Respiratory Infection (ARI) in Indonesia in 2007, 2012, and 2017,” Global Pediatric Health, vol. 9, p. 2333794X221078700, 2022, doi: 10.1177/2333794X221078700.

N. Dinda, L. Sulityorini, and I. Rahmawati, “The Overview of the Nurse’s Discharge Planning Implementation in Children with ARI in Indonesia: Literature Review,” Nursing and Health Sciences Journal, vol. 2, no. 1, Art. no. 1, 2022, doi: 10.53713/nhs.v2i2.91.

Pradana Ratnasari, N. (2020). ARI (1,1) Model for Predicting Covid19 in Indonesia. Indonesian Journal of Life Sciences, vol. 2 no. 2, pp. 83-88. https://doi.org/https://doi.org/10.54250/ijls.v2i2.51

F. Tumiwa, Finni Fitria Tumiwa, Angelia Pondaag, Grace I. V. Watung, and Ake R. C. Langingi, “Edukasi Bagi Masyarakat Desa Mopuya Tentang Pencegahan Tuberkulosis”, Jurnal Pengabdian kepada Masyarakat Nusantara, vol. 3, no. 2, pp. 1157-1163, 2022, doi: https://doi.org/10.55338/jpkmn.v3i2.1%20Desember.

F. Mulya, “Analisis Program Penanggulangan TBC di Indonesia dalam Upaya Pencapaian Target Eliminasi TBC Tahun 2030.” Accessed: Oct. 29, 2023. [Online]. Available: https://www.researchgate.net/publication/366876908_Analisis_Program_Penanggulangan_TBC_di_Indonesia_dalam_Upaya_Pencapaian_Target_Eliminasi_TBC_Tahun_2030

KEMENKES, “Melalui Kegiatan INA–TIME 2022 Ke-4, Menkes Budi Minta 90% Penderita TBC Dapat Terdeteksi di Tahun 2024 – P2P Kemenkes RI.” Accessed: Apr. 05, 2023. [Online]. Available: http://p2p.kemkes.go.id/melalui-ina-time-2022-ke-4-menkes-budi-minta-90-penderita-tbc-dapat-terdeteksi-di-tahun-2024/

K. Han and S. Hong, “Cough Detection Using Millimeter-Wave Fmcw Radar,” in ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2023, pp. 1–5. doi: 10.1109/ICASSP49357.2023.10095631.

R. V. Sharan, “Cough sound detection from raw waveform using SincNet and bidirectional GRU,” Biomed. Signal Process. Control, vol. 82, p. 104580, Apr. 2023, doi: 10.1016/j.bspc.2023.104580.

Y. Chen et al., "Robust Cough Detection with Out-of-Distribution Detection," in IEEE Journal of Biomedical and Health Informatics, vol. 27, no. 7, pp. 3210-3221, July 2023, doi: 10.1109/JBHI.2023.3264783.

J. Shen et al., "Piecewise Position Encoding in Convolutional Neural Network for Cough-Based Covid-19 Detection," ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Rhodes Island, Greece, 2023, pp. 1-5, doi: 10.1109/ICASSP49357.2023.10095316.

N. Sharma et al., “Coswara -- A Database of Breathing, Cough, and Voice Sounds for COVID-19 Diagnosis,” in Interspeech 2020, Oct. 2020, pp. 4811–4815. doi: 10.21437/Interspeech.2020-2768.




DOI: https://doi.org/10.15575/telka.v10n1.12-21

Refbacks

  • Saat ini tidak ada refbacks.


Jurnal TELKA terindex oleh :


     moraref logo       Crossref logo        sinta logo     base logo


Onesearch logo     IPI logo      Dimensions logo




Didukung oleh :







Lisensi Creative Commons
Ciptaan disebarluaskan di bawah Lisensi Creative Commons Atribusi-NonKomersial-BerbagiSerupa 4.0 Internasional.