Yan�k hastalar�n�n mortalite �ng�r�s�nde makine ��renimi algoritmalar�n�n kullan�lmas�

Yaz�c�, Hilmi; Ugurlu, Onur; Aygul, Yesim; Yildirim, Mehmet; U�ar, Ahmet Deniz

30/4Gelecek Say� Ar�iv Kapaklar Pop�ler Makaleler

ICMJE COI Form

H�zl� Arama

Yan�k hastalar�n�n mortalite �ng�r�s�nde makine ��renimi algoritmalar�n�n kullan�lmas� [Ulus Travma Acil Cerrahi Derg]

Ulus Travma Acil Cerrahi Derg. 2023; 29(10): 1130-1137 | DOI: 10.14744/tjtes.2023.79968

Yan�k hastalar�n�n mortalite �ng�r�s�nde makine ��renimi algoritmalar�n�n kullan�lmas�

Hilmi Yaz�c�¹, Onur Ugurlu², Yesim Aygul³, Mehmet Yildirim⁴, Ahmet Deniz U�ar⁴
¹Marmara �niversitesi Pendik E�itim ve Ara�t�rma Hastanesi, Genel Cerrahi Klini�i, �stanbul, T�rkiye
²�zmir Bak�r�ay �niversitesi, Mimarl�k ve M�hendislik Fak�ltesi, �zmir, T�rkiye
³Ege �niversitesi, Matematik B�l�m�, �zmir, T�rkiye
⁴Sa�l�k Bilimleri �niversitesi, �zmir Bozyaka E�itim ve Ara�t�rma Hastanesi, Genel Cerrahi Klini�i, �zmir, T�rkiye

AMA�: Yan�klar d�nya �ap�nda en yayg�n travmalardan biridir. Ciddi yan�k hastalar�nda, mortalite ve morbidite riski olduk�a y�ksektir. Bu �al��mada, yan�k mortalitesi i�in risk fakt�rlerini de�erlendirmeyi ve bu fakt�rler i�in alt� farkl� Makine ��renimi (M�) Algoritmas�n�n tahmin performanslar�n� kar��la�t�r�lmas� ama�land�.
GERE� VE Y�NTEM: Yan�k tedavi merkezinde tedavi edilen yan�k hastalar�n�n t�bbi kay�tlar� retrospektif olarak incelendi. Hastalar�n ya� ve cinsiyet, Toplam Yan�k Y�zey Alan�, �nhalasyon Yaralanmas�, tam kat yan�klar�, yan�k tipleri gibi verileri kaydedildi ve M� modellerinde girdi �zelli�i olarak kullan�ld�. Hastalar Ya�ayanlar ve Ya�amayanlar olarak iki grup alt�nda incelendi. Mortaliteyi tahmin etmek i�in k-Nearest Neighbor, Decision Tree, Random Forest, Support Vector Machine, Multi-Layer Perceptron, ve AdaBoost olmak �zere alt� M� algoritmas� kullan�ld�. Her algoritma i�in bir�ok farkl� ��reti kombinasyonu de�erlendirildi.
BULGULAR: �al��maya dahil edilen hasta say�s� 363 idi. M� algoritmalar�na dahil edilen alt� parametrenin t�m� tek de�i�kenli analizde anlaml� bir fark g�sterdi (p<0.001). Sonu�lar, t�m girdi �zelliklerini kullanan AdaBoost algoritmas�n�n %90 Do�ruluk ve %92 AUC ile en iyi tahmin performan-s�na sahip oldu�unu g�stermektedir.
SONU�: Do�ru verilerin ��retildi�i M� algoritmalar�, yan�k mortalitesinde y�ksek tahmin performans� g�stermi�tir. Yan�k hastalar� i�in uygun verilerin ��retildi�i M� modellerinin, klinik pratikte kullan��l� olabilece�ini d��nmekteyiz. Bu nedenle, bu konuda daha ileri ara�t�rmalara ihtiya� vard�r.

Anahtar Kelimeler: Makine ��renimi, sa�kal�m; tahminleme; yan�k.

The effect of well-known burn-related features on machine learning algorithms in burn patients� mortality prediction

Hilmi Yaz�c�¹, Onur Ugurlu², Yesim Aygul³, Mehmet Yildirim⁴, Ahmet Deniz U�ar⁴
¹Department of General Surgery, Marmara University, Pendik Training and Research Hospital, �stanbul-T�rkiye
²Faculty of Engineering and Architecture, Izmir Bakircay University, �zmir-T�rkiye
³Department of Mathematics, Ege University, �zmir-T�rkiye
⁴Department of General Surgery, University of Health Sciences, Izmir Bozyaka Training and Research Hospital, �zmir-T�rkiye

BACKGROUND: Burns is one of the most common traumas worldwide. Severely injured burn patients have an increased risk for mortality and morbidity. This study aimed to evaluate well-known risk factors for burn mortality and comparison of six machine learn-ing (ML) Algorithms� predictive performances.
METHODS: The medical records of patients who had burn injuries treated at Izmir Bozyaka Training and Research Hospital's Burn Treatment Center were examined retrospectively. Patients� demographics such as age and gender, total burned surface area (TBSA), Inhalation injury (II), full-thickness burns (FTBSA), and burn types (BT) were recorded and used as input features in ML models. Pa-tients were analyzed under two groups: Survivors and Non-Survivors. Six ML algorithms, including k-Nearest Neighbor, Decision Tree, Random Forest, Support Vector Machine, Multi-Layer Perceptron, and AdaBoost (AB), were used for predicting mortality. Several different input feature combinations were evaluated for each algorithm.
RESULTS: The number of eligible patients was 363. All six parameters (TBSA, Gender, FTBSA, II, Age, BT) that were included in ML algorithms showed a significant difference (p<0.001). The results show that AB algorithm using all input features had the best predic-tion performance with an accuracy of 90% and an area under the curve of 92%.
CONCLUSION: ML algorithms showed strong predictive performance in burn mortality. The development of an ML algorithm with the right input features could be useful in the clinical practice. Further investigations are needed on this topic.

Keywords: Burn, machine learning; mortality; prediction.

Hilmi Yaz�c�, Onur Ugurlu, Yesim Aygul, Mehmet Yildirim, Ahmet Deniz U�ar. The effect of well-known burn-related features on machine learning algorithms in burn patients� mortality prediction. Ulus Travma Acil Cerrahi Derg. 2023; 29(10): 1130-1137

Sorumlu Yazar: Hilmi Yaz�c�, T�rkiye
Makale Dili: �ngilizce

ARA�LAR Tam Metin PDF Yazd�r Al�nt�y� �ndir RIS EndNote BibTex Medlars Procite Reference Manager E-Postala Payla� Yazara e-posta g�nder Benzer makaleler PubMed Google Scholar