Acil servise göğüs ağrısı şikayeti ile başvuran hastaların esı triajı ve EDACS değerlendirmelerinin yapay zeka ile karşılaştırılması
Yükleniyor...
Tarih
2024
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Necmettin Erbakan Üniversitesi, Tıp Fakültesi
Erişim Hakkı
info:eu-repo/semantics/openAccess
Özet
Grş
Acl servse göğüs ağrısı le başvuran hastalarla yapılan bu çalışmada acl servslerde yaygın olarak kullanılan br
traj sstem olan Emergency Severty Indeks (ESI) traj kategorsnn ve Emergency Department Assessment of
Chest Pan Score (EDACS) kardyak rsk grubunun yapay zeka algortmalarıyla karşılaştırıp lteratüre katkı
sağlamak amaçlanmıştır.
Gereç Yöntem
Bu çalışma, 01 Ağustos 2024 le 01 Ocak 2025 tarhler arasında Necmettn Erbakan Ünverstes Erşkn Acl
Servs’nde, göğüs ağrısı olan 18 yaş üstü hastalar üzernde tek merkezl, prospektf ve kestsel br tasarımla
toplam 396 hasta üzernde gerçekleştrlmştr. Gebelk durumu, travmaya bağlı göğüs ağrısı, kend steğyle
taburcu olan ve çalışma çn onam vermeyen hastalar çalışmaya dahl edlmemştr. Çalışma kapsamında,
ayaktan ve ambulans le acl servse başvuran hastaların vtal bulguları, demografk özellkler, ESİ traj
kategorler, EDACS göğüs ağrısı rsk kategors ve acl servs sonlanımları hasta takp formuna kaydedlmştr.
Ayaktan başvuran hastaların trajını traj görevller, ambulansla gelen hastaların trajını se hekmler
gerçekleştrmştr. Toplanan verler, standart br metn formatında düzenlenerek ChatGPT-4o’ya brbrnden
bağımsız olarak sunulmuş ve sstematk br şeklde hasta takp formuna kaydedlmştr.
Bulgular
Çalışma grubunun yaş ortalaması 51,9±17,6 yıl olup katılımcıların %56,6’sı erkektr. Hastaların %78,3’ü ayaktan
başvurmuştur. Traj görevls, ayaktan başvurup taburcu edlen hastaların %57,8'n ESİ Kategor-2 olarak
değerlendrrken bu oran yapay zekada %29,1, EDACS vers verlen yapay zekada se %18,4'tür (p<0,001).
Hekm ve yapay zeka, ambulansla acl servse başvurup taburcu edlen hastalarda benzer oranlarda ESİ Kategor-
2 olarak değerlendrme yaparken; EDACS vers verlen yapay zeka, hastaların %15'n ESİ Kategor-2 olarak
değerlendrmştr (p<0,001). Krtk hastaların ayırt edlmesnde yapay zeka, hekm le benzer br performans
serglemş olup aralarında anlamlı br fark saptanmamıştır (p=0,772). Yapay zekaya, kardyak rsk gruplaması
yapması çn EDACS le aynı verler verldğnde hastaları daha yüksek rsk grubunda değerlendrme eğlmnde
olduğu gözlemlenmştr (p<0,001).
Sonuç
Yapay zeka tabanlı sstemlern, hastaları ver tabanlı analz yeteneğ ve sstematk değerlendrme özellkler
sayesnde traj görevller ve hekmlere kıyasla hastaların klnk sonlanımıyla daha örtüşen br hasta trajı yaptığı
gözlenmştr. Bu blgler ışığında yapay zeka tabanlı sstemlern ve EDACS’ın acl servs trajında
kullanılablrlğ, hasta sonuçlarını yleştrme ve kaynakları verml kullanma potansyel açısından önem
taşımaktadır.
Introducton Ths study amed to contrbute to the lterature by comparng the Emergency Severty Index (ESI) trage categores, a commonly used trage system n emergency departments, and the Emergency Department Assessment of Chest Pan Score (EDACS) cardac rsk groups wth artfcal ntellgence algorthms n patents presentng to the emergency department wth chest pan. Materal and methods Ths study was conducted between August 1, 2024, and January 1, 2025, at Necmettn Erbakan Unversty Adult Emergency Department as a sngle-center, prospectve, and cross-sectonal study nvolvng a total of 396 patents aged 18 years and older wth chest pan. Patents who were pregnant, had chest pan due to trauma, were dscharged upon ther own request, or dd not provde consent for the study were excluded. The study recorded the vtal sgns, demographc characterstcs, ESI trage categores, EDACS chest pan rsk categores, and emergency department outcomes of patents who presented to the emergency department ether on foot or va ambulance. Trage for patents presentng on foot was conducted by trage staff, whle trage for patents arrvng by ambulance was performed by physcans. The collected data were organzed nto a standardzed text format, ndependently entered nto ChatGPT-4o, and systematcally recorded n the patent follow-up form. Results The mean age of the study group was 51.9±17.6 years, wth 56.6% of the partcpants beng male. A total of 78.3% of the patents presented to the emergency department as walk-ns. The trage staff classfed 57.8% of the walk-n patents who were dscharged as ESI Category-2, whle ths rate was 29.1% for the AI system and 18.4% for the AI system wth EDACS data (p<0.001). For patents presentng va ambulance and subsequently dscharged, physcans and the AI system showed smlar rates n classfyng patents as ESI Category-2. However, the AI system wth EDACS data classfed 15% of these patents as ESI Category-2 (p<0.001). In dstngushng crtcal patents, the AI system demonstrated a performance comparable to that of physcans, wth no sgnfcant dfference between them (p=0.772). When provded wth the same EDACS data for cardac rsk stratfcaton, the AI system tended to classfy patents n hgher rsk groups (p<0.001). Concluson Artfcal ntellgence-based systems were observed to perform patent trage more algned wth clncal outcomes compared to trage staff and physcans, thanks to ther data-drven analyss capablty and systematc evaluaton features. In lght of ths nformaton, the use of AI-based systems and EDACS n emergency department trage holds sgnfcant potental for mprovng patent outcomes and optmzng resource utlzaton.
Introducton Ths study amed to contrbute to the lterature by comparng the Emergency Severty Index (ESI) trage categores, a commonly used trage system n emergency departments, and the Emergency Department Assessment of Chest Pan Score (EDACS) cardac rsk groups wth artfcal ntellgence algorthms n patents presentng to the emergency department wth chest pan. Materal and methods Ths study was conducted between August 1, 2024, and January 1, 2025, at Necmettn Erbakan Unversty Adult Emergency Department as a sngle-center, prospectve, and cross-sectonal study nvolvng a total of 396 patents aged 18 years and older wth chest pan. Patents who were pregnant, had chest pan due to trauma, were dscharged upon ther own request, or dd not provde consent for the study were excluded. The study recorded the vtal sgns, demographc characterstcs, ESI trage categores, EDACS chest pan rsk categores, and emergency department outcomes of patents who presented to the emergency department ether on foot or va ambulance. Trage for patents presentng on foot was conducted by trage staff, whle trage for patents arrvng by ambulance was performed by physcans. The collected data were organzed nto a standardzed text format, ndependently entered nto ChatGPT-4o, and systematcally recorded n the patent follow-up form. Results The mean age of the study group was 51.9±17.6 years, wth 56.6% of the partcpants beng male. A total of 78.3% of the patents presented to the emergency department as walk-ns. The trage staff classfed 57.8% of the walk-n patents who were dscharged as ESI Category-2, whle ths rate was 29.1% for the AI system and 18.4% for the AI system wth EDACS data (p<0.001). For patents presentng va ambulance and subsequently dscharged, physcans and the AI system showed smlar rates n classfyng patents as ESI Category-2. However, the AI system wth EDACS data classfed 15% of these patents as ESI Category-2 (p<0.001). In dstngushng crtcal patents, the AI system demonstrated a performance comparable to that of physcans, wth no sgnfcant dfference between them (p=0.772). When provded wth the same EDACS data for cardac rsk stratfcaton, the AI system tended to classfy patents n hgher rsk groups (p<0.001). Concluson Artfcal ntellgence-based systems were observed to perform patent trage more algned wth clncal outcomes compared to trage staff and physcans, thanks to ther data-drven analyss capablty and systematc evaluaton features. In lght of ths nformaton, the use of AI-based systems and EDACS n emergency department trage holds sgnfcant potental for mprovng patent outcomes and optmzng resource utlzaton.
Açıklama
Anahtar Kelimeler
Kaynak
WoS Q Değeri
Scopus Q Değeri
Cilt
Sayı
Künye
Çınar, M. O. (2025). Acil servise göğüs ağrısı şikayeti ile başvuran hastaların esı triajı ve EDACS değerlendirmelerinin yapay zeka ile karşılaştırılması. (Yayınlanmamış tıpta uzmanlık tezi) Necmettin Erbakan Üniversitesi, Tıp Fakültesi Dahili Tıp Bilimleri Bölümü Acil Tıp Anabilim Dalı, Konya.
