ChatGPT 4, kritik hastaları yönetmekle ilgili soruları uygunsuz bir şekilde yanıtlayabilir

Öz

Yoğun bakım hastaları son derece karmaşık patolojilere sahiptir ve bu da tedavi yönetimlerini zorlaştırır. Bu çalışmada, büyük dilli bir model olan ChatGPT 4'ü kritik hasta yönetimi konusundaki bilgi birikimi açısından değerlendirmeyi amaçladık. Yoğun bakım ünitelerinde çalışan bir yoğun bakım uzmanı, anestezist ve nörolog tarafından kafa travması, pulmoner emboli, miyokard enfarktüsü, kronik obstrüktif akciğer hastalığı, kronik böbrek hastalığı ve enfektif endokardit, septik şok ve status epileptikus dahil olmak üzere mekanik ventilasyon içeren senaryolar oluşturuldu. Bu senaryolarla ilgili hasta yönetimiyle ilgili sorular ChatGPT 4'e önerildi. ChatGPT 4 soruların çoğunu doğru yanıtlasa da, yine de tıbbi girdi gerektiriyor.

Anahtar Kelimeler

• Büyük dil modelleri
• ChatGPT 4
• yoğun bakım üniteleri
• kritik bakım
• yapay zeka

ChatGPT 4 may inappropriately answer questions about managing critically ill patients

Öz

Intensive care patients have extremely complex pathologies, making their treatment management difficult. In this study, we aim to evaluate ChatGPT 4, a large language model, regarding its knowledge of critically ill patient management. Scenarios involving mechanical ventilation were created by an intensivist, anesthesiologist, and neurologist working in intensive care units, including head trauma, pulmonary embolism, myocardial infarction, chronic obstructive pulmonary disease, chronic kidney disease, and infective endocarditis, septic shock, and status epilepticus. Questions about patient management related to these scenarios were proposed to ChatGPT 4. Although ChatGPT 4 answered most of the questions correctly, it still requires medical input.

Anahtar Kelimeler

• Large language models
• ChatGPT 4
• intensive care units
• critical care
• artificial intelligence

Destekleyen Kurum

This research received no external funding.

Etik Beyan

This study does not require copyrighting for ChatGPT 4 outputs. According to OpenAI's Content Policy and Terms of Use, ChatGPT users own all outputs they create with large language models, including the text and images. We obtained ChatGPT 4’s responses to our designed questions. ChatGPT's answers to our questions are available at: https://github.com/barisaakan/llm-patients

Kaynakça

1. Lu Y, Wu H, Qi S, Cheng K. Artificial Intelligence in Intensive Care Medicine: Toward a ChatGPT/GPT-4 Way? Ann Biomed Eng. 2023;51:1898-903.
2. https://github.com/barisaakan/llm-patients
3. Monjezi M, Jamaati H. The effects of pressure- versus volume-controlled ventilation on ventilator work of breathing. Biomed Eng Online. 2020;19:72.
4. Chiumello D, Meli A, Pozzi T, Lucenteforte M, Simili P, Sterchele E, Coppola S. Different Inspiratory Flow Waveform during Volume-Controlled Ventilation in ARDS Patients. J Clin Med. 2021;10:4756.
5. McDonald EG, Aggrey G, Aslan AT, Casias M, Cortes-Penfield N, Dong MQD et al. Guidelines for Diagnosis and Management of Infective Endocarditis in Adults: A WikiGuidelines Group Consensus Statement. JAMA Netw Open. 2023;6:e2326366.
6. Zhu X, Wang Z, Ferrari MW, Ferrari-Kuehne K, Hsi et al. Management of anticoagulation in patients with infective endocarditis. Thromb Res. 2023;229:15-25.
7. Ibrahim F, Murr NI. Lafora Disease 2024. In: StatPearls [Internet]. Treasure Island (FL): StatPearls Publishing; 2025.
8. Parihar R, Ganesh S. Lafora progressive myoclonus epilepsy: Disease mechanism and therapeutic attempts. J Biosci. 2024;49:22.

9. Meo SA, Al-Masri AA, Alotaibi M, Meo MZS, Meo MOS. ChatGPT Knowledge Evaluation in Basic and Clinical Medical Sciences: Multiple Choice Question Examination-Based Performance. Healthcare (Basel). 2023;11:2046.
10. Wang X, Tang J, Feng Y, Tang C, Wang X. Can ChatGPT-4 perform as a competent physician based on the Chinese critical care examination? J Crit Care. 2025;86:155010.