
MISCELLANEOUS
Artificial Intelligence in Diagnosing and
Managing Vascular Surgery Patients: An
Experimental Study Using the GPT-4 Model
Vangelis G. Alexiou, MD,
1
Bauer E. Sumpio, MD,
2
Areti Vassiliou, MD,
1
Stavros K. Kakkos, MD,
3
George Geroulakos, MD
4
.
1
Ioannina University
Hospital, Ioannina, Greece;
2
Yale University School of Medicine, New
Haven, Connecticut, USA;
3
University Hospital of Patras, Patras,
Greece;
4
National and Kapodistrian University of Athens, Athens,
Greece
Objectives: The introduction of artificial intelligence (AI) has led to
groundbreaking advancements across many scientific fields. Machine
learning algorithms have enabled AI models to learn, adapt, and solve
complex problems in previously unimaginable ways. Natural language
processing (NLP) allows these models to comprehend and respond to in-
quiries in a natural and humanly understandable way. We sought to
investigate the application and performance of an AI chatbot in the diag-
nosis and management of vascular surgery patients.
Methods: An experimental study was undertaken to evaluate the GPT-4
AI model’s performance across 57 clinical scenarios derived from a text-
book on vascular surgery. Specific prompts were devised to address the
AI model and task it to identify symptoms, diagnose conditions, and
select appropriate therapeutic approaches. Answers were scored,
descriptive statistics were produced, and means were compared across
topics. The reasoning and evidence used in the cases in which AI per-
formed poorly were critically reviewed.
Results: The AI model correctly answered over 65% of the 385 ques-
tions. Performance variation between and within 13 vascular surgery
topics did not show any statistically signi ficant differences. Analysis of
the questions where the model failed by more than 50% suggests a
gap in the ability to interpret and process multifaceted medical informa-
tion. Twenty-seven percent of these errors were attributed to a potential
lack of understanding of complex clinical scenarios. The AI model also
quoted incorrect or outdated information in 14% of cases and showed
an inability to comprehend context, nuances, and medical classification
systems in 11% of cases.
Conclusions: GPT-4 demonstrated the potential to provide clinically
relevant answers for most of the tested scenarios. However, its reasoning
must still be carefully analyzed for exactitude and clinical validity. While
language models show promise as valuable tools for clinicians, it is essen-
tial to recognize their role as supportive mechanisms rather than stand-
alone solutions.
Author Disclosures: V. Alexiou: None; B. Sumpio: None; A. Vassiliou:
None; S. Kakkos: None; G. Geroulakos: None.
Endovascular Management of Pelvic Injuries
Nasser Algharem, MD,
1
Mohammed Almotlaq, MD,
1
Salem Al Obayah, MD,
1
Afnan Alwaresh, MBBS,
1
Najib Garad,
MBBCh,
2
Ammar Farhan, MBBCh,
1
Hassan Salman, BS,
1
Hussain
Balharith, BS,
1
Amal Alyami, BS
1
.
1
King Khalid Hospital, Najran, Saudi
Arabia;
2
Alfaisal University, Riyadh, Saudi Arabia
Objective: Pelvic vascular injuries are associated with significant mor-
tality and increasing morbidity. Pelvic injuries, often resulting from
trauma such as that incurred in motor vehicle accidents (Fig 1), war injury
(Fig 2), or falls can lead to complex vascular and orthopedic complica-
tions. Over the last nine years, we have treated pelvic vascular injuries
with embolization (Fig 2) and covered stents.
Methods: In this study, 110 cases of signifi cant pelvic vascular injury were
included and classified anatomically according to the extent of the injury
treated with stent-graft placement, embolization, or combined therapy.
Of the patients, 45 underwent combined treatment with embolization
with gelfoam and coils (Fig 2), 30 unde rwent embolization with gelfoam
only, 20 underwent stent graft placement alone, and 15 unde rwent com-
bined embolization and stenting. The anatomical territory was recorded
and studied.
Results: The anatomical classification was common iliac artery in 15
cases, external iliac artery in 21 (Fig 1), internal iliac artery in 32 (Fig 1), in-
ternal pudendal artery in 15 (Fig 2), and combined small an d possible
bone marrow pelvic bleeders in 60. Of the patients, 22 were women
and 88 were men, with ages ranging from 18 to 60 years. Significant
blood loss r equirin g bloo d transfusion was reported in 76 cases, and
PSA were found in 10 cases. The pelvic vascular injury was caused by
road traffic accident in 62 cases, surgery in 6, and war-related injury in
42.
Conclusion: Interventional radiology plays a crucial role in the m an-
agement of pelvic injuries, offering minimally invasive techniques that
can significantly reduce morbidity and mortality. Gelfoam embolization
of open -book pelvic fractures and possibl e bone marrow bleeding
proved advantageous in savi ng the patient and min imizing blood trans-
fusions. Stent grafts were mainly used in injuries involving large- and
medium-sized arteries to retain vascular integrity and flow.
Author Disclosures: N. Algharem: None; M. Almotlaq: None; S. Al
Obayah: None; A. Alwaresh: None; N. Garad: None; A. Farhan: None; H.
Salman: None; H. Balharith: None; A. Alyami: None.
Embolization of a Giant Renal Arteriovenous
Fistula: Case Report
Guilherme Almeida, MD,
1
Carlos Macedo, MD,
1
Davi Colen, MD,
1
Filipe Terra, MD,
2
Guilherme Carvalho, MD,
1
Helady Pinheiro, PhD,
1
Salim Ferreira, MD
1
.
1
Juiz de Fora Federal University, Juiz de Fora,
Brazil;
2
SUPREMA, Juiz de Fora, Brazil
Objectives: T o report a case of embolization of a giant arteriovenous fis-
tula (AVF) that remained unreported for 23 years after a percutaneous
renal biopsy.
Methods: Case Report:
A 35-year-old male patient had Berger’s disease that developed at the
age of 13. He had undergone a percutaneous needle biopsy (PNB) of the
left kidney, which was uneventful and led to the diagnosis of Berger’s dis-
ease. A kidney transplant was performed without complications, and the
patient remained asymptomatic on immunosuppressants.
Twenty-three years later, an abdominal ultrasound revealed a mass in
the left flank. Computed tomography angiography (CTA) showed a giant
Fig 1. Post-road traffic accident transection of the left external iliac
artery with profuse bleeding.
Fig 2. War-related left internal iliac artery anterior division and pu-
dendal artery injury.
Journal of Vascular Surgery
Abstracts
S45
Volume 81, Number 5S