Comparing Predictive Accuracy of Bleeding in Total Abdominal Hysterectomy Among Anesthesiologists, Gynecologists and AI: A Clinical Observational Study

Abstract

Aim: Accurate prediction of perioperative blood loss is critical for optimizing outcomes in total abdominal hysterectomy (TAH). Traditional estimation methods by clinicians are subjective and prone to variability, while artificial intelligence (AI) offers a potential data-driven alternative. This study compares the predictive accuracy of anesthesiologists, gynecologists, and the AI algorithm ChatGPT4.0 for blood loss during TAH. Material and Methods: This single-center, prospective observational study evaluated 50 patients who underwent TAH for benign conditions in 2023. Clinical data, including uterine size, surgical duration, and surgeon experience, were retrospectively collected. Participating gynecologists and anesthesiologists predicted intraoperative blood loss based on anonymized patient data. Predictions were compared to ChatGPT4.0’s outputs and actual recorded blood loss, categorized into mild, moderate, and severe bleeding levels. Sensitivity, positive predictive value, and overall accuracy were analyzed using statistical tests appropriate for data distribution. Results: Anesthesiologists achieved the highest overall accuracy (40%), excelling in moderate bleeding predictions. Gynecologists demonstrated moderate performance across all categories, with 38% accuracy. ChatGPT4.0 showed the lowest overall accuracy (34%) but outperformed clinicians in predicting severe bleeding (37.5% positive predictive value). Variability in clinician predictions highlighted the challenges of subjective estimation, while AI predictions demonstrated consistency but limited precision. Conclusions: AI offers promise in enhancing objective blood loss prediction, particularly for severe cases. However, its performance remains inferior to clinician estimates in most scenarios, underscoring the need for further algorithm refinement and integration into clinical workflows. Future research should focus on long-term validation and addressing ethical challenges in AI adoption.

Keywords

• AI
• anesthesiology
• artificial intelligence
• blood loss prediction; ChatGPT; gynecology; total abdominal hysterectomy

References

1. 1. Clarke-Pearson DL, Geller EJ. Complications of hysterectomy. Obstetrics & Gynecology. 2013;121(3):654–673.
2. 2. Meeks GR, Harris RL. Surgical approach to hysterectomy: abdominal, laparoscopy-assisted, or vaginal. Clinical obstetrics and Gynecology. 1997;40(4):886–894.
3. 3. Maresh M, Metcalfe M, McPherson K, Overton C, Hall V, Hargreaves J, et al. The VALUE national hysterectomy study: description of the patients and their surgery. BJOG. An International Journal of Obstetrics & Gynaecology. 2002;109(3):302–312.
4. 4. English EM, Bell S, Kamdar NS, Swenson CW, Wiese H, Morgan DM. Importance of estimated blood loss in resource utilization and complications of hysterectomy for benign indications. Obstetrics & Gynecology. 2019;133(4):650–657.
5. 5. Bonilla DJ, Mains L, Whitaker R, Crawford B, Finan M, Magnus M. Uterine weight as a predictor of morbidity after a benign abdominal and total laparoscopic hysterectomy. The Journal of reproductive medicine. 2007;52(6):490–498.
6. 6. Beck TL, Morse CB, Gray HJ, Goff BA, Urban RR, Liao JB. Route of hysterectomy and surgical outcomes from a statewide gynecologic oncology population: is there a role for vaginal hysterectomy? American Journal of Obstetrics and Gynecology. 2016;214(3):348. e1–348. e9.
7. 7. Catanzarite T, Saha S, Pilecki MA, Kim JY, Milad MP. Longer operative time during benign laparoscopic and robotic hysterectomy is associated with increased 30-day perioperative complications. Journal of minimally invasive gynecology. 2015;22(6):1049–1058.
8. 8. Vree FE, Cohen SL, Chavan N, Einarsson JI. The impact of surgeon volume on perioperative outcomes in hysterectomy. JSLS. Journal of the Society of Laparoendoscopic Surgeons. 2014;18(2):174.

9. 9. Yüzkat N, Soyalp C, Gülhas N. Comparison of the Error Rates of an Anesthesiologist and Surgeon in Estimating Perioperative Blood Loss in Major Orthopedic Surgeries: Clinical Observational Study. Journal of Anesthesia/Anestezi Dergisi ( JARSS). 2019;27(4)
10. 10. Stehrer R, Hingsammer L, Staudigl C, Hunger S, Malek M, Jacob M, et al. Machine learning based prediction of perioperative blood loss in orthognathic surgery. Journal of Cranio- Maxillofacial Surgery. 2019/11/01/. 2019;47(11):1676–1681.
11. 11. Kane S, Collins S, Sproat LA, Mangel J. Predictors of transfusion requirement among patients who undergo hysterectomy for benign disease. Journal of Gynecologic Surgery. 2012;28(2):113–115.
12. 12. Faul F, Erdfelder E, Buchner A, Lang A-G. Statistical power analyses using G* Power 3. 1: Tests for correlation and regression analyses. Behavior research methods. 2009;41(4):1149–1160.
13. 13. Cantuaria GH, Angioli R, Frost L, Duncan R, Penalver MA. Comparison of bimanual examination with ultrasound examination before hysterectomy for uterine leiomyoma. Obstetrics & Gynecology. 1998;92(1):109–112.
14. 14. Gunasekaran K, Punnagai K, Vijaybabu K. A comparative study of diluents (crystalloid vs colloid) in acute normovolaemic haemodilution (ANH). International Journal. 2014;5(2):64.
15. 15. Landis J. The Measurement of Observer Agreement for Categorical Data. Biometrics. 1977;
16. 16. Gluck O, Mizrachi Y, Kovo M, Divon M, Bar J, Weiner E. Major underestimation and overestimation of visual blood loss during cesarean deliveries: can they be predicted? Archives of Gynecology and Obstetrics. 2017;296:907–913.
17. 17. Sato M, Koizumi M, Inaba K, Takahashi Y, Nagashima N, Ki H, et al. Gynecologists may underestimate the amount of blood loss during total laparoscopic hysterectomy. Obstetrics and Gynecology International. 2018;2018(1):3802532. 18. Goh E, Gallo R, Hom J, Strong E, Weng Y, Kerman H, et al. Large language model influence on diagnostic reasoning: a randomized clinical trial. JAMA Network Open. 2024;7(10):e2440969–e2440969.
18. 19. Horita K, Hida K, Itatani Y, Fujita H, Hidaka Y, Yamamoto G, et al. Real-time detection of active bleeding in laparoscopic colectomy using artificial intelligence. Surgical Endoscopy. 2024;38(6):3461–3469.
19. 20. Wang M, Yi G, Zhang Y, Li M, Zhang J. Quantitative prediction of postpartum hemorrhage in cesarean section on machine learning. BMC Medical Informatics and Decision Making. 2024;24(1):166.,
20. 21. Alowais SA, Alghamdi SS, Alsuhebany N, Alqahtani T, Alshaya AI, Almohareb SN, et al. Revolutionizing healthcare: the role of artificial intelligence in clinical practice. BMC medical education. 2023;23(1):689.
21. 22. Xu H, Shuttleworth KMJ. Medical artificial intelligence and the black box problem: a view based on the ethical principle of “do no harm”. Intelligent Medicine. 2024;4(1):52–57.