Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Comparative Study
. 2025 Jul 1;15(1):21728.
doi: 10.1038/s41598-025-06591-9.

Comparison of AI chatbot predicted and realworld survival outcomes in hepatocellular carcinoma

Engin Eren Kavak  1 , Efe Cem Erdat  2 , Zeynep Altundağ Derin  3 , İsmail Dilli  3 , Pınar Kubilay Tolunay  2 , Berna Öksüzoğlu  3 , Hakan Akbulut  2
Affiliations
Comparative Study

Comparison of AI chatbot predicted and realworld survival outcomes in hepatocellular carcinoma

Engin Eren Kavak et al. Sci Rep. .

Abstract

This study compares survival predictions made by an artificial intelligence (AI) based chatbot with real-world data in hepatocellular carcinoma (HCC) patients. It aims to evaluate the reliability and accuracy of AI technologies in HCC prognosis. A retrospective analysis was conducted on patients diagnosed with HCC. The estimated survival times for each patient were calculated using an artificial intelligence chatbot. The follow-up periods and mortality data for the patients were used to obtain real-life survival data. The predicted and actual survival times were statistically compared. ChatGPT-4o consistently overestimated the overall survival (OS) times compared to real-world outcomes.A statistically significant discrepancy was observed between the predicted and actual survival times (p < 0.05). Nevertheless, while the survival predictions of AI were more accurate in patients with advanced-stage HCC, the predictions differed significantly in patients with early-stage HCC. AI has the potential to play an important role in the prognosis of complex diseases such as HCC. However, this study's findings indicate that AI's predictions are not entirely consistent with real-world data, particularly in the context of early-stage diseases. Further large-scale studies may enhance the dependability of incorporating AI into clinical decision-support systems. The use of AI-assisted predictions may prove to be a valuable tool for the prediction of survival in patients with HCC. Nevertheless, further studies are required to substantiate the accuracy and reliability of these systems before they can be implemented in clinical practice.

Keywords: Artificial intelligence; Hepatocellular carcinoma; Predicted overall survival; Prognostic model.

PubMed Disclaimer

Conflict of interest statement

Declarations. Competing interests: The authors declare no competing interests. Ethics approval: Before the commencement of data collection, the study design was presented to the Ankara Etlik City Hospital review board and approved with the decision number AEŞK-BADEK-1024-844.

Figures

Fig. 1
Fig. 1
Kaplan-Meier overall survival curve for real-world data in HCC patients. Median OS: 10.6 months (range: 7.0–20.0).
Fig. 2
Fig. 2
Kaplan-Meier survival prediction curve generated by ChatGPT-4o. Median predicted OS: 15.0 months (range: 10.5–NR).
Fig. 3
Fig. 3
Forest plot of univariate Cox regression analysis. Significant prognostic variables (p < 0.05) are highlighted.
Fig. 4
Fig. 4
Forest plot of multivariate Cox regression. Only ECOG performance score remained statistically significant (p < 0.0001).
Fig. 5
Fig. 5
Bland-Altman analysis comparing AI-predicted vs. actual survival in the full cohort. Mean bias: 11.89 months.
Fig. 6
Fig. 6
Bland-Altman plot for patients with advanced-stage HCC. AI predictions closely aligned with actual outcomes (bias = 1.22 months).
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. Sung, H. et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. Cancer J. Clin.71(3), 209–249 (2021). - PubMed
    1. Siegel, R. L., Giaquinto, A. N. & Jemal, A. Cancer statistics, 2024. CA: A Cancer journal for clinicians. ;74(1):12–49. (2024). 10.3322/caac.21820 - PubMed
    1. McGlynn, K. A., Petrick, J. L. & El-Serag, H. B. Epidemiology of hepatocellular carcinoma. Hepatology73(Suppl 1), 4–13. 10.1002/hep.31288. (2021). - PMC - PubMed
    1. Zhou, K. et al. Association of nativity with survival among adults with hepatocellular carcinoma. J. Natl. Cancer Inst.115(7), 861–869. 10.1093/jnci/djad067 (2023). - PMC - PubMed
    1. Ryerson, A. B. et al. and B.A. Kohler. Annual Report to the Nation on the Status of Cancer, 1975–2012, featuring the increasing incidence of liver cancer. Cancer122(9), 1312–1337. 10.1002/cncr.29936 (2016). - PMC - PubMed

Publication types

LinkOut - more resources