EUVIMEDEUVIMED entdecken
ÜbersichtsarbeitSystematische Übersicht

Artificial Intelligence Models to Predict Recurrence in Patients with Differentiated Thyroid Cancer: A PROBAST + AI Systematic Review of Clinical Applicability

KI-Modelle zur Vorhersage eines Rezidivs bei differenziertem Schilddrüsenkarzinom: systematische Übersichtsarbeit zur klinischen Anwendbarkeit mit PROBAST + AI

Paola Solis-Pazmino iD, Alison Fernanda Samaniego iD, Lenara Golbert, Erika Laurini de Souza Meyer iD, Eduardo A Pilatuna iD, Vanessa Suñé Mattevi, Freddy Moposita Molina, Debora Nienow, Kamille Guidolin, Laura Berton Eidt, Virgilio Gonzales Zanella, Camilo Gonzalez-Velazquez iD und 6 weitere

Thyroid® · 12.11.2026

DOI: 10.1177/10507256261504546 ↗

Auf einen Blick

34 Studien zu KI-Modellen für DTC-Rezidive zeigten meist gute Diskrimination, aber häufig methodische Mängel und fehlende Kalibrierung. Die Evidenz reicht laut Autor:innen nicht für den Routinedeinsatz.

Fragestellung

Wie gut sind KI-Modelle zur Vorhersage eines Rezidivs bei differenziertem Schilddrüsenkarzinom hinsichtlich Qualität, Verzerrungsrisiko und Anwendbarkeit?

Methode

Systematische Übersichtsarbeit; Suche in Ovid MEDLINE, Embase, Cochrane, Scopus und Web of Science bis 13. Juli 2026; Bewertung mit PROBAST + AI.

Ergebnisse

34 Studien mit 21.267 Patient:innen; 21 einzigartige Datensätze und 13 UCI-abgeleitete Publikationen. AUC 0,67–1,00 in 22 von 34 Studien; nur 10 prüften Kalibrierung.

Einschränkungen

26 von 34 Entwicklungsstudien mit hohen Qualitätsbedenken; 17 von 34 nur geringe Anwendungsbedenken; keine standardisierten Kalibrierungsmaße.

Schlussfolgerung der Autor:innen

KI-Modelle für DTC-Rezidive zeigen vielversprechende Diskriminierung, haben aber wesentliche methodische Schwächen; robuste, validierte und klinisch kalibrierte Modelle fehlen.

Kernaussagen

  • AUC in 22 von 34 Studien: 0,67–1,00
  • Nur 10 Studien berichteten Kalibrierung
  • Routinemäßige klinische Nutzung derzeit nicht gestützt

Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.

Kurz gefasst

Übersichtsarbeit in der Zeitschrift Thyroid®, online erschienen am 12.11.2026. Verfasst von Paola Solis-Pazmino und 17 weiteren Autor:innen. Beteiligte Einrichtung: CaTaLiNA: Cancer de tiroides en Latinoamerica, Quito, Ecuador. Die Arbeit nennt 54 Literaturangaben.

Auf einen BlickAbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Differentiated thyroid cancer (DTC) generally carries an excellent prognosis, yet recurrence occurs in up to 8–9% of patients. Artificial intelligence (AI) models are increasingly proposed to predict DTC recurrence; however, their methodological quality and clinical readiness are poorly defined. We conducted a systematic review to evaluate the methodological quality, risk of bias, and applicability of these models using the PROBAST + AI framework.

Methods

We searched Ovid MEDLINE, Embase, Cochrane databases, Scopus, and Web of Science from inception to July 13, 2026. Eligible studies included original research developing or validating AI models for recurrence in DTC. Model quality (for development), risk of bias (for evaluation), and applicability were assessed using the PROBAST + AI framework, with a focus on discrimination, calibration, and validation strategies (PROSPERO; registration number [CRD420251004644]).

Results

Of 673 records screened, 34 studies were included (21 unique datasets and 13 UCI-derived publications), encompassing 21,267 patients with a recurrence rate of 7.8%. Most were developed in Asian cohorts (17 of 21 datasets) and were structured clinical/tabular variables (82.4%), including demographic, clinical, laboratory, and pathological data, whereas the remaining models incorporated multimodal inputs such as imaging-derived or radiomic features. Notably, nearly one quarter of included studies (13 of 34, 38%) represented reanalysis of a single dataset. Discrimination was reported in 22 of 34 studies (64.7%) (area under the receiver operating characteristic curve [AUC] 0.67–1.00), though only 10 studies assessed calibration, none with standardized metrics. The PROBAST + AI assessment revealed high overall quality concerns in 26 of 34 (76.5%) development studies due to reliance on single-center cohorts, inadequate handling of missing data, and absence of external validation. Only 6 of 34 included studies (17.6%) were judged to have low overall quality concern, whereas 17 of 34 (50.0%) were rated as having low applicability concern.

Conclusions

AI models for predicting DTC recurrence show promising discriminatory performance but have substantial methodological shortcomings in development, and a lack of calibration and external validation necessary for clinical use. Given the guidelines’ emphasis on risk-adapted de-escalation of therapy, the development of robust, validated, and clinically calibrated AI tools is a priority. Current evidence remains insufficient to support routine clinical use.

Schlagwörter

Stichwörter der Autor:innen

PROBAST + AIartificial intelligencemachine learningpredictionrecurrencethyroid cancer

Autor:innen und Institutionen

  • CaTaLiNA: Cancer de tiroides en Latinoamerica, Quito, Ecuador.
  • Division of Head and Neck Surgery, Santa Casa de Porto Alegre, Porto Alegre, Brazil.
  • Knowledge and Evaluation Research Unit, Mayo Clinic, Rochester, Minnesota, USA.
  • Graduate Program in Pathology, Federal University of Health Sciences of Porto Alegre (UFCSPA), Porto Alegre, Brazil.
  • Medical School, Universidad Central del Ecuador, Quito, Ecuador.
  • Division of Endocrinology, Santa Casa de Porto Alegre, Porto Alegre, Brazil.

ORCID-Profile: Paola Solis-Pazmino, Alison Fernanda Samaniego, Erika Laurini de Souza Meyer, Eduardo A Pilatuna, Camilo Gonzalez-Velazquez, Luis A Figueroa, David Toro-Tobon

Zitieren

APA

Solis-Pazmino, P., Samaniego, A. F., Golbert, L., Laurini de Souza Meyer, E., Pilatuna, E. A., Suñé Mattevi, V., Moposita Molina, F., Nienow, D., Guidolin, K., Berton Eidt, L., Gonzales Zanella, V., Gonzalez-Velazquez, C., Lincango, E. P., Figueroa, L. A., Tuttle, R. M., Toro-Tobon, D., Brito, J. P., & Ponce-Ponte, O. J. (2026). Artificial Intelligence Models to Predict Recurrence in Patients with Differentiated Thyroid Cancer: A PROBAST + AI Systematic Review of Clinical Applicability. Thyroid®. https://doi.org/10.1177/10507256261504546

Vancouver

Solis-Pazmino P, Samaniego AF, Golbert L, Laurini de Souza Meyer E, Pilatuna EA, Suñé Mattevi V, et al. Artificial Intelligence Models to Predict Recurrence in Patients with Differentiated Thyroid Cancer: A PROBAST + AI Systematic Review of Clinical Applicability. Thyroid®. 2026. doi:10.1177/10507256261504546

BibTeX
@article{SolisPazmino2026,
  author = {Solis-Pazmino, Paola and Samaniego, Alison Fernanda and Golbert, Lenara and Laurini de Souza Meyer, Erika and Pilatuna, Eduardo A and Suñé Mattevi, Vanessa and Moposita Molina, Freddy and Nienow, Debora and Guidolin, Kamille and Berton Eidt, Laura and Gonzales Zanella, Virgilio and Gonzalez-Velazquez, Camilo and Lincango, Eddy P and Figueroa, Luis A and Tuttle, R Michael and Toro-Tobon, David and Brito, Juan P and Ponce-Ponte, Oscar J},
  title = {Artificial Intelligence Models to Predict Recurrence in Patients with Differentiated Thyroid Cancer: A PROBAST + AI Systematic Review of Clinical Applicability},
  journal = {Thyroid®},
  year = {2026},
  doi = {10.1177/10507256261504546}
}
Kostenlos starten →