The previous paper under discussion makes a positive and timely contribution, not despite the emergence of “unsure” responses in recent surveys, but precisely because of them. Rather than treating uncertainty as a failure of measurement or a lack of conviction, the authors correctly identify it as a distinct cognitive and emotional state—one that machine learning models are uniquely equipped to capture and predict.
In the aftermath of the Kahramanmaraş earthquakes (February 2023), medical students faced compounded stressors: disruption of training, exposure to mass casualty realities, and confrontation with their own professional vulnerability. Traditional survey analyses often collapse “unsure” into missing data or a neutral midpoint. However, recent surveys—including those cited in the previous work—show that students actively select “unsure” when asked about career decisions, ethical judgments, or readiness to practice in disaster settings. This is not evasion. It is an authentic expression of complex, context-driven indecision.
The previous paper’s machine learning approach is positive because it moves beyond binary or Likert-scale assumptions. By treating decisional uncertainty as a predictable target variable, the authors achieve several advances:
-
Ecological validity – Post-disaster environments generate real, legitimate ambiguity. Predicting who remains uncertain, and under what conditions, is as clinically relevant as predicting certainty.
-
Feature importance – ML models (random forests, gradient boosting, etc.) can reveal which factors—prior disaster training, proximity to epicenter, personal loss, social support—most strongly discriminate between “certain,” “uncertain,” and “unsure” states.
-
Temporal dynamics – Decisional uncertainty is not static. With repeated cross-sectional or longitudinal surveys, ML can identify whether uncertainty resolves, persists, or worsens—and which interventions might accelerate resolution.
https://healthpr.org/journal/HPR/articles/online_first/7917
how to cite:
Kadir Uludag, Fatih Kara, Taşkın Soyaslan, Ömer Çelik, Enes Küçükbey, Hongxing Wang. Machine learning-driven prediction of decisional uncertainty among medical students post-Kahramanmaras earthquake. Health Psychology Research 0262. https://doi.org/10.36922/hpr.0262
