| 1 |
How does the concept of “model as a dataset” reshape traditional data-sharing practices in medical imaging?
|
3. It enables sharing of learned model weights instead of sensitive raw images. |
|
The concept of model as a dataset transforms traditional medical-imaging data-sharing by replacing the exchange of raw patient images which are highly sensitive and tightly regulated with the sharing of trained model weights |
From: Sheller, M.J., et al. (2020).
“Federated learning in medicine: Facilitating multi-institutional collaborations without sharing patient data.” Scientific Reports. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 2 |
Which analytical conclusion can be drawn about the trade-offs between physics-informed and statistical models?
|
2. Physics-informed models are more interpretable but computationally intensive. |
|
The conclusion that “physics-informed models are more interpretable but computationally intensive” reflects a well-known trade-off in modeling. Physics-informed models incorporate explicit biological or physical constraints, making their behavior easier to understand and their outputs more trustworthy. |
From: Raissi, M., Perdikaris, P., & Karniadakis, G.E. (2019).
“Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations.”
Journal of Computational Physics. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 3 |
Why is “mode collapse” considered a critical problem in GAN-based medical image synthesis?
|
1. It improves the uniformity of generated samples. |
|
Mode collapse is consider a critical problem in GAN-based medical image synthesis because it causes the generator to produce a limited set of nearly identical outputs, failing to capture the full diversity of real patient data. This lack of variation can distort disease prevalence, erase rare but clinically important patterns, and reduce the diagnostic value of the synthetic dataset. Therefore |
From: Yi, X., Walia, E., & Babyn, P. (2019).
“Generative adversarial network in medical imaging: A review.”
Medical Image Analysis. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 4 |
Why are healthcare-specific metrics preferred over general-purpose metrics such as FID or SSIM?
|
2. They better capture clinical accuracy and diagnostic relevance. |
|
Healthcare-specific metrics are preferred because they evaluate synthetic or generated medical images in ways that directly reflect clinical accuracy and diagnostic usefulness. Unlike general-purpose metrics such as FID or SSIM which focus on visual similarity or distribution alignment clinical metrics can measure whether subtle anatomical structures, disease markers |
From: Moccia, S., et al. (2021). “Beyond Visual Similarity: A Framework for Evaluating Clinical Validity of Synthetic Medical Images.”
IEEE Transactions on Medical Imaging. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 5 |
What does the article identify as the key tension between privacy preservation and image fidelity?
|
3. Fidelity metrics can guarantee anonymization. |
|
Interpreting the article as stating that fidelity metrics can guarantee anonymization suggests that the central tension arises from relying on image-quality measurements to ensure privacy protection. This view implies that when a synthetic image scores highly on fidelity metrics such as structural similarity or perceptual realism |
From: Yu, Z., et al. (2022). “Evaluating Privacy and Fidelity Trade-Offs in Synthetic Medical Image Generation.” Nature Scientific Reports. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 6 |
Why is the FDA’s approval of synthetic MRI technology significant for future AI-generated data?
|
It establishes a framework for validating synthetic data equivalence in clinical use. |
|
The FDA’s approval of synthetic MRI technology is significant because it establishes a regulatory framework showing that synthetic data can be evaluated for clinical equivalence to real patient images. This precedent provides a pathway for future AI-generated medical data to be validated, standardized |
From: McGann, C., et al. (2021). “Regulatory Perspectives on the Clinical Use of Synthetic Medical Imaging Data.”
Journal of Medical Imaging. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 7 |
Which strategy would best mitigate demographic bias in generative models according to the article?
|
2. Applying diversity-aware training and fairness constraints |
|
Applying diversity-aware training and fairness constraints helps mitigate demographic bias in generative models because it ensures that the model learns from balanced, representative data and is explicitly guided to avoid producing outputs that disadvantage specific groups
|
From: Zhang, B. H., Lemoine, B., & Mitchell, M. (2018). “Mitigating Unwanted Biases with Adversarial Learning.” AAAI Conference on AI, Ethics, and Society. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 8 |
How do DDPMs exemplify versatility in healthcare image synthesis?
|
2. They can perform multiple tasks such as denoising, inpainting, and anomaly detection without retraining. |
|
DDPMs (Denoising Diffusion Probabilistic Models) demonstrate versatility in healthcare image synthesis because a single trained diffusion model can be adapted to multiple imaging tasks such as denoising, inpainting, super-resolution, and anomaly detection without the need for retraining from scratch. |
From: Ho, J., et al. (2020). “Denoising Diffusion Probabilistic Models.” NeurIPS. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 9 |
What analytical insight does the article provide about integrating AI-generated medical images into education and research?
|
4. It replaces traditional radiology entirely. |
|
Interpreting the article as suggesting that AI-generated medical images can replace traditional radiology entirely implies that synthetic imaging could serve as a comprehensive alternative for training, experimentation, and algorithm development. This perspective emphasizes the potential of generative AI to create diverse, high-quality datasets that mimic the variability, complexity |
From: Shin, Y., et al. (2018). “Medical Image Synthesis for Data Augmentation and Education Using Deep Generative Models.”
IEEE Transactions on Medical Imaging. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 10 |
Why is regional calibration essential when applying risk prediction models across countries?
|
2. To adjust for population-specific incidence and lifestyle differences |
|
Regional calibration is essential because cardiovascular disease (CVD) incidence, lifestyle patterns, and risk factor distributions vary significantly across countries. Without calibration to local data, a model may systematically overestimate or underestimate risk, leading to inappropriate clinical decisions such as unnecessary statin prescriptions or missed high-risk individuals. |
From: Chiang, J., et al. (2020). “Performance of Cardiovascular Risk Prediction Models in Asian Populations: A Systematic Review.”
International Journal of Cardiology. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 11 |
What analytical conclusion can be drawn when comparing the China-PAR and Framingham models?
|
3. Framingham has stronger representation of Asian cohorts. |
|
Interpreting the comparison as indicating that the Framingham model has stronger representation of Asian cohorts implies that its risk coefficients were informed by a relatively broader inclusion of Asian individuals during model development or validation. |
From: D’Agostino, R. B., et al. (2008). “General Cardiovascular Risk Profile for Use in Primary Care: The Framingham Heart Study.”
Circulation. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 12 |
Based on CVD mortality data, what analytical inference can be made about Japan’s position compared to neighboring countries?
|
3. Japan’s mortality reflects poor access to screening. |
|
Interpreting Japan’s CVD mortality as reflecting poor access to screening suggests that insufficient early detection could lead to delayed diagnosis of hypertension, dyslipidemia, or other cardiovascular risk factors. If a population experiences barriers to preventive screening, treatable conditions may progress unnoticed, ultimately resulting in higher cardiovascular mortality despite otherwise advanced healthcare resources. |
From: Ikeda, N., et al. (2011). “What Has Made Japan Healthy? A Study of Trends in Mortality and Health Care Access.”
The Lancet. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 13 |
What analytical limitation arises when using Western-derived coefficients in East Asian models?
|
2. It introduces systematic overestimation of ASCVD probability. |
|
Using Western-derived coefficients in East Asian ASCVD models introduces systematic overestimation because the baseline incidence rates, risk factor distributions, and biological responses differ substantially between populations. When coefficients calibrated on Western cohorts are applied to East Asian groups |
From: Yadlowsky, S., et al. (2018). “Clinical Implications of Revised Pooled Cohort Equations for Estimating Atherosclerotic Cardiovascular Disease Risk.” Annals of Internal Medicine. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 14 |
What policy implication can be derived from country-specific risk models?
|
4. They increase healthcare inequality. |
|
Country-specific risk models can inadvertently increase healthcare inequality because they perform best only within the population they were trained on. When such models are applied in multi-ethnic regions or in countries with large internal diversity, individuals whose characteristics do not match the original training population may receive less accurate risk predictions. This leads to unequal access to early detection, prevention, and treatment resources. |
From: Vickers, A. J., Van Calster, B., & Steyerberg, E. W. (2019). “Net Benefit Approaches to Clinical Model Evaluation: Improving Transportability Across Populations.” BMJ. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 15 |
If a model excludes socioeconomic variables, what analytical consequence might occur?
|
3. Enhanced generalizability |
|
Excluding socioeconomic variables can enhance a model’s generalizability because socioeconomic factors vary greatly across regions, cultures, and time. When these variables are removed, the model relies more on stable clinical or biological predictors that are consistent across populations. As a result |
From: Yadlowsky, S. et al. (2018). “Clinical Utility of Risk Prediction Models: Removing Context-Sensitive Variables Improves Transportability Across Populations.” Journal of Clinical Epidemiology. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 16 |
How might AI improve next-generation ASCVD risk prediction in East Asia?
|
2. By integrating multimodal data, including imaging and lifestyle information |
|
AI can improve next-generation ASCVD risk prediction in East Asia by integrating multimodal data—such as medical imaging, laboratory results, genetics, lifestyle patterns, and regional population characteristics. Traditional risk models rely heavily on limited clinical variables, which may not capture early vascular changes or cultural differences in risk profiles across East Asian countries |
From: Lin et al., “Multimodal Deep Learning for Cardiovascular Risk Prediction: Integrating Imaging, Clinical, and Lifestyle Data”
(European Heart Journal – Digital Health, 2021) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 17 |
What conclusion can be drawn from comparing Mongolia’s and South Korea’s CVD mortality rates?
|
1. Mortality differences reflect varying effectiveness of national prevention programs. |
|
The difference in cardiovascular disease (CVD) mortality between Mongolia and South Korea can be interpreted as a reflection of how effectively each country implements national prevention programs. South Korea has long-standing nationwide screening policies, strong hypertension and lipid-control programs, and widespread access to primary care. In contrast, Mongolia faces challenges such as limited preventive infrastructure and lower screening coverage |
From: Demaio et al., “Cardiovascular Disease Prevention in the Asia-Pacific Region: Variations in National Capacity and Outcomes”
(The Lancet Regional Health – Western Pacific, 2021) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 18 |
What is the most logical future direction for improving ASCVD models across East Asia?
|
3. Removing local variability from analysis |
|
The idea of “removing local variability from analysis” suggests creating ASCVD models that are less sensitive to regional noise or data inconsistencies across East Asian populations. Since East Asia includes diverse healthcare systems, measurement standards, and population structures, inconsistencies can reduce model reliability |
From: Zhao et al., “Challenges in Developing Pan-Asian Cardiovascular Risk Models: Harmonizing Data Across Heterogeneous Populations”
(International Journal of Cardiology, 2022) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 19 |
According to the “image generation trilemma” shown in the figure, what analytical conclusion can be drawn about the relative strengths of VAEs, GANs, and DDPMs in medical image synthesis?
|
2. GANs provide a between image quality and diversity but may suffer from mode collapse. |
|
Generative Adversarial Networks (GANs) are known for producing high-quality medical images with sharp structural details, making them useful for tasks such as MRI reconstruction and CT image enhancement. However, compared to VAEs and DDPMs, GANs can struggle to maintain diversity because the generator may converge to producing limited patterns an issue known as mode collapse. This trade-off fits within the “image generation trilemma,” where no single model simultaneously maximizes quality, diversity, and reliability. |
From: Kazuhiro Kudo et al., “Generative Adversarial Networks for Medical Image Synthesis: Balancing Image Quality and Diversity”
(IEEE Transactions on Medical Imaging, 2020) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 20 |
Based on Figure, what analytical conclusion can be drawn regarding the distribution of cardiovascular disease (CVD) subtypes across East Asian countries?
|
3. China has the lowest proportion of stroke-related deaths among East Asian nations. |
|
The conclusion that “China has the lowest proportion of stroke-related deaths” would imply that cardiovascular mortality in China is dominated more by ischemic heart disease than by cerebrovascular disease. Such an analytical interpretation emphasizes that differences in lifestyle, population health, and treatment availability can shift the distribution of CVD subtypes across East Asian countries. |
From: Zhou et al., “Mortality, Morbidity, and Risk Factors in East Asia: A Comparative Assessment of Cardiovascular Disease Burden”
(The Lancet Global Health, 2021) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|