| 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. |
|
This concept allows researchers to exchange trained model parameters instead of actual medical images, which keeps patient data private. It still lets others use the knowledge within the model while avoiding exposure of personal or hospital-specific information. |
This approach is based on the concept of privacy-preserving AI and federated learning, as described in this research https://www.nature.com/articles/s41746-020-00323-1 , which promotes secure data collaboration without sharing raw patient records. |
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. |
|
Physics-informed models combine scientific laws with data, making their predictions more understandable and realistic. However, this also makes them slower and more complex to train than regular statistical models, which only rely on correlations in data. |
This trade-off is discussed in Raissi et al. (2019), Journal of Computational Physics, explaining that physics-informed neural networks improve interpretability but require greater computational power.
Reference: https://www.sciencedirect.com/science/article/abs/pii/S0021999118307125 |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 3 |
Why is “mode collapse” considered a critical problem in GAN-based medical image synthesis?
|
2. It reduces image realism and variety by producing repetitive outputs. |
|
Mode collapse means the AI keeps making almost the same image again and again. This is bad for medical use because doctors need different examples to study real patient differences. |
Goodfellow et al. (2014) explained that mode collapse lowers diversity and makes generated images less realistic. (NeurIPS 2014)
|
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 metrics don’t just check if images look real they check if the image actually helps with diagnosis. That makes them more useful for doctors and medical AI than general image scores. |
Yi et al. (2019) mentioned that clinical metrics focus on diagnostic value instead of just visual quality. |
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?
|
1. Higher realism may risk reproducing identifiable patient data. |
|
When AI makes medical images that look very real, sometimes it can accidentally copy parts from real patients. This makes it possible for someone to recognize who the data came from, which is a privacy problem. |
Researchers (Chen et al., 2022) said that realistic synthetic data can still leak information about real people if the model remembers patient details. |
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?
|
1. It establishes a framework for validating synthetic data equivalence in clinical use. |
|
The FDA’s approval is important because it shows that synthetic data can be safe to use in hospitals if it passes certain quality tests. It also helps create clear rules for other AI systems in the future. |
According to the U.S. FDA (2023), this approval helps build trust and gives a standard for testing AI-generated medical data before using it in real life.
|
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 |
|
Adding fairness rules and making sure the AI learns from many different groups helps reduce bias. If the data comes from only one type of population, the model might not work well for others. |
According to Yang et al. (2023, The Lancet Regional Health), improving data diversity and using fairness-aware methods helps minimize demographic bias in healthcare AI models. |
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 are flexible because they don’t need to be trained again for each new task. They can fix noisy images, fill missing parts, and even detect unusual patterns all from the same model. |
Ho et al., NeurIPS Conference Paper, 2020. The paper introduced diffusion models and showed that they can adapt to several imaging tasks without retraining, making them efficient and general-purpose tools for healthcare AI. |
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?
|
2. It enhances training by providing diverse, realistic datasets without ethical breaches. |
|
Using AI-generated medical images helps students and researchers learn from realistic and varied examples without needing real patient data. This supports both education and research while protecting privacy and avoiding ethical problems. |
Lee et al., Nature Medicine, 2022. The study shows that synthetic medical images can improve medical training and research collaboration while reducing ethical risks from using real patient data. |
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 |
|
People in different countries have different genetics, diets, and lifestyles. Calibrating models for each region helps make risk predictions more accurate and fair. Without adjustment, models built for one country might not work well in another. |
Yang et al., The Lancet Regional Health – Western Pacific, 2023. The research highlights that cardiovascular risk models need regional calibration to reflect variations in population health and lifestyle across East Asia.
|
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 11 |
What analytical conclusion can be drawn when comparing the China-PAR and Framingham models?
|
2. China-PAR uses local epidemiological data, leading to improved predictive validity. |
|
The China-PAR model was created using data from Chinese people, which makes it more accurate for predicting cardiovascular risk in East Asian populations. The Framingham model, on the other hand, was made based on Western data, so it may not fully fit Asian lifestyles or genetics. |
Yang et al., The Lancet Regional Health – Western Pacific, 2023 . The study found that locally developed models like China-PAR perform better for Asian populations compared to Western-based models. |
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?
|
1. Japan’s low CVD mortality suggests effective prevention and healthcare systems. |
|
Japan has one of the lowest rates of cardiovascular deaths among East Asian countries. This shows that its strong public health policies, healthy diets, and early screening systems help prevent heart disease and stroke better than in nearby countries. |
World Heart Federation, 2022 Global cardiovascular reports highlight Japan’s success in prevention strategies and efficient healthcare systems as key reasons for its low CVD mortality rate. |
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. |
|
When models developed in Western countries are applied to East Asian populations, they often predict higher cardiovascular risks than the actual rate. This happens because the data used to build those models come from people with different lifestyles, diets, and genetics. So, using them without adjustment can lead to overestimation of risk. |
Yang et al., The Lancet Regional Health – Western Pacific, 2023. The study reported that Western-based risk equations like Framingham tend to overpredict ASCVD risk among Asian populations due to epidemiological differences. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 14 |
What policy implication can be derived from country-specific risk models?
|
1. They allow for targeted national prevention programs. |
|
Country-specific models help governments design health programs that fit local populations. For example, if certain risk factors are common in one country, public health policies can focus on reducing those risks directly, making prevention more effective. |
World Heart Federation, 2022 .Global cardiovascular policy reports emphasize that regionally tailored risk models can guide more accurate prevention strategies and resource allocation for national health systems. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 15 |
If a model excludes socioeconomic variables, what analytical consequence might occur?
|
2. Ignored non-biological determinants of disease |
|
If the model doesn’t include socioeconomic factors like income, education, or access to healthcare, it misses important reasons why people might get sick. These factors affect lifestyle and health behavior, so ignoring them makes predictions less realistic. |
World Heart Federation, 2022. The report highlights that social and economic conditions strongly influence cardiovascular health and should be included in risk prediction models to improve fairness and accuracy.
|
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 combine different types of data such as medical images, genetic data, and lifestyle habits to make more accurate predictions. This helps doctors understand risk from many angles instead of using just one factor like cholesterol. |
Yang et al., The Lancet Regional Health – Western Pacific, 2023. The study shows that AI-powered models using multimodal datasets improve ASCVD risk prediction for East Asian populations by capturing more complex health relationships. |
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. |
|
South Korea has a lower cardiovascular disease death rate because it has stronger prevention programs, better access to healthcare, and public awareness about healthy lifestyles. Mongolia’s higher mortality shows limited healthcare access and fewer national-level prevention policies. |
World Heart Federation, Global Cardiovascular Report 2022. Countries with well-developed prevention systems, like South Korea, have lower CVD mortality compared to developing nations such as Mongolia.
|
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 18 |
What is the most logical future direction for improving ASCVD models across East Asia?
|
1. Establishing multinational data-sharing platforms to harmonize regional models |
|
Since each East Asian country has different health patterns, combining data from multiple nations can help build more accurate models. This approach supports collaboration and allows AI systems to learn from diverse populations, improving fairness and prediction accuracy. |
Yang et al., The Lancet Regional Health – Western Pacific, 2023. The study suggests that multinational data integration is essential for creating reliable and generalizable cardiovascular prediction models across Asia. |
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 balance between image quality and diversity but may suffer from mode collapse. |
|
In the image generation trilemma, each model has its strengths and weakness
- VAEs are fast but usually produce blurry or low-quality images.
- DDPMs can create high-quality and diverse images, but they take a lot of time to generate.
- GANs stay in the middle - they balance realism and diversity well, which makes them useful for medical image synthesis. However, they sometimes face mode collapse, where the model generates similar images repeatedly. |
Karras et al., IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021. The study explains the trade-off between image quality, diversity, and speed in generative models, showing that GANs achieve a good balance but remain vulnerable to mode collapse. |
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?
|
1. Ischemic heart disease (IHD) accounts for a higher proportion of CVD deaths in Japan and South Korea compared with China, suggesting regional lifestyle or prevention differences. |
|
The chart shows that Japan and South Korea have a larger share of deaths from ischemic heart disease around 38–39% compared to China 41% stroke, 48% IHD. This suggests that lifestyle, diet, and healthcare systems differ between countries. For example, Japan’s population may have better stroke prevention but higher rates of heart disease due to aging and diet patterns. |
Yang et al., The Lancet Regional Health – Western Pacific, 2023. The study reports variation in CVD subtypes across East Asia, linking these differences to national prevention programs, diet, and healthcare quality. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|