| 1 |
How does the concept of “model as a dataset” reshape traditional data-sharing practices in medical imaging?
|
It enables sharing of learned model weights instead of sensitive raw images. |
|
Traditional models required actual images in order to be trained, which meant that the data of, in some cases, actual patients had to be sent across multiple networks, which may cause a leak or harm patient privacy. By sharing weights/parameters of models, patient data can be protected, and data sharing may be more efficient as well. |
This can be inferred by reading about the "model as a dataset" concept and comparing it to traditional methods. |
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?
|
Physics-informed models are more interpretable but computationally intensive. |
|
Physics-informed models rely on physic principles and mathematical equations to generate data that resembles biological situations, but it may be intensive in the fact that it requires a hgh level of expertise in the subject and can lean towards computational calculations rather than pattern recognition. |
This can be inferred by reading about how physics-informed models and statistical models function to see the differences and trade-offs. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 3 |
Why is “mode collapse” considered a critical problem in GAN-based medical image synthesis?
|
It reduces image realism and variety by producing repetitive outputs. |
|
The GANs model relies on the generator and discriminator to see what generated images are real and fake, and by using the same data it can lead to a repetition of the same, most realistic data. |
This can be inferred by reading how the GANs model works. |
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?
|
They better capture clinical accuracy and diagnostic relevance. |
|
Healthcare-specific metrics evaluate clinical accuracy and validty while FID and SSIM mainly evaluates how the synthetic data looks compared to real images. Healthcare-specific metrics are more safe for actual use. |
This can be inferred by reading how different metrics in evaluating generated data functions. |
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?
|
Higher realism may risk reproducing identifiable patient data. |
|
Image fidelity refers to how accurate the generated data may be. While training the models, it may copy the data of real images, such as the facial structures of a patient and replicate that into the synthetic data in order to make it realistic, which may cause the patient's identity to be known, compromising patient privacy. |
This can be inferred by reading the challenges of using generated synthetic data. |
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. |
|
Having the FDA evaluating and providing clearance for medical technology ensures that it is safe to use in real clinical practices because they provide a framework, structure, and standards as to how it should function and what outputs are received. |
The article cites the FDA and explains that it is important for the FDA to approve of the medical technology. It can be understood by analysing the passage. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 7 |
Which strategy would best mitigate demographic bias in generative models according to the article?
|
Applying diversity-aware training and fairness constraints |
|
Data can be more accurate if it is tailor-made for certain subgroups, because it specifies certain risk factors and situations. By applying diversity-aware training and fairness constraints, more researches can specify their data into making it most accurate for their targetted groups. |
The article mentions that by making diversity factors known, it can be a more accurate representation for the population. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 8 |
How do DDPMs exemplify versatility in healthcare image synthesis?
|
They can perform multiple tasks such as denoising, inpainting, and anomaly detection without retraining. |
|
DDPMs use the process of denoising to create high-quality samples, which can make them multifunctional for many process in healthcare, generating synthetic data of many various aspects. |
This can be inferred by reading the potentials and promises of generated data, as well as the function of DDPMs models. |
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?
|
It enhances training by providing diverse, realistic datasets without ethical breaches. |
|
Synthetic data, if trained and generated under strict rules and frameworks, gives way to many different benefits that may result in higher levels of patient privacy, as well as by increasing and predicting among many populations. |
Inferring from the passage on the potentials and promises of generated data. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 10 |
Why is regional calibration essential when applying risk prediction models across countries?
|
To adjust for population-specific incidence and lifestyle differences |
|
It is important that risk prediction models across countries use local data for training, as each population may vary in terms of genetics, lifestyles, mutations, environment, and much more. Generalising and average many sub-groups into the same prediction model risks overestimating or underestimating certain groups, as well as disregarding some important risk factors in minorities as it was not part of the major training data. |
This can be inferred by how Western models wrongly predicted ASCVD for East Asians, but models specified for each country proved more accurate for the people. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 11 |
What analytical conclusion can be drawn when comparing the China-PAR and Framingham models?
|
China-PAR uses local epidemiological data, leading to improved predictive validity. |
|
The Framingham model uses US studies and cohorts, which overestimates the risk for EA individuals as they have different baselines. The China-PAR model uses data from their people and country, which allows them to be more specific, accurate, and include keyy risk factors for the Chinese people. |
This can be inferred by reading the Framingham risk prediction model and comparing it to the China-PAR model and accuracy. |
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?
|
Japan’s low CVD mortality suggests effective prevention and healthcare systems. |
|
Compared to other countries in EA, Japan as a lower CVD mortality rate, so it can be inferred that they may have better screening, healthcare systems, or even a more healthy way of life among the Japanese people |
This can be seen by comparing the rates of CVD mortality among EA countries. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 13 |
What analytical limitation arises when using Western-derived coefficients in East Asian models?
|
It introduces systematic overestimation of ASCVD probability. |
|
Western populations tend to have a higher baseline of ASCVD compared to EA populations, causing there to be an overestimating of ASCVD risk in EA people since the training data was based mainly on Western cohorts. |
This was inferred reading the results of the Framingham model on East Asians (10 years lower) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 14 |
What policy implication can be derived from country-specific risk models?
|
They allow for targeted national prevention programs. |
|
Country-specific risk assessment allows the government to determine what are the leading causes of disease in their country and launch tailor-made preventions to reduce illness within the country, specific to the lifestyles and other factors of the population. |
This was inferred by interpreting the information of the article and applying it to real-world knowledge in public health safety. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 15 |
If a model excludes socioeconomic variables, what analytical consequence might occur?
|
Ignored non-biological determinants of disease |
|
Socioeconomic variables play a role in determining the conditions of each individual and their health. For example, wealthy people may access treatments and healthcare more easily, making their health significantly better than those who cannot. It is another form of bias and generalisation. |
This was inferred by reading the article and applying real-world considerations. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 16 |
How might AI improve next-generation ASCVD risk prediction in East Asia?
|
By integrating multimodal data, including imaging and lifestyle information |
|
Using AI may improve it by using training data to predict future statistics, by analysing current lifestyles to provide specific preventions or treatments, and it may be used in many situations for many diseases or functions in healthcare. |
This was inferred by reading about the future directions, potentials, and promises in regards to generated data and the usage of AI in the future. |
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?
|
Mortality differences reflect varying effectiveness of national prevention programs. |
|
East Asians may have similar lifestyles or conditions in terms of risks in health. It could be inferred that South Korea's natonal prevention program is more effective due to their lower CVD mortality rates. |
This can be inferred by looking at the graph comparing dfferent countries of EA, as well as real-world inferrences. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 18 |
What is the most logical future direction for improving ASCVD models across East Asia?
|
Establishing multinational data-sharing platforms to harmonize regional models |
|
The best direction to improve ASCVD models is to specify the data to fit the population of each country to find more accurate risk factors and being able to launch nationwide prevention interventions. |
Inferred from future directions and how specified data is more accurate for risk prediction models. |
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?
|
VAEs and DDPMs perform identically in generating high-fidelity images. |
|
From the chart, VAEs and DPPMs leans towards quality. |
Inferred from the given chart. |
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?
|
Stroke dominates as the primary cause of CVD death in all East Asian countries equally. |
|
The chart provides information that stroke (red) is most frequent. |
Analysing the given chart. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|