| 1 |
An AI model for depression detection shows high accuracy in laboratory testing but performs poorly in community clinics. Based on the article, what is the most likely explanation?
|
3. AI algorithms cannot process emotional data |
|
Artificial intelligence models, unlike humans, have subpar or even no capability to detect emotions. Furthermore, clinical experts are skeptical on using artificial intelligence as a basis for their diagnosis. |
Source 11 (Zhang Z et al.), Source 10 (Timmons AC et al.) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 2 |
Which scenario best illustrates the ethical risk of over-reliance on AI in mental health care?
|
3. AI making autonomous decisions without human oversight |
|
Current medical datasets regarding mental health are not sufficient due to their reduced sample size. Not only that, biased datasets that lack cultural, ethnic or linguistic diversity can cause problems when dealing with more than one type of patient. Artificial intelligence models have an opaque nature, described as "black boxes," which poses clinical concerns. If left alone to its own devices, future harm to patient health can arise as a consequence. |
Source 10 (Timmons AC, et al), Source 98 (Mautant T et al). |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 3 |
If an AI mental health tool consistently underdiagnoses symptoms in minority populations, which corrective strategy aligns best with the article’s recommendations?
|
5. Disabling automated predictions |
|
Besides having used skewed data that contains little to no data diversity, the issue of ethical concerns, namely weakness to being misused and privacy concerns, pose as barriers to AI usage. Furthermore, the AI models have been reported that they act like "black boxes" when used. It is no doubt that using automated predictions over human-made predictions should be cancelled until further developments can be made. If so, the automated predictions still should be used as an assisting tool to psychiatrists, NOT a full on replacement. Furthermore, public trust has been skewed, which creates more of an incentive to remove the automatic predictions entirely. |
Source 98 (Mautang T), Source 105 (Looi JC et al.), Source 113 (Shen J et al.), Source 10 (Timmons AC et al.) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 4 |
Why does the article argue that AI tools are more suitable for screening rather than standalone intervention in mental health?
|
3. Mental health conditions are dynamic and context-dependent |
|
Certain conditions that effect dynamic AI tracking such as unpredictability in emotion, cultural and social context, and challenges with clinical integration limit the AI tool to being used as an assistant in screening. Furthermore, if the AI tools are consistently unregulated through lack of developer transparency, unexpected results could appear which can worsen patient conditions. |
Source 10 (Timmons AC et al.), Source 101 (Roudriges F.), Paragraph on Data-centric AI (Ali et al.) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 5 |
Which AI feature would most increase clinician trust, according to the article?
|
3. Explainable decision pathways |
|
With explainable decision pathways, developers are more likely to be held accountable for their mistakes, and can be fixed effectively in a short amount of time. When holding the source of the problems accountable, it allows for more customers to trust in the tool used to assist in making predictions of patient status and further treatment. |
Source 124 (Kocak B et al.) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 6 |
What unintended consequence may arise if AI mental health apps are widely adopted without regulation?
|
2. Increased stigma due to data misuse |
|
In contrast to the statement above, where developers are held accountable, instead the following scenario will have a developer never accounted for. Bugs within code run rampant, and loopholes are constantly being discovered. With such dangers in place, it is not a surprise that people with malicious intent will begin using data breaches for their own interests. |
Source 105 (Jooi JC et al.), Source 106 (Javaid M et al.), Source 107 (Sig HBD et al.) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 7 |
Which model of care best aligns with the authors’ proposed future of AI in mental health?
|
3. Human–AI collaborative decision-making |
|
Although artifical intelligence has a swiss army knife worth of capabilities including machine learning and deep learning, it lacks certain qualities of a human psychiatrist. Firstly, a human psychiatrist is able to read emotions, while AI can do so at a limited capacity of not at all. Secondly, a human psychiatrist is able to interpret more based on their current knowledge of minority groups or cultures. However, AI is solely based off of the training data given by developers. If a mishap in dealing with a minority group is found, then the entire model would have to be re-trained in order to be equipped with such data to deal with those scenarios. However, humans are unable to process loads of data at once, which is where artificial intelligence steps in to do such work for them. |
Source 10 (Timmons AC et al), Source 100 (Shumate J.), Source 8 (Etkin A.) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 8 |
Why might an AI system incorrectly label normal stress as a mental disorder?
|
2. Absence of physiological data |
|
Small data sizes can lead to skewed data results. Although it is equipped with a wide variety of tools and resources, it cannot be an effective tool if trained using a small training dataset. |
Source 9 (Bouderhem R.) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 9 |
Which research focus would most effectively improve clinical adoption of AI mental health tools?
|
2. Enhancing data privacy frameworks |
|
Even though other problems such as speed or reliability are in the scope of developer concerns, data privacy is an even bigger one. Since digitized mental healthcare adheres to the policy of privacy and consent, one shall not be surprised that they are lying to their customers when they have an unregulated, weak data privacy system. |
Source 112 (Witowski K et al.), Source 97 (Ludenberg SM et al.) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 10 |
From a public health perspective, what is the strongest justification for integrating AI into mental health systems?
|
3. Expanding access and early detection |
|
By expanding early access, more people in rural areas are more exposed to mental healthcare, therefore they can get the treatment they need. Early detection expansion allows for psychiatrists to get a better insight on their patients. |
I have assisted my friend in a science competition, and he noticed that in rural areas, people receive a lot less mental healthcare due to lack of trained mental health professionals and budget constraints. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 11 |
Why are mRNA vaccines particularly suited for rapidly mutating respiratory viruses?
|
3. Their sequences can be rapidly updated |
|
Unlike DNA, mRNA sequences only require the viral genotype to create the vaccine. With this, virus mRNA vaccines require a shorter development cycle, so they can be used on rapidly mutating viruses in a short manner of time. |
DNA analysis requires complex techniques from PCR, Gel electrophoresis and eventually gene sequencing. Due to RNA being single stranded as well as limited to a specific set of proteins, scientists do not need to further analyze its sequence. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 12 |
What would most likely happen if lipid nanoparticles failed to protect mRNA during delivery?
|
3. Rapid degradation of mRNA |
|
Without the nanoparticles' stable nanostructure, the mRNA would be exposed to the extracellular fluid, which would degrade the mRNA molecule before it reaches the target cell(s). |
Source 27 (Hou X et al.) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 13 |
Which limitation most restricts global deployment of current mRNA vaccines?
|
1. Low immunogenicity |
|
If vaccines are not triggering an immune response or a weak one, then the body is not able to combat the disease. Conversely, having too low immunogenicity can cause extra side effects such as inflammatory responses. |
Source 14 (Chaudhary N et al.) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 14 |
Why does self-amplifying mRNA (saRNA) reduce required vaccine doses?
|
2. It replicates within host cells |
|
saRNA functions sort of like a virus, with the fact that it can self-replicate and spread throughout the host. When little saRNA is injected into the host, it can cause the RNA to be spread to more cells with relative ease, eliminating the need for more vaccines. |
src 23 (Oda Y et al.) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 15 |
Which feature makes circular mRNA vaccines potentially transformative?
|
3. Enhanced stability without a cold chain |
|
When the need for a 5' cap and 3' poly-A tail is removed, it allows for the structure to be more stable, effectively eliminating the need for a cold chain. |
Source 24 (Xie J et al.) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 16 |
Why might respiratory-route vaccine administration improve protection against respiratory viruses?
|
2. It targets mucosal immunity at infection sites |
|
By targeting the infection sites, the vaccine can spread to a target area before other areas of the body, making the effect appear earlier on the virus' target area. |
Source 33, 34 |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 17 |
What is the primary advantage of combination mRNA vaccines?
|
3. Protection against multiple pathogens in one dose |
|
By having more than one antigen inside a single virus, it can effectively target multiple pathogens at once. |
Section 2.4 (Zheng et al. [AUTOR]) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 18 |
Why is post-market surveillance especially important for mRNA vaccines?
|
2. Rare adverse events may emerge over time |
|
Factors such as individual immune system differences and the novelty of the vaccine platform make it easier for adverse effects to appear. As a result, surveilling the patients after administration is necessary. |
After having a covid-19 or any other vaccine, you are required to stay in a quarantine zone for a specific amount of time before you are free to go. This is what the question calls "post-market surveillance," to check for any adverse effects on an individual. |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 19 |
Which trend best reflects the future direction of mRNA vaccine research described in the article?
|
3. Multivalent and broadly protective vaccines |
|
Combination vaccines are discussed in the article. They are in research by moderna and other big pharma corporations, as well as institutes such as UPenn are working on vaccines that can target 20 different variants of influenza. |
Sources 25, 26, 16 |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|
| 20 |
From a global health perspective, which improvement would most enhance equity in mRNA vaccine access?
|
3. Improved stability at room temperature |
|
Implementing circular RNA would be the corresponding development to choice #3. If cricRNA is widely implemented, cold storage platforms will be eliminated from the scene, allowing people in hot or rural areas to access the vaccines, ultimately improving their quality of life as a whole. |
Source 24 (Xie J et al.) |
7 |
-.50
-.25
+.25
เต็ม
0
-35%
+30%
+35%
|