pros and cons of AI Flashcards
Precision medicine can help with:
cancer, cardiovascular disease, neurological disorder, psychiatry treatments
acronym CCNP
A study found that black patients indicated ____ levels of pain that could not be described by their x-rays
Higher
Why does 97% of healthcare data go unused?
Because it is unstructured
What system was designed to categorize clinical AI research published on the Medline database? (showed 25.3% of medical papers were in oncology)
NLP system
Which 2 fields rely heavily on imaging for diagnosis?
Oncology and neurology
acronym ON
What was the original problem of the Sepsis Watch, when used in Duke University’s Hospital?
Alarm fatigue - doctors ignored alerts sent by nurses
What app improved the Thrombolysis service in the Royal Berkshire NHS Foundation Trust? What did the app do?
Brainomix: uses ML to view brain CT scans anywhere on the app
Which healthcare professions are most likely to lose their jobs with the rise of AI in healthcare & why?
Radiologists and pathologists because of the digitisation & high volume of data available
Why is it difficult to identify bias in AI?
Because it is a ‘black box’ (the internal workings as it generates output from input is unknown)
What type of bias occurs when data is not representative of the entire population or is biased in some way?
Data bias
What type of bias occurs when the algorithm itself is biased, resulting in unfair outcomes?
Algorithmic bias
How does user bias occur?
It occurs when users introduce their personal biases into a system and thus receive biased results or implement results in biased ways
Procedural fairness can combat what type of bias?
Algorithmic bias (when a system has procedural fairness, the processes used to make decisions are considered fair)
What type of fairness is concerned with the outcomes of decisions?
Distributive fairness
If a system is representationally fair, what type of bias is it unlikely to exhibit?
Data bias
