Application of AI for sepsis Flashcards
Describe the characteristics of sepsis.
- Severe immune response to infection
- Damage to organs
- 49 million cases per year
Describe the definition of:
- Artificial intelligence
- Machine learning
- Deep learning
- Artificial intelligence → a field, which combines computer science and robust datasets, to enable problem-solving.
- Machine learning → a branch of AI and computer science which focuses on the use of data and algorithms to imitate the way that humans learn, gradually improving its accuracy.
- Deep learning → a subset of machine learning, which is essentially a neural network with three or more layers. These neural networks attempt to simulate the behavior of the human brain—albeit far from matching its ability—allowing it to “learn” from large amounts of data.
Supervised learning falls under the scope of machine learning. How is supervised learning defined?
It is defined by its use of labeled datasets to train algorithms that to classify data or predict outcomes accurately.
Supervised learning can be used for early detection of sepsis. What are the challenges of the use of supervised learning for the early detection of sepsis?
- Prediction
- Incorporation bias (occurs when the gold standard uses (or incorporates) the test you are studying.)
- Optimistic
Why not just use blood cultures to predict sepsis?
- A yield of 1.4-17.4% positives
- Contamination rate of 30-55%
- Impact on managment → 0.18%
Unsupervised learning falls under the scope of machine learning. How is unsupervised learning defined?
Unsupervised learning is a type of machine learning in which models are trained using unlabeled dataset and are allowed to act on that data without any supervision.
What is the difference between supervised- and unsupervised learning?
- In supervised learning, input data is provided to the model along with the output.
- In unsupervised learning, only input data is provided to the model.
You are developing a prediction tool for sepsis. How would this tool look like if:
- modelled according to supervised learning
- modelled according to unsupervised learning
- Supervised learning → the predictive variables of sepsis are incorporated into the prediction tool such as liver function and blood values (note: incorporation bias!). With this, sepsis can be predicted based on the chosen predictive variables.
- Unsupervised learning → here, the algorithm is given an input dataset containing patients with different types of medications. The algortihm is never trained upon the given dataset, which means it does not have any idea about the features of the dataset. The task of the unsupervised learning algorithm is to identify the patients’ features on their own. Unsupervised learning algorithm will perform this task by clustering the patient dataset into the groups according to similarities between the patients.
What is important for developing an unsupervised algorithm regarding the clusters that are chosen?
The clusters should be biologically:
- plausible
- identifiable
- reproducible
What is reinforcement learning?
A machine learning training method based on rewarding desired behaviors and/or punishing undesired ones (based on human learning). Here, based on the feedback the algorithm receives, it alters it action for the next time.
Name a situation where you would use the following algorithm:
- reinforcement learning
- supervised learning
- unsupervised learning
- reinforcement learning → to make a choice for a certain situation. For example: when you need to make multiple choices within a certain timeframe.
- supervised learning → used to make decisions/predictions about prognosis.
- unsupervised learning → used to create different clusters within a certain population and to determine which treatment or outcome there should be.
Why is prediction of sepsis with the use of AI difficult?
The incorporation bias causes the tool to be highly sensitive to small deviating changes in certain predictive variables. Therefore, the tool is way too sensitive to predict sepsis accurately.
