Brainscape's Knowledge GenomeTM

Browse over 1 million classes created by top students, professors, publishers, and experts.


See full index

Clinical Informatics Board > Domain 4: Data Governance & Data Analytics > Flashcards

Domain 4: Data Governance & Data Analytics Flashcards

1
Q

What is the ACID test for reliable database (DB) transactions?

A

Atomicity - transaction is individisble

Consistency - database is always in a valid state

Isolation - 2 transactions run simultaneously will not interfere with each other

Durability - persistent data store will always contain up-to-date info

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is included in the OSI 7 layer model

A
  1. Application (e.g. POP3/IMAP4, HTTP, FTP, SSH and HTTPS)
  2. Presentation (Encryption, decryption, conversion to character sets like ASCII)
  3. Session (Lightweight Directory Access Protocol = LDAP, SSL)
  4. Transport (TCP, UDP)
  5. Network (IPv4, IPv6, DHCP)
  6. Data Link (Address Resolution Protocol = ARP)
  7. Physical (transmission of binary bits via copper wire, coaxial or fiber optic cable)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Global variables referenced by what symbol?

A

Up arrow symbol (later became a caret “^”)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

In an event monitor, a high test result is a?

A

Condition

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What is evoking strength in INTERNIST-1/QMR

A

PPV

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is the observed value for an item in a ML data set?

A

Label

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What is an example of an expert system?

A

CADUCEUS

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Examples of reinforcement learning

A

Markov decision process (use with known model)

Monte Carlo (use when one or more elements are unknown)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What are measures of information retrieval success?

A

Precision (PPV) = % of returned documents that are relevant to the query

Recall (sensitivity) = % of all relevant documents in the corpus that were found

Fall-out (false positive rate) = % of irrelevant documents that are retrieved

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Difference between bias and variance

A

Bias = measure of inaccuracy
Variance = measure of imprecision

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is the null error rate

A

Rate of being wrong if you ALWAYS pick the majority class.

Ex) If majority class has 105 instance out of 165 total instances, null error rate = (165-105)/165

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What statistical models to use to screen for low incidence conditions

A
  • Fbeta score
  • Matthews Correlation Coefficient
  • Stratified K-fold cross-validation
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Measures to evaluate regression methods (numerical output)

A
  • Root mean squared error (RMSE): lower = better fit
  • Correlation coefficient (r): strength of relationship between x and y on a scatter plot. No correlation is r = 0.
  • Coefficient of determination (r2 squared): Goodness of fit. Represents % variation in y that is not explained by variation in x. 100% = perfectly fit model.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Measures to evaluate classification methods (categorical output)

A
  • Cohen’s Kappa: measure of how well the classifier performs as compared to random chance
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Measure used to evaluate both supervised & unsupervised models

A
  • Receiver Operated Characteristic (ROC) curve: Plot sensitivity against 1-specificity
  • Area Under the Curve (AUC): AKA concordance (c) statistic. AUC = 1 –> perfect discrimination.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is the most important, time-consuming and expensive part of developing ML model

A

Gathering appropriate data (instances)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

Difference between validation and testing data set

A
  • Validation data set used to evaluate PRELIM model to tune model
  • Testing data set used for final evaluation of model. No further change is anticipated.
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

General rule for feature to instance ratio

A

Select <= 1 feature for each 10 instances in the development data set

High # of features –> overfitting

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

Methods for feature selection

A
  • Forward selection (iterative inclusion)
  • Backward selection (iterative removal)
  • Stepwise selection (combination of forward & backward selection)
  • Forced inclusion
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

Methods for reducing number of features when an unsupervised model is overfit

A

Dimensionality reduction method

21
Q

Methods for validating model when data (instance) is limited

A
  • K-fold cross-validation: training data randomly partitioned into equal # (k) subsamples (folds)
  • Leave One Out Cross-Validation (LOOCV): extreme case of k-fold
  • Bootstrapping
22
Q

Methods for optimizing (aka tuning) model

A
  • Tune hyperparameters
  • Dimensionality reduction
  • Regularization: force algorithm to build a less complex model (i.e. more generalizable –> less likely to overfit data)
23
Q

Types of deployed model

A
  • Static model (most common in medicine)
  • Incremental/continuous model
24
Q

ML algorithm that can be used for both supervised and unsupervised algorithms

A

Neural networks (aka connectionist systems)

25
Q

ML algorithm used for supervised algorithms

A
  • Regression Methods
  • Classification Methods
  • Ensemble Methods
26
Q

ML algorithm used for unsupervised algorithms

A
  • Clustering Methods
  • Association Rules
  • Dimensionality Reduction Methods
27
Q

How many hidden layers does Deep Artificial Neural Network usually have?

A

> 3

28
Q

Different types of ANN

A
  • Feed-forward network
  • Multilayer Perception (MLP)
  • Convolutional neural Network
  • Recurrent neural network (RNN)
  • Generative Adversarial Network (GAN)
29
Q

Common output of convolutional neural network (CNN)

A

Image classification and/or image feature selection (e.g. saliency map showing which features considered more relevant)

30
Q

Common use of recurrent neural network (RNN)

A

NLP

31
Q

Common use of generative adversarial network

A

Use to generate DeepFake images and simulate cat-and-mouse fraud schemes

32
Q

How does generative adversarial network work?

A

Pairs of deep learning neural networks (discriminator network vs. generator network) trained in tandem repeatedly

33
Q

What are examples of classification models

A
  • Logistic regression
  • Naive Bayes Classifier
  • Support Vector Machines
  • k-Nearest Neighbor
  • Decision Trees (most common method)
34
Q

What does an internal and leaf node represent in a decision tree

A

Internal node = 1 feature (independent variable)

Leaf node = outcome class (dependent variable)

35
Q

Method for checking quality of decision tree model

A
  • Gini impurity (0 with single class populations)
  • Entropy (high when large number of evenly mixed classes)
36
Q

Types of ensemble methods

A

Parallel ensembles

Sequential (series) ensembles

37
Q

What are methods to create diversity in parallel ensembles to help decrease overfitting

A
  • Bagging (bootstrap aggregating): boostrap each model in ensmble
  • Random subspaces: Use random subset of features per model
  • Random forest: Ensemble of randomly selected decision trees to make a “forest.” Uses BOTH bagging and random subspaces.
38
Q

How is sequential (series) ensembles constructed?

A
  • Combines constrained (weak learner) models into a single strong learner
  • Constructed using boosting methods, which uses weighted voting
39
Q

Difference between hard and soft clustering

A
  • Exclusive (hard) clustering: An instance can only belong to 1 cluster
  • Fuzzy (soft) clustering: An instance can have more than 1 cluster assignment
39
Q

Types of boosting methods

A
  • AdaBoost (Adaptive Boosting): misclassified data from each algorithm have weights increased
  • Gradient boosting: similar to adaboost except model trains on residual errors of the previously run model
  • CatBoost
40
Q

Types of clustering methods

A
  • Hierarchical clustering: instances grouped based on similarities and differences
  • Probabilistic clustering: instances clustered based on probability that they belong to a particular distribution
  • K-Means clustering: Assign instances to manually defined number (K) of clusters based on similarity –> Compute distance of all instances in cluster from the centroid (center of the cluster) using a defined distance metric. Move instances closest to centroid to that k-cluster. Continue until centroids stop moving location.
41
Q

What is a silhouette coefficient

A

Ratio of cluster sum of squared error (SSE) & cluster separation

1: clusters well apart
0: clusters indifferent
-1: clusters assigned incorrectly

42
Q

Types of association rules

A
  • Market basket analysis (e.g. customers who bought X also bought Y)
  • Apriori algorithm: Uses a hash tree to count item sets navigating through data set in breadth first manner
43
Q

Dimensionality reduction methods

A
  • Principle Components Analysis (principle component = axis through data that is a function of contribution of variability in a population. Each principle component has to be orthogonal with all other principle components)
  • Singular value decomposition
  • Autoencoders
44
Q

What is denormalization of data

A

Intentional duplication of data to improve database performance

45
Q

3 Database integrity requirements

A
  1. Entity integrity: every table in the database has a unique primary key
  2. Referential integrity: Whenever a database column refers to a row in another table, that row exits
  3. Domain integrity: specific list of values that are acceptable for a particular column
46
Q

What is F1 score and how is it calculated?

A

F-Score (aka F-measure or F1 score) = harmonic mean between precision and recall

F1 = 2 * (Precision x Recall) / (Precision + Recall)

47
Q

What is tokenization?

A

Process of breaking documents into searchable items. Tokenization must be done before items are placed into an index.

Clinical Informatics Board (6 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2024 Bold Learning Solutions. Terms and Conditions