Glossary

Healthcare in 3D Terms and Phrases

  • Classification (Machine Learning) — In HCn3D classification occurs when patient journeys of others are counterfactually analyzed to categorize the attributes or features of a subject patient journey (e.g. whether the patient is likely to have heart disease). A type of reinforcement learning built on supervised and unsupervised learning techniques.
  • Clustering (Machine Learning) — Is the result of unsupervised learning methods that attempt to find similarities in unlabeled data sets. HCn3D uses clustering methods to discover previously unknown correlations or classifications for use in supervised learning methods.
  • Decision Tree (machine learning) — Is a flowchart-like structure in which each “decision” represents the evaluation (or test) of a feature (e.g. whether a patient is male or female), each branch of the tree represents an outcome of the test, and each “leaf” (last outcome on the branch) represents a classification (e.g. Males over 40 years old who have heart disease).
  • Feature (machine learning) — Is a property or characteristic that can be measured or categorized in the data being analyzed (e.g. person's age).
  • Feature Engineering (machine learning) — Is a process by which knowledge of the data (or business domain) is used to create new (non-natural) features in an effort to improve the function (predictive power and/or efficiency) of machine learning algorithms?
  • Feature Hierarchy (Machine Learning) — The resulting feature output of the traversal of a deep-learning neural network. The output of each layer aggregates and recombines forming recognizable patterns from high-dimensional data sets. For example, pixels in a picture combine to form facial features (eyes, mouth, nose, etc…) which in turn combines to form a recognizable face. In HCn3D this method can be applied to the free-from text fields in the patient EHR to obtain classifiable information that in turn is used in Supervised Learning techniques.
  • Feedforward Neural Network (Machine Learning) — The simplest form of a neural network in which the information moves in only one direction through the node. A feedforward neural net is considered “deep” when it constitutes 3 or more layers.
Page 1 of 2 1 2