Skip to main content

Table 4 Summarized description of the features chosen by the feature reduction process for module 3

From: Sparsifying machine learning models identify stable subsets of predictive features for behavioral detection of autism

Feature

Category

A2

Language (intonation, volume, rhythm)

A4

Language (words/phrases)

A8

Language (conversation)

B3-missing

Language: examiner could not test nonverbal communication linked to language production

B2

Social interaction (facial expressions)

B7

Social interaction (social overtures)

B8

Social interaction (social overtures, attention)

D4

Behavior (repetitive behavior)

D3-missing

Behavior: examiner could not test self injurious behavior

Gender

Gender of the patient