Table 2.

Performance of the SVM machine learning algorithm in discriminating between the different group constellations.

	Group differentiation
	ND vs. ASD	ND vs. ASD + ADHD	ADHD vs. ASD	ADHD vs. ASD + ADHD	ND vs. ADHD	ASD vs. ASD + ADHD
All-feature model: AUC (Sensitivity, Specificity)	0.91 (0.90, 0.88)	0.86 (0.41, 0.72)	0.91 (0.68, 0.73)	0.84 (0.89, 0.78)	0.62 (0.04, 0.39)	0.42 (0, 0)
Five-feature model: AUC (Sensitivity, Specificity)	0.77 (0.88, 0.81)	0.86 (0.51, 0.78)	0.91 (0.73, 0.82)	0.89 (0.90, 0.87)	0.40 (0, 0)	0.48 (0, 0)
DeLong test (all-feature vs. five-feature model): p-value	< 0.001	ns	ns	0.007	< 0.001	ns
Five most discriminative ADOS/ADI features (rank ordered):	Facial expressions directed to others Amount of reciprocal social communication Unusual eye contact Imaginative play (ADI) Hand and finger mannerisms (ADI)	Unusual eye contact Head shaking (ADI) Hand, finger and other complex mannerisms Shared enjoyment in interactions Compulsions/Rituals (ADI)	Facial expressions directed to others Stereotyped/ idiosyncratic use of words or phrases Amount of reciprocal social communication Conversation Friendships (ADI)	Quality of social overtures Response to approaches of other children (ADI) Descriptive, conventional, instrumental or informational gestures Hand and finger mannerisms (ADI) Conversation	Overactivity Abnormal development at or before 36 months (ADI) Tantrums, aggression, negative or disruptive behavior Offering comfort (ADI) Amount of reciprocal social communication	Asks for information Offers information Pronominal reversal (ADI) Language production and linked nonverbal communication Insight

Additionally, a test with a sensitivity (i.e., positive test result in presence of the condition) and specificity (i.e., negative test result in absence of the condition) of around 75% to 90% can be considered to have reasonable to good diagnostic performance.

ND no psychiatric diagnosis, ASD autism spectrum disorder, ADHD attention-deficit/hyperactivity disorder, ns not significant; AUC area under the curve: AUC < 0.7 (poor discrimination), 0.7 ≤ AUC < 0.8 (acceptable discrimination), 0.8 ≤ AUC < 0.9 (excellent discrimination), AUC ≥ 0.9 (outstanding discrimination) according to Hosmer et al. .⁴⁴.