Considering patient safety in autonomous e-mental health systems – detecting risk situations and referring patients back to human care

Table 4 Multilevel analysis results on intent to contacting the agent again, and feeling agent has heard them in situations that user reject advice to seek health care (4, 7, and 8): model comparisons

Model comparison	n	χ ²	p
Intent to contact the agent again
Agent strategies (M₀ vs M₁)	152	χ²₂ = 1.00	.605
Sub set agent strategies
Accept rejection vs Persuasion (M₀ vs M₁)	99	χ²₁ = 0.59	.441
Accept rejection vs Facilitate (M₀ vs M₁)	102	χ²₁ = 0.03	.874
Feeling agent has heard them
Agent strategy (M₀ vs M₁)	152	χ²₂ = 16.60	<.001
Sub set agent strategy
Accept rejection vs Persuasion (M₀ vs M₁)	99	χ²₁ = 12.03	<.001
Accept rejection vs Facilitate (M₀ vs M₁)	102	χ²₁ = 0.003	.953

ISSN: 1472-6947