Novel Use of Natural Language Processing (NLP) to Predict Suicidal Ideation and Psychiatric Symptoms in a Text-Based Mental Health Intervention in Madrid
Table 2
Structured variable predictors of suicidal ideation and average GHQ-12 ≥ 4 (,453).
Suicidal ideation ever (OR, 95% CI)
value
GHQ ≥ 4 (OR, 95% Cl)
value
Age (cont)
1.02 (1.01–1.03)
<0.001
0.99 (0.98–1.00)
0.2
Female
1.12 (0.84–1.49)
0.456
1.02 (0.74–1.40)
0.913
Average nightly sleep (hours)
1.11 (1.01–1.22)
0.037
0.98 (0.88–1.09)
0.735
Sleep quality (0–100)
0.99 (0.99–1.00)
0.094
0.99 (0.98–1.00)
0.029
Anger rarely (0–100)
0.97 (0.96–0.98)
<0.001
0.98 (0.97–0.99)
<0.001
Changes in appetite (0–100)
1.00 (1.00–1.00)
0.482
1.00 (0.99–1.00)
0.793
Medication adherence (0–100)
1.00 (1.00–1.00)
0.746
1.00 (0.96–1.00)
0.912
WHO_5 wellbeing scale
0.96 (0.95–0.97)
<0.001
0.94 (0.93–0.95)
<0.001
Both models were completed using logistic regression (STATA 14 software).