Can passive measurement of physiological distress help better predict suicidal thinking?