Table 3.
Statistical insights on differential language expression prior to an inpatient visit*.
Feature | Cohen’s d | p-value (corrected) | Mean diff-of-diff | 95% CI |
---|---|---|---|---|
Inpatient visits | ||||
Change in Linguistic Style/ Mental Well-being | ||||
Categories that increase in usage before inpatient visit | ||||
Depressive | 0.306 | 0.001 | 0.089 | [0.09, 0.52] |
Anxious | 0.286 | 0.001 | 0.084 | [0.07, 0.50] |
Family (‘baby’, ‘ma’,‘son’, ‘family’) | 0.306 | 0.001 | 0.003 | [0.09, 0.52] |
Health (‘tired’, ‘pain’, ‘sick’, ‘ill’) | 0.255 | 0.032 | 0.001 | [0.04, 0.48] |
Categories that decrease in usage before inpatient visit | ||||
1st person singular (‘lol’, ‘:)’, ‘b’) | −0.392 | <0.001 | −0.014 | [−0.61, −0.17] |
Informal Speech (‘lol’, ‘:)’, ‘b’) | −0.392 | <0.001 | −0.014 | [−0.61, −0.17] |
Hear (‘say’, ‘hear’, ‘listen’, ‘heard’) | −0.365 | 0.003 | −0.001 | [−0.58, −0.15] |
Affiliation (‘we’, ‘our’, ‘friends’) | −0.361 | 0.001 | −0.003 | [−0.58, −0.14] |
Extraverted | −0.357 | <0.001 | −0.080 | [−0.57, −0.14] |
Drives (‘up’, ‘get’, ‘love’) | −0.354 | <0.001 | −0.006 | [−0.57, −0.14] |
Netspeak (‘u’, ‘lol’, ‘da’, ‘smh’) | −0.35 | <0.001 | −0.01 | [−0.57, −0.13] |
Nonfluencies (‘ugg’, ‘well’, ‘oh’, ‘er’) | −0.335 | 0.001 | −0.001 | [−0.55, −0.12] |
Leisure (‘fun’, ‘play’, ‘nap’) | −0.321 | 0.006 | −0.002 | [−0.54, −0.1] |
Reward (‘get’, ‘take’, ‘best’, ‘win’) | −0.314 | 0.001 | −0.002 | [−0.53, −0.1] |
Affective Processes (‘:)’, ‘ugh’, ‘happy’) | −0.242 | 0.023 | −0.004 | [−0.46, −0.03] |
Positive Emotion (‘love’, ‘good’, ‘lol’, ‘better’) | −0.228 | 0.013 | −0.004 | [−0.44, −0.01] |
Swear Words (‘a**’, ‘f**k’, ‘hell’, ‘wtf’) | −0.209 | 0.023 | −0.002 | [−0.43, 0.00] |
Change in Linguistic Topics | ||||
Topics that increase in usage before inpatient visit | ||||
check, yes, doctors, office, waiting, appointment | 0.504 | <0.001 | 0.001 | [0.29, 0.72] |
hospital, pain, surgery, blood, meds, nurse | 0.380 | <0.001 | 0.001 | [0.16, 0.6] |
baby, mommy, girl, son, boy, daughter | 0.377 | <0.001 | 0.001 | [0.16, 0.59] |
days, more, two, weeks, until, couple | 0.301 | 0.014 | 0.001 | [0.08,0.52] |
even, still, tho, yet, blah, mad | 0.275 | 0.006 | 0.001 | [0.06, 0.49] |
kids, child, their, children, mother, father | 0.275 | 0.006 | 0.001 | [0.06,0.49] |
hurt, head, bad, body, stomach,:(, ugh | 0.267 | 0.026 | 0.001 | [0.05, 0.48] |
Topics that decrease in usage before inpatient visit | ||||
calling, phone, answer, hear, ooo, talking | −0.488 | <0.001 | −0.001 | [−0.71, −0.27] |
better, feeling, little, hope, bit, type | −0.429 | <0.001 | −0.001 | [−0.65, −0.21] |
lol, ha, ctfu, lmao, funny, haha | −0.429 | <0.001 | −0.001 | [−0.65, −0.21] |
cool, funny, tho, used, remember, seem | −0.395 | <0.001 | <0.001 | [−0.61, −0.18] |
no, what, matter, how, always, end | −0.381 | <0.001 | <0.001 | [−0.6, −0.16] |
:), show, crew, awesome, fashion, guys | −0.381 | <0.001 | <0.001 | [−0.6, −0.16] |
*Effect sizes of individual linguistic features (diff-of-diff b/w true and null events) for inpatient visits. Significance was measured using paired, two-tailed t-test with Benjamini-Hochberg p-correction.