Table 4. Bivariate comparisons of characteristics by willingness to adopt large language models among patients and caregivers. Statistical comparisons were performed using t tests for continuous variables and Chi-square tests for categorical variables.
| Variables | Participants | P valuea | |
|---|---|---|---|
| Willing (n=187) | Unwilling/uncertain (n=293) | ||
| Sex, n (%) | .547 | ||
| Male | 98 (52.4) | 144 (49.1) | |
| Female | 89 (47.6) | 149 (50.9) | |
| Role, n (%) | .787 | ||
| Patient | 142 (75.9) | 218 (74.4) | |
| Caregiver | 45 (24.1) | 75 (25.6) | |
| Education level, n (%) | <.001 | ||
| Primary school or below | 19 (10.2) | 86 (29.4) | |
| Junior high school | 28 (15) | 66 (22.5) | |
| High school | 40 (21.4) | 65 (22.2) | |
| College | 58 (31) | 45 (15.4) | |
| Bachelor’s degree or above | 42 (22.5) | 31 (10.6) | |
| Hospital affiliation, n (%) | .886 | ||
| Nonaffiliated | 78 (41.7) | 119 (40.6) | |
| Affiliated | 109 (58.3) | 174 (59.4) | |
| Hospital type, n (%) | .462 | ||
| Oncology | 33 (17.6) | 61 (20.8) | |
| General | 154 (82.4) | 232 (79.2) | |
| Province economic level, n (%) | <.001 | ||
| Low | 23 (12.3) | 67 (22.9) | |
| Medium | 97 (51.9) | 61 (20.8) | |
| High | 67 (35.8) | 165 (56.3) | |
| Frequency of digital tool use, n (%) | <.001 | ||
| Rarely | 4 (2.1) | 90 (30.7) | |
| Occasionally | 24 (12.8) | 138 (47.1) | |
| Often | 159 (85) | 65 (22.2) | |
| Previous awareness of LLMsb, n (%) | .177 | ||
| Yes | 102 (54.5) | 140 (47.8) | |
| No | 85 (45.5) | 153 (52.2) | |
| Previous use of LLMs, n (%) | .118 | ||
| Yes | 91 (48.7) | 120 (41) | |
| No | 96 (51.3) | 173 (59) | |
| Age (years), mean (SD) | 49.5 (12.0) | 50.5 (13.0) | .410 |
| Trust in LLMs (1-5), mean (SD) | 3.02 (1.18) | 2.36 (1.04) | <.001 |
| Perceived usefulness (1-5), mean (SD) | 3.97 (0.70) | 2.10 (0.88) | <.001 |
| Privacy concerns (1-5), mean (SD) | 2.21 (1.77) | 3.04 (1.25) | <.001 |
a
P<.05 considered statistically significant.
b
LLM: large language model.