Page 123 - AIH-2-1

Artificial Intelligence in Health Does improving diagnostic accuracy increase AI adoption?

Table 1. The proportion of individuals choosing the AI and/or AI+ alternative over the saliva test per sensitivity and specificity
levels
AI versus saliva test AI+ versus saliva test AI or AI+ versus saliva test
n 1482 1743 3225
Proportion of yes to AI/AI+ 45.82% 28.92% 36.68%
+++ +++
Sensitivity
Proportion of yes to AI/AI+ when sensitivity=60% 36.95%*** 32.77%** 35.04%***
Proportion of yes to AI/AI+ when sensitivity=70% *** ** 31.63%***
44 33.% +++ 26 43.% +++
Proportion of yes to AI/AI+ when sensitivity=95% (***) (**) 48.68%***
59 39.% (+++ ) 30 04.% (+++ )
Specificity
Proportion of yes to AI/AI+ when specificity=60% 41.24% 29.68%***
+++ 18 2.% (***)
(+++ )
Proportion of yes to AI/AI+ when specificity=70% 47.67% 28.18%***
+++ 12 73.% (***)
(+++ )
Proportion of yes to AI/AI+ when specificity=95% 47.04% 44.24%***
+++ 41 94.% (***)
(++ )
Notes: (1) Chi-square statistical test of difference of acceptance rate per type of scenario offered (AI vs. saliva test or AI+ vs. saliva test): *P<0.1,
**P<0.05, and ***P<0.01; (2) Pairwise z-test of difference in proportions per sensitivity or specificity level (60%, 70%, or 95%) between AI-versus-
saliva-test and AI+-versus-saliva-test: P<0.1, P<0.05, +++ P<0.01.
++
+
Abbreviations: AI: Artificial intelligence using electronic health records (EHRs) data; AI+: Artificial intelligence using electronic health records (EHRs)
data and digital consumer data.

Proportion of yes to AI when
sensitivity/specificity = 95%

Proportion of yes to AI when
sensitivity/specificity = 70%

Proportion of yes to AI when
sensitivity/specificity = 60%

0.00% 10.00% 20.00% 30.00% 40.00% 50.00% 60.00%
SPECIFICITY SENSITIVITY
Figure 2. The proportion of artificial intelligence adoption (vs. saliva test) across different levels of accuracy

5. Discussion The results strongly showed the influence of AI
sensitivity (how well the AI test can identify true positives)
The findings of this study provide valuable insights into
public acceptance of AI-based diagnostic alternatives on acceptance rates. As AI sensitivity increased from 60%
compared to conventional saliva tests. By analyzing to 95%, support for AI/AI+ diagnostics rose considerably,
3225 pairwise choices, we observed that only 36.68% of reaching 48.68% at 95% sensitivity. A similar pattern
participants preferred AI/AI+ alternatives over traditional emerged with specificity (how well an AI test can identify
saliva tests, specifically with a significant preference for AI true negatives), where acceptance rates increased from
(45.82%) compared to AI+ (28.92%). 29.68% at 60% specificity to 44.24% at 95% specificity.

Volume 2 Issue 1 (2025) 117 doi: 10.36922/aih.3561