Table 1 Descriptive characteristics of included articles (N = 52)
n | % | n | % | ||
|---|---|---|---|---|---|
Site (patient data source) | AI type | ||||
China | 25 | 48.1% | Deep learning | 48 | 92.3% |
Korea | 7 | 13.5% | Machine learning | 4 | 7.7% |
USA | 5 | 9.6% | AI explainability | ||
Japan | 3 | 5.8% | AI | 49 | 94.2% |
Open-source dataset | 3 | 5.8% | XAI | 2 | 3.8% |
Other | 9 | 17.3% | Both | 1 | 1.9% |
Study cohort (kappa = 0.52) | Task (kappa = 0.53) | ||||
Retrospective | 42 | 80.8% | Image | 43 | 82.7% |
Prospective | 9 | 17.3% | Video | 3 | 5.8% |
Unclear | 1 | 1.9% | Text | 2 | 3.8% |
Clinician number in HMT | Mixed | 4 | 7.7% | ||
≤ 10 | 30 | 57.7% | Teaming mode (kappa = 0.73) | ||
11–20 | 15 | 28.8% | Simultaneous | 31 | 59.6% |
21–30 | 1 | 1.9% | Sequential | 17 | 32.7% |
31–60 | 4 | 7.7% | Both | 2 | 3.8% |
> 100 | 2 | 3.8% | Unclear | 2 | 3.8% |
Clinician expertise (kappa = 0.81) | Number of patients/tasks | ||||
Senior | 3 | 5.8% | ≤100 | 9 | 17.3% |
Junior | 7 | 13.5% | 101–200 | 12 | 23.1% |
Unclear | 15 | 28.8% | 201–300 | 11 | 21.2% |
Mixed | 27 | 51.9% | 301–400 | 7 | 13.5% |
401–500 | 3 | 5.8% | |||
501–1000 | 5 | 9.6% | |||
>1000 | 5 | 9.6% | |||
