Table 2.

Analysis grid used to extract relevant information from reports. aTRL: adapted technology readiness level

Domain	Criterion
Publication type	Original article, communication, conference paper, book, technical note
Publication reliability	Peer-reviewed publication, date of the publication
Data sources	Real or generated data, subject types (humans or animals)
Population/sample study	Representativeness of the population/sample, inclusion, and exclusion criteria
	Size of the population/sample
Input data	Features (input data) used for the model
	Datasets used and distribution of data (balanced vs. unbalanced data)
	Processing of missing data
Outcome	Description of the outcome variable
Model development	Architecture of the model
	How overfitting is handled
Model performance	Metrics used to assess the performance of the model
Model evaluation	Value of the performance metrics
Real application	Is the model used in medicolegal practice? Does it perform better compared to other non-AI methods?
	Is the model applied on a population/sample that is included into the population/sample study?
Maturity of the application	Maturity of the application estimated from the aTRL scale