Table 4. Performance of the LLM and Text String Search Compared With Criterion Standard (n = 400)^a.

	Precision			Recall			F1 Score			Cohen κ (95% CI)
	Wearing helmet	Not wearing helmet	Unknown	Wearing helmet	Not wearing helmet	Recall unknown	Wearing helmet	Not wearing helmet	Unknown
LLM	0.98	1.00	0.98	0.98	0.96	1.00	0.98	0.980	1.00	0.98 (0.96-1.00)
Text string search	0.98	1.00	0.98	0.98	0.96	1.00	0.98	0.98	1.00	0.98 (0.96-1.00)

Abbreviation: LLM, large language model.

^a These 400 records were a random draw from the high-detail prompt (December 7, 2023).