Development of diagnostic algorithm using machine learning for distinguishing between active tuberculosis and latent tuberculosis infection

Table 1 Demographic and clinical characteristics of the recruited participants

Variables	Discovery cohort		P*	Validation cohort		P*	P^†
Variables	ATB (n = 468)	LTBI (n = 424)	P*	ATB (n = 125)	LTBI (n = 138)	P*	P^†
Age, years	52.38 ± 14.04	53.08 ± 14.47	0.573	51.70 ± 13.68	53.51 ± 13.59	0.209	0.785
Sex, male, %	289 (61.75%)	247 (58.25%)	0.287	81 (64.80%)	83 (60.14%)	0.436	0.508
Underlying condition or illness
Diabetes mellitus	92 (19.66%)	77 (18.16%)	0.569	27 (21.60%)	24 (17.39%)	0.389	0.872
Virus hepatitis or cirrhosis	52 (11.11%)	39 (9.20%)	0.346	16 (12.80%)	16 (11.59%)	0.765	0.364
Nephritis or renal failure	36 (7.69%)	22 (5.19%)	0.13	8 (6.40%)	11 (7.97%)	0.623	0.68
Solid tumor	30 (6.41%)	27 (6.37%)	0.979	7 (5.60%)	9 (6.52%)	0.755	0.858
Heart disease	29 (6.20%)	18 (4.25%)	0.193	6 (4.80%)	6 (4.35%)	0.861	0.648
Positive culture for Mtb	398 (85.04%)	N/A	N/A	112 (89.60%)	N/A	N/A	N/A
Positive GeneXpert MTB/RIF	381 (81.41%)	N/A	N/A	106 (84.80%)	N/A	N/A	N/A

ATB: active tuberculosis; LTBI: latent tuberculosis infection; Mtb: Mycobacterium tuberculosis; N/A: not applicable. *Comparisons were performed between ATB and LTBI groups using Mann–Whitney U test or Chi-square test. ^†Comparisons were performed between discovery cohort and validation cohort using Mann–Whitney U test or Chi-square test. Data were presented as means ± standard deviation or numbers (percentages)

ISSN: 1471-2334