Application of an Interpretable Machine Learning Model to Predict Lymph Node Metastasis in Patients with Laryngeal Carcinoma

Table 2

A univariate logistic regression analysis of variables related to LNM.

Variables	OR	95% CI	value
Age at diagnosis
≥65	Reference
50–65	1.495	1.070–2.089	0.019
50	1.605	1.414–1.821	<0.001
Race
White	Reference
Black	1.299	1.102–1.530	0.002
Other	1.349	1.016–1.791	0.038
Sex
Male	Reference
Female	1.172	1.003–1.369	0.046
Primary site
Supraglottis	Reference
Glottis	0.236	0.204–0.273	<0.001
Subglottis	0.351	0.239–0.513	<0.001
Larynx	0.796	0.637–0.995	0.045
Other	0.80	0.625–1.024	0.076
Histology
Squamous cell carcinoma	Reference
Nonsquamous cell carcinoma	0.903	0.66–1.234	0.522
Number of tumours
1	Reference
>1	0.701	0.616–0.799	<0.001
Tumour size
≤3 cm	Reference
>3 cm	3.617	3.184–4.109	<0.001
Grade
Grade I	Reference
Grade II	3.478	2.647–4.568	<0.001
Grade III and IV	7.405	5.594–9.802	<0.001
T-stage
T1	Reference
T2	4.607	3.571–5.945	<0.001
T3	7.187	5.656–9.132	<0.001
T4	9.077	7.248–11.367	<0.001

Notes: represents a statistically significant difference. Abbreviations: LNM, lymph node metastasis.