Prediction of dysphagia aspiration through machine learning-based analysis of patients’ postprandial voices

Table 1 Demographic characteristics

	Normal	Aspiration	p-value
Gender (N (%))
Men	41 (32.03%)	52 (74.29%)	< 0.001* (χ²: 30.76, df: 1)
Women	87 (67.97%)	18 (25.71%)	< 0.001* (χ²: 30.76, df: 1)
Age (mean ± SD)
Total	61.16 ± 13.00	72.30 ± 12.03	< 0.001**
Men	63.27 ± 13.57	72.25 ± 11.68	0.001**
Women	60.16 ± 12.66	72.44 ± 13.34	0.001**
Comorbid conditions (N (%))
Total
Central nervous system disorders	17 (13.28%)	18 (25.71%)	< 0.001*** (χ²: 36.10, df: 5)
Digestive system and dental disorders	3 (2.34%)	12 (17.14%)
Pulmonary disorders	4 (3.12%)	9 (12.86%)
Other cancers	7 (5.47%)	3 (4.29%)
Aging-related disorders	12 (9.38%)	8 (11.43%)
None	85 (66.41%)	20 (28.57%)
Men
Central nervous system disorders	5 (12.20%)	11 (21.15%)	0.002*** (χ²: 18.54, df: 5)
Digestive system and dental disorders	1 (2.44%)	12 (23.08%)
Pulmonary disorders	2 (4.88%)	8 (15.38%)
Other cancers	2 (4.88%)	2 (3.85%)
Aging-related disorders	5 (12.20%)	6 (11.54%)
None	26 (63.41%)	13 (25.00%)
Women
Central nervous system disorders	12 (13.79%)	7 (38.89%)	0.140*** (χ²: 8.31, df: 5)
Digestive system and dental disorders	2 (2.30%)	0 (0.00%)
Pulmonary disorders	2 (2.30%)	1 (5.56%)
Other cancers	5 (5.75%)	1 (5.56%)
Aging-related disorders	7 (8.05%)	2 (11.11%)
None	59 (67.82%)	7 (38.89%)

^*The Chi-square test results show a significant difference. To address gender bias, separate models were constructed for each gender (male and female). The data was then divided into 10 folds for each gender. After that, the results were combined in the gender-neutral model, effectively removing any gender-related biases
^**The Mann–Whitney U test results indicate a significant difference between the two groups. However, to eliminate bias, participants under the age of 40 were excluded from the analysis
^***Regarding the comorbid conditions, a Chi-square test was conducted for analysis. While there are no significant differences observed among females, statistically significant differences are found in the overall dataset or males. However, vocal fold-related conditions were excluded, and dysphagia can occur in conjunction with various other conditions, which may account for differences when compared to the normal group

ISSN: 1743-0003