Table 1 Summary of patient characteristics in the Royal National Ear, Nose and Throat dataset (N = 54927) in comparison with the Massachusetts Eye and Ear dataset (N = 116400). The demographics are presented for the entire database N = 116400 patients, however of these, only 66252 patients had audiograms used in the GMM model. The demographics of this subcohort are not presented in the published study so we present the demographics of the entire dataset only.

From: Identification of sensorineural hearing loss subtypes using unsupervised machine learning and assessment of their replicability

 

RNENT

(N = 54927)

MEE

(N = 116400)

Age (years)

Mean (SD)

Range

Proportion ≥ 50 years (%)

61.9 (± 19.7)

18–100

71

-

18–80

63

Gender (%)

Male

Female

44

56

46–49 as function of age group

51–54 as function of age group

GMM a model parameters

Cluster number

Covariance Type

Regularisation Value

Convergence Thresholds

9

Full

0.01

le-3

10

Full

0.01

le-6

  1. a GMM, Gaussian Mixture Model.