Skip to main content
. 2024 May 17;12:50117. doi: 10.2196/50117

Table 1. Comparison between imbalanced data versus imbalance mitigation strategies.

Studies Participant distribution Techniques Conclusion
Moon et al [29] 524 patients with PDa and 43 patients with essential tremor
  • SMOTEb

  • F1-score improved

Veeraragavan et al [30] 93 patients with idiopathic PD and 73 controls; 10 patients with H&Yc 3; 28 patients with H&Y 2.5; and 55 patients with H&Y 2
  • SMOTE

  • Test accuracy improved

Falchetti et al [31] 388 patients with idiopathic PD and 323 controls
  • Oversampling

  • Undersampling

  • Combination of oversampling and undersampling

  • Without any sampling, the combination of oversampling and undersampling methods is comparable

Jeancolas et al [32] 115 patients with PD and 152 controls
  • Data augmentation

  • Performed better for free speech task

  • No consistent improvement in the sentence repetition task

a

PD: Parkinson disease.

b

SMOTE: synthetic minority oversampling technique.

c

H&Y: Hoehn and Yahr.