Table 18 Multimodal datasets

From: Machine learning for Parkinson’s disease: a comprehensive review of datasets, algorithms, and challenges

Article

Repository/Source

#PD

#HC

#PDF

#PDM

#HCF

#HCM

Information

170

PPMI220

885

–

–

–

–

–

Longitudinal data (years 0–4), 981 features: motor, non-motor, and imaging

171

PPMI220

423

196

146

277

70

126

7-year data collection from PPMI

PDEP288

610

196

218

392

70

126

PDBP dataset used for validation

169

PPMI220

648

434

256

392

232

202

HC: average age 62.38

PD: average of 64.38

172

PhysioNet250,251

93

72

32

40

35

58

HC: average age (63.65 ± 8.58)

PD: average age (66.30 ± 9.45)

15

PPMI220

73

59

21

52

16

43

Only individuals with complete MRI, SPECT, and CSF data were included

173

290

93

73

–

–

–

–

an average age of 66.3 years

Sarkar dataset192

20

20

6

14

10

10

Audio features: Jitter, Shimmer, HNR, Auto-Correlation

HandPD291

74

18

15

59

12

6

From São Paulo State University, Brazil

174

PPMI220

396

168

136

260

59

109

58 individuals with SWEDD (35 M, 23 F) with an average age of (60.6 ± 10) years,

Participants lacking clinical and imaging features

HC: average age (61.1 ± 11.3)

PD: average age (61.7 ± 9.65)

175

Sarkar dataset192

20

20

6

14

10

10

26 features, including parameters related to time and frequency

292

28

28

–

–

–

–

From Dandenong Neurology, Melbourne, VIC, Australia

HandPD291

74

18

15

59

12

6

From São Paulo State University, Brazil

NewHandPD235

31

35

–

–

–

–

A total of 66 individuals

293

–

–

–

–

–

–

–

294

–

–

–

–

–

 

1108 images,

659 Covid patients,

277 bacterial pneumonia cases,

175 healthy lungs

176

PPMI220

264

–

155

109

–

–

264 patients with known LRRK2

129

–

79

50

–

–

–

295,296

–

–

–

–

–

–

–

177

PPMI220

421

213

–

–

–

–

SWEED Group: 81 individuals,

465 M and 250 F

178

PPMI220

460

160

178

282

53

107

Data was obtained on January 6,

2023, from the PPMI database,

A total of 675 entities,

SWEED Group: 55 individuals

(23 F, 32 M)

179

Clinical

50

25

–

–

–

–

14 neurocognitive tests, health questionnaires, and movement evaluations

180

PPMI220

427

171

–

–

–

–

The model was trained with PPMI data and validated with PDBP data

PDEP288

804

442

–

–

–

–

 
  1. In this table, we detailed all the datasets in papers and compared participant demographics (number, gender, and health status: #PD => Parkinson’s Disease, #HC => Number of Healthy Control Participants, #PDF => Number of Parkinson’s disease Female Participants, #PDM => Number of Parkinson’s disease Male Participants, #HCF => Number of Healthy Control Female Participants, #HCM => Number of Healthy Control Male Participants).
Back to article page