Table 1 The model includes variables assessing cognitive function (ADAS and MMSE), as well as laboratory, clinical, and background variables.

From: Machine learning for comprehensive forecasting of Alzheimer’s Disease progression

Name Category Type Temporal Statistics Missing%
Commands ADAS Ordinal Yes 0.58 (0.84) 0.1
Comprehension 0.40 (0.72) 0.1
Construction 0.94 (0.90) 0.1
Delayed Word Recall 7.94 (2.51) 0.4
Ideational 0.54 (0.88) 0.1
Instructions 0.78 (1.18) 0.1
Naming 0.67 (0.88) 0.1
Orientation 2.48 (2.02) 0.1
Spoken Language 0.30 (0.69) 0.1
Word Finding 0.66 (0.90) 0.1
Word Recall 6.04 (1.78) 0.1
Word Recognition 6.35 (3.30) 0.1
Attention and Calculation MMSE Ordinal Yes 2.88 (1.69) 16.8
Language 7.90 (0.92) 16.8
Orientation 6.56 (1.92) 16.8
Recall 0.82 (0.88) 16.8
Registration 2.90 (0.34) 16.8
Alanine aminotransferase Laboratory Continuous Yes 0.32 (0.14) μkat/l 18.2
Alkaline phosphatase 1.29 (0.46) μkat/l 18.2
Aspartate aminotransferase 0.37 (0.10) μkat/l 18.2
Cholesterol 5.5 (1.0) mmol/l 17.9
Creatine kinase 0.99 (0.62) mg/dl 0.7
Creatinine 0.95 (0.22) mg/dl 17.9
Gamma glutamyl transferase 2.3 (1.8) iu/dl 32.0
Hematocrit 0.42 (0.04) counts 14.7
Hemoglobin 14.0 (1.2) g/dl 0.8
Hemoglobin a1c 5.81 (0.73)% 48.4
Indirect bilirubin 0.51 (0.24) mg/dl 48.4
Potassium 4.34 (0.35) mmol/l 18.0
Sodium 1.41 (0.02) mmol/cl 31.8
Triglycerides 1.53 (0.83) g/l 18.1
Blood pressure (diastolic) Clinical Continuous Yes 75.9 (8.3) mmHg 1.8
Blood pressure (systolic) Continuous 135 (15) mmHg 1.8
Heart rate Continuous 67.3 (8.2) bpm 1.8
Weight Continuous 71 (15) kg 3.0
Dropout Binary none 0.1
Age at baseline Background Continuous No 73.4 (8.4) years 0.9
Geographic region Categorical 67% North America 0
Initial diagnosis (AD or MCI) Binary 69% AD/31% MCI 0
Past cardiovascular event Binary 37% Y/63% N 0
ApoE ε4 allele count Ordinal 36% 0/48% 1/16% 2 72.4
Race Categorical 93% White 0.2
Sex Binary 54% F/46% M 0
Height Continuous 165 (10) cm 1.9
  1. The statistics column gives the mean and standard deviation of the data (combining training, validation, and test data) at baseline, along with any units. For geographic region and race, the dominant category frequency is given. The missing percentage column gives the percentage of missing data for each variable at baseline.