Sepsis

This dataset comprises vital signs, lab tests, administered fluid boluses and vasopressors for 2,164 patients with sepsis in the intensive care unit (ICU).

You can access our synthetic sepsis dataset through PhysioNet.
(see https://www.physionet.org/content/synthetic-mimic-iii-health-gym/1.0.0/)

What you should know about this dataset

These synthetic data were generated based on the MIMIC-III Clinical Database. The cohort was defined according to [1]Gottesman O, Futoma J, Liu Y, Parbhoo S, Celi L, Brunskill E, Doshi-Velez F. Interpretable Off-Policy Evaluation in Reinforcement Learning by Highlighting Influential Transitions. Proceedings of the … Continue reading, and included adults with an initial ICU stay of at least 24 hours and seven or more mean arterial pressure (MAP) values of 65mmHg or less, indicating probable acute hypotension. All synthetic records have a length of 48 hours.

Variables

The following tables contain a list of variables in the synthetic data together with descriptive statistics.

Variable Name Data Type Unit Numeric Statistics
Median Q1 Q3
Age numeric year 65.40 58.29 72.95
Heart Rate (HR) numeric bpm 89.09 78.46 99.82
Systolic BP numeric mmHg 123.67 114.43 133.03
Mean BP numeric mmHg 81.02 75.18 86.91
Diastolic BP numeric mmHg 58.90 50.40 66.95
Respiratory Rate (RR) numeric bpm 21.46 18.69 24.28
Potassium (K) numeric meq/L 4.12 3.78 4.45
Sodium (Na) numeric meq/L 140.01 136.59 143.57
Chloride (Cl−) numeric meq/L 105.23 102.08 108.03
Calcium (Ca) numeric mg/dL 8.02 7.37 8.66
Ionised Ca numeric mg/dL 1.11 1.04 1.18
Carbon Dioxide (CO2) numeric meq/L 25.27 23.44 27.29
Albumin numeric g/dL 3.01 2.68 3.32
Hemoglobin (Hb) numeric g/dL 10.20 9.17 11.23
Potential of Hydrogen (pH) numeric - - 7.39 7.34 7.44
Arterial Base Excess (BE) numeric meq/L 0.16 −2.04 2.48
Bicarbonate (HCO3) numeric meq/L 24.38 22.63 26.13
FiO2 numeric fraction 0.45 0.38 0.55
Glucose numeric mg/dL 134.11 108.21 167.06
Blood Urea Nitrogen (BUN) numeric mg/dL 25.38 19.89 31.92
Creatinine numeric mg/dL 1.13 0.90 1.44
Magnesium (Mg) numeric mg/dL 2.04 1.83 2.29
Serum Glutamic Oxaloacetic Transaminase (SGOT) numeric u/L 50.78 31.53 88.97
Serum Glutamic Pyruvic Transaminase (SGPT) numeric u/L 39.99 26.20 65.66
Total Bilirubin (Total Bili) numeric mg/dL 1.19 0.66 2.32
White Blood Cell Count (WBC) numeric E9/L 10.60 7.99 13.92
Platelets Count (Platelets) numeric E9/L 184.44 141.97 239.41
PaO2 numeric mmHg 109.07 84.22 139.63
Partial Pressure of CO2 (PaCO2) numeric mmHg 39.32 34.92 44.97
Lactate numeric mmol/L 1.82 1.41 2.40
Total Volume of Intravenous Fluids (Input Total) numeric mL 4867.46 1887.84 11155.76
Intravenous Fluids of Each 4-Hour Period (Input 4H) numeric mL 58.66 13.83 229.01
Maximum Dose of Vasopressors in 4H (Max Vaso) numeric mcg/kg/min 0.0002 0.0 0.0017
Total Volume of Urine Output (Output Total) numeric mL 2505.54 585.47 6733.69
Urine Output in 4H (Output 4H) numeric mL 159.33 44.74 361.69
Variable Name Data Type Unit Numeric Statistics
Gender binary - Male: 73.41% True: 26.59%
Readmission of Patient (Readmission) binary - False: 60.20% True: 39.80%
Mechanical Ventilation (Mech) binary - False: 56.89% True: 43.11%
GCS categorical point 13 Classes 3 : 8.71% 4 : 0.38% 5 : 0.50% 6 : 6.30% 7 : 0.74% 8 : 2.27% 9 : 1.52% 10 : 9.31% 11 : 9.12% 12 : 6.31% 13 : 2.53% 14 : 15.45% 15 : 36.85%
Peripheral Oxygen Saturation (SpO2) categorical % 10 Classes (C) C1: [0.00, 93.83) : 13.38%; C2: [93.83, 95.14) : 8.12%; C3: [95.14, 96.00) : 4.48%; C4: [96.00, 96.70) : 10.64%; C5: [96.70, 97.33) : 12.61%; C6: [97.33, 98.00) : 11.36%; C7: [98.00, 98.60) : 11.52%; C8: [98.60, 99.22) : 11.84%; C9: [99.22, 99.86) : 8.39%; C10: [99.86, 100.0] : 7.66%;
Temperature (Temp) categorical Celcius 10 Classes (C) C1: [15.11, 35.95) : 7.83%; C2: [35.95, 36.28) : 6.55%; C3: [36.28, 36.50) : 12.87%; C4: [36.50, 36.69) : 16.56%; C5: [36.69, 36.88) : 4.21%; C6: [36.88, 37.06) : 8.21%; C7: [37.06, 37.28) : 7.10%; C8: [37.28, 37.56) : 9.37%; C9: [37.56, 37.93) : 10.96%; C10: [37.93, 40.52] : 16.33%;
Partial Thromboplastin Time (PTT) categorical s 10 Classes (C) C1: [17.80, 24.53) : 7.69%; C2: [24.53, 26.63) : 6.71%; C3: [26.63, 28.20) : 10.02%; C4: [28.20, 29.60) : 12.44%; C5: [29.60, 31.45) : 5.46%; C6: [31.45, 34.00) : 9.27%; C7: [34.00, 37.10) : 9.99%; C8: [37.10, 42.80) : 11.47%; C9: [42.80, 57.90) : 12.38%; C10: [57.90, 150.00] : 14.58%;
Prothrombin Time (PT) categorical s 10 Classes (C) C1: [9.90, 12.20) : 7.89%; C2: [12.20, 12.90) : 8.2%; C3: [12.90, 13.30) : 11.02%; C4: [13.30, 13.80) : 9.84%; C5: [13.80, 14.30) : 9.45%; C6: [14.30, 14.90) : 6.59%; C7: [14.90, 15.90) : 10.37%; C8: [15.90, 17.51) : 10.51%; C9: [17.51, 22.00) : 13.27%; C10: [22.00, 146.70] : 12.85%;
International Normalised Ratio (INR) categorical - 10 Classes (C) C1: [0.00, 1.00) : 0.19%; C2: [1.00, 1.10) : 8.88%; C3: [1.10, 1.20) : 23.35%; C4: [2.21, 17.60] : 0.09% C5: [1.20, 1.30) : 15.64%; C6: [1.30, 1.31) : 10.22%; C7: [1.31, 1.50) : 7.53%; C8: [1.50, 1.70) : 9.71%; C9: [1.70, 2.21) : 10.67%; C10: [2.21, 17.60] : 13.70%;

Comparison of marginal distributions

Comparison of marginal distributions between the real and synthetic data for continuous, categorical and binary variables.

Numeric variable comparisons

E.g. MAP, BP, Systolic BP, Urine, ALT, AST, PO2, Lactic Acid, Serum Creatinine

Categorical variable comparisons

E.g. Fluid Boluses, Vasopressors, FiO2, GCS

Binary variable comparisons

E.g. Urine (M), ALT/AST (M), FiO2 (M), CGS (M), PO2 (M), Lactic Acid (M), Serum Creatinine (M)

Comparison of correlations

The matrices below show:

1) Static correlations between pairs of variables for all patients at any time, in the real and synthetic data.

2) Dynamic correlations between variables in the real and synthetic data. Time series for each patient were linearly decomposed into trends (indicating a general upward or downward slope) and cycles (indicating periodic patterns). Correlations were then computed between trends and cycles for all patients.

Real Data

Synthetic

Real Data

Synthetic

Real Data

Synthetic

Setup

Health Gym data can be downloaded either directly from this website or using the Python API.

Install the health gym python package through pip by running the following in your terminal.

				
					pip install healthgym
				
			

To download and access the data from within python.

				
					import healthgym as hg

sepsis_data = hg.datasets.Sepsis(root: 'path/to/data/', download = True)

				
			

All datasets support the following parameters.

				
					root (string)
# Root directory of dataset where dataset exists or will be saved to if download is set to True. 

download (bool, optional) 
# If true, downloads the dataset from the internet and puts it in root directory. If dataset is already downloaded, it is not downloaded again.
				
			

Download the dataset

Direct download as CSV file or view GitHub repository

Be part of the community
here is how to reach out.

Join our community on GitHub or contact us

References[+]