Update README.md
Browse files#################################################################################################################################
### A 100,000-patient database that contains in total 100,000 patients, 361,760 admissions, and 107,535,387 lab observations. ###
### Created by: Uri Kartoun, PhD, Copyright 2014. ###
#################################################################################################################################
The Problem
It is difficult and expensive to access Electronic Medical Records (EMRs) due to privacy concerns and technical problems.
I am a student or a researcher working at a university that does not have yet an access to EMR system and I am interested in evaluating machine learning algorithms.
Tedious bureaucracy.
I am teaching a computer science course and I wish to let my 150 students to experiment with electronic medical records.
Not possible due to privacy issues.
I am in a process of founding a company focused on developing a new EMR management platform and I want to demonstrate to a venture capital company and to potential customers the ability of my product to handle big data.
Current simulated medical databases are limited and are hard to configure.
The Solution
A database of artificial patients.
The data is generated according to pre-defined criteria and is not based on any human data.
The database contains the same characteristics that exist in a real medical database such as patients admission details, demographics, socioeconomic details, labs, medications, etc.
The database is customizable. For example, it is possible to generate a population of 100,000 patients of which 60% are male, 40% are African American, 15% are diabetic, specific lab range distributions can be set, etc.
The number of records can range from several thousands to millions, depending on the desired configuration.
#################################
### PatientCorePopulatedTable ###
#################################
[PatientID] - a unique ID representing a patient.
[PatientGender] - Male/Female.
[PatientDateOfBirth] - Date Of Birth.
[PatientRace] - African American, Asian, White.
[PatientMaritalStatus] - Single, Married, Divorced, Separated, Widowed.
[PatientLanguage] - English, Icelandic, Spanish.
[PatientPopulationPercentageBelowPoverty] - given in %.
####################################
### AdmissionsCorePopulatedTable ###
####################################
[PatientID] - a unique ID representing a patient.
[AdmissionID] - an admission ID for the patient.
[AdmissionStartDate] - start date.
[AdmissionEndDate] - end date.
#############################################
### AdmissionsDiagnosesCorePopulatedTable ###
#############################################
[PatientID] - a unique ID representing a patient.
[AdmissionID] - an admission ID for the patient.
[PrimaryDiagnosisCode] - ICD10 code for admission's primary diagnosis.
[PrimaryDiagnosisDescription] - admission's primary diagnosis description.
##############################
### LabsCorePopulatedTable ###
##############################
[PatientID] - a unique ID representing a patient.
[AdmissionID] - an admission ID for the patient.
[LabName] - lab's name, including:
CBC: WHITE BLOOD CELL COUNT
CBC: RED BLOOD CELL COUNT
CBC: HEMOGLOBIN
CBC: HEMATOCRIT
CBC: MEAN CORPUSCULAR VOLUME
CBC: MCH
CBC: MCHC
CBC: RDW
CBC: PLATELET COUNT
CBC: ABSOLUTE NEUTROPHILS
CBC: ABSOLUTE LYMPHOCYTES
CBC: NEUTROPHILS
CBC: LYMPHOCYTES
CBC: MONOCYTES
CBC: EOSINOPHILS
CBC: BASOPHILS
METABOLIC: SODIUM
METABOLIC: POTASSIUM
METABOLIC: CHLORIDE
METABOLIC: CARBON DIOXIDE
METABOLIC: ANION GAP
METABOLIC: GLUCOSE
METABOLIC: BUN
METABOLIC: CREATININE
METABOLIC: TOTAL PROTEIN
METABOLIC: ALBUMIN
METABOLIC: CALCIUM
METABOLIC: BILI TOTAL
METABOLIC: AST/SGOT
METABOLIC: ALT/SGPT
METABOLIC: ALK PHOS
URINALYSIS: SPECIFIC GRAVITY
URINALYSIS: PH
URINALYSIS: RED BLOOD CELLS
URINALYSIS: WHITE BLOOD CELLS
[LabValue] - lab's value
[LabUnits] - lab's units.
[LabDateTime] - date.