Home

The M@tric project is a joint and ambitious “Big Data” project of the 3 ICUs of the university hospitals of Antwerpen (UZA), Ghent (UZ Gent) and Leuven (UZ Gasthuisberg Leuven) with the aim of performing advanced clinical research in critically ill patients.
The core of the M@tric is a database containing high-quality and complexly interrelated data from all adult patients annually admitted since 2013 to the intensive care units (ICUs) of the participating university hospitals.

The project was funded in 2009 by the Flemish government via the Hercules program of the Research Foundation – Flanders (FWO).

 

Mission

Intensive care medicine deals with the treatment of patients with acute life-threatening illness or severe trauma. It is now a separate, mature and high-technological medical discipline in which intensive monitoring and support is provided to different organ systems such as brain, heart, lungs, liver and kidneys, through a combination of pharmacological therapies and sophisticated medical devices.

Intensive Care medicine has an essential place in every health care system, as it is treating the sickest patients of all the medical disciplines. On average 10% of all beds in Belgian hospitals are intensive care beds. Intensive care medicine is extremely expensive (around 0.5% of GDP) and consumes a large fraction of the health care resources.

The deployment of “Information and communication technology (ICT)” in health care has become essential to improve the quality of care and efficiency. Intensive Care medicine is especially suitable for computerization because it is an extremely data-rich environment. The amount of data generated in the ICU is so overwhelming that it often leads to data-overload which can even result in information loss. More than 236 different variable categories are reported in one single medical ICU record and this far exceeds human cognitive capacities.

“Patient data management systems (PDMSs)” with focus on the ICU were increasingly implemented since the beginning of this century. Such a PDMS requires that each ICU patient has a dedicated bed-side PC client. Interfaces are provided with all monitoring devices, ventilators and syringe pumps with automated data collection occurring every few minutes. Additional data are inputted manually if necessary. These systems also include “Computerized Physician Order Entry (CPOE)” and interfaces with the hospital information system, including the lab system, resulting in a database in which each patient generates between 10,000 and 20,000 data values on a daily basis.

In 2007 the 3 ICUs of the university hospitals of Antwerpen (UZA), Ghent (UZ Gent) and Leuven (UZ Gasthuisberg Leuven) started a joined and ambitious project under the name M@tric. The aim of this “Big Data” project was to create an interuniversity database containing high-quality and complexly interrelated data from all 9,000 adult patients annually admitted to the ICU of the participating university hospitals.

The project was funded in 2009 by the Flemish government via the Hercules program of the Research – Flanders (FWO).

A multidisciplinary team from the 3 participating ICUs was set up to integrate all clinically relevant data from the 3 PDMSs of the 3 participating ICUs with focus on data quality, data semantics and standardization, data privacy and security. The project was supported by an external company. The M@tric database contains more than 800 variables per patient with a high granularity (e.g. physiologic variables every minute).

The project was officially launched in 2017 and is now fully operational and contains already more than 25,000 patients (January 2018).

The M@tric allows high-level clinical research and the research of the M@tric consortium focusses on the following domains:

  • epidemiological and clinical research
  • knowledge discovery, data mining, machine learning and predictive modeling
  • longitudinal data analysis and causality
  • translational research
  • health economic research

The M@tric consortium is open to collaborate with other research groups from Belgium and abroad, including data sharing projects.

Organisation

The Unitversity Hospitals of:

Antwerp

  • Prof. dr. Philippe Jorens, intensivist
  • Dr. Walter Verbrugghe, intensivist, data scientist
  • Dirk Reyntiens, data scientist

Ghent

  • Prof. dr. Johan Decruyenaere, intensivist
  • Prof. dr. Kirsten Colpaert, intensivist
  • Christian Danneels, data scientist
  • MSc. Bram Gadeyne, data scientist

Leuven

  • Prof. dr. Greet Van den Berghe, intensivist
  • Prof. dr. Geert Meyfroidt, intensivist
  • Dr. Philippe Meersseman, intensivist
  • Dominiek Cottem, data scientist

General Figures

These panels show some more detailed information regarding the contents of the M@tric project (last updated January 2018).

  • Nr of patients: 22249
  • Nr of hospital admissions: 24176
  • Nr of ICU admissions: 27019
Gender

Age on admission

Length

Weight

Regions

This map shows a distribution of the residence of the patients

Origin

Where do patients come from before hospital admission.

ICU Survival

Hospital Survival

Medical Admission Diagnosis

Surgical Admission Diagnosis

Structure

Anonimisation

Anonymity is a big issue when dealing with sensitive data. A few manipulations were made to the data to ensure anonymity.

  • The data does not contain any names or free text fields that could mention other sensitive data.
  • The birthdates are changed to the first day of the month. These will only be given if the research really requires it. Otherwise the age or a categorical age will be given.

Both the project and the database were registered with the privacy committee and approved by the ethics committee. The project will be screened for GPDR compliance in the near future.

Patients, hospital and ICU-admissions

A patient is uniquely defined (within a center) and is assigned a GUID.

One patient can have multiple hospital admissions. This is defined as a period where the patient at some point in time entered the hospital. A patient might go to different departments and might be (re) admitted to different ICU’s within one hospital admission. Each hospital admission is assigned a sequence number within the patient.

An ICU admission is one uninterrupted stay at the ICU and is assigned a sequence number within the hospital admission.

The different patient information tables

Figure: The different patient information tables

Next to the identifiers, each level has several attributes like mortality, admission/dischargetime, admission/discharge wards, etc.

You can download the full database schema here

Data tables

Table Description Nr of values
ComorbidityLists comorbidities applicable to the admission51783
AllLabDataLaboratory values17465370
AllMonDataHigh frequency values that are comming from monitoring devices.1382396432
DiagnosisProceduresDiagnoses and procedures with a start- and endtime indication (if applicable)148320
MedTreatmentAll doses of medication that were given during a specific time period and at the same rate.4278206
ScoresCalculated scores like SOFA and TISS28.2989974

(last updated January 2018)

Free Joomla! template by L.THEME