top of page

Analysis of SIRS in Sepsis Data Set Using Tableau


As we all know, sepsis is a life-threatening disease caused by either bacterial or viral infection, potentially leading to malfunctioning of various organs, shock, and possibly death.

Levels of Sepsis:

There are three stages of sepsis:

1. Sepsis

2. Severe Sepsis

3. Septic shock

Figure 1. Stages of Sepsis

Before Sepsis, there is another stage called SIRS (Systematic Inflammatory Response Syndrome). When Sepsis is not documented with SIRS then it is due, to utilized infection, then you must ask for clarification from the physician. Sepsis may be caused by a severe bacterial infection (sepsis), trauma, or pancreatitis. It is marked by fast heart rate, low blood pressure, low or high body temperature, and low or high white blood cell count. The condition may lead to multiple organ failure and shock.

How SIRS is related to Sepsis?

Sepsis is the systemic response to infection and is defined as the presence of SIRS in addition to a documented or presumed infection. As SIRS is the initial infection stage to enter Sepsis, it is one of the most difficult aspects of sepsis care in diagnosing it on time and providing proper treatment. The health system developed a program where the nursing staff completes a manual assessment of their patients' vitals regularly, intending to identify SIRS faster.

Identifying SIRS in a patient include the following criteria:

  1. Temperature >38.5ºC or <35ºC (body temperature alterations which are known as hyperthermia or hypothermia)

  2. Heart rate >90 beats/min (Tachycardia)

  3. Respiratory rate >20 breaths/min or PaCO2 <32 mmHg (Tachypnoea)

  4. WBC >12,000 cells/mm3 or < 4000 cells/mm3, or >10 percent immature (band) forms (leukocytosis or leukopenia)

Finding two or more of any of these criteria in a patient indicates the possibility of sepsis.

Insights of Sepsis Data set:

The Sepsis dataset has 43 features or fields which can broadly be classified into

• Demographics

• Vital Signs or Clinical

• Laboratory values

In the Sepsis dataset, there is a feature called Sepsis Label which indicates whether a patient has sepsis on admission or not. The sepsis label holds either a 0 or 1 value. When a patient is admitted to the hospital, one of the following happens:

• Sepsis label is 0 (No sepsis on admission)

• Sepsis label is converted from 0 to 1(No sepsis on admission and then converted to sepsis)

• Sepsis label 1 (Sepsis on Admission)

Analysis of SIRS using Tableau:

As all of us experience COVID-19 for the past two years, we know that when our near and dear ones are infected with COVID-19 it gets really difficult for us to get a timely diagnosis. Proper treatment which might include getting an ICU bed in a hospital is extremely challenging. At the same time, it is also difficult for the hospital administration and the doctors to decide on the allocation of ICU beds as it depends upon the severity of the patients.

The same thing happens with SIRS-infected patients. Here we will create four visualizations in Tableau public. The first Tableau visualization will segregate the patients with sepsis labels 0 and 1, which can be represented using a pie chart.

Figure 2. Segregation of Sepsis Patients

The segregation of sepsis patients is illustrated in Figure.2. In the above Tableau visualization, 98.02% of patients have no sepsis, and 1.98% of the patients have sepsis. The goal of this visualization is to provide timely diagnosis and proper treatment to this 98.02% of patients with sepsis label 0 so that they can be cured before entering the sepsis stage. As 98.02% of patients get admitted to the hospital that means they got some kind of infection or SIRS. The second Tableau visualization is for the use of hospital administration, doctors, and nurses. This visualization can be used to decide on whom to allocate the next available ICU bed.

Figure 3. Visualization of SIRS patients

The Tableau Visualization of SIRS patients or the possibility of getting sepsis is explained in Figure.3. One calculated field is created, which filters the patients that satisfy two or more SIRS criteria. This visualization includes the patient ID, temperature as Temp, heart rate as HR, respiration as Resp, Paco2, white blood cell count as WBC, and sepsis label. Here colors used are against each condition. Then it will be easier for the doctors to identify which of the two SIRS criteria are satisfied by the patient among the four SIRS criteria. Now let's discuss more the calculated field. Sepsis possibility is the name of the calculated field, which contains the logic of the nested if statement.

The third visualization is only for the use of doctors to find out the percentage of the patients satisfying any of the four conditions.

Figure 4. Percentage of Patients Satisfying SIRS conditions

In Figure.4, a tree map is used to find out the percentage of patients who satisfy a common combination of the SIRS criteria. In that way, it will be easier for the doctors to provide custom treatment to that set of patients.

The fourth visualization is for use of doctors to hourly monitor the SIRS parameters of individual patients using their patient ID.

Figure 5. Hourly monitor of SIRS parameters with a Patient ID

In Figure.5, a combination of bar and line graphs is used to hourly monitor the temperature, heart rate, WBC, respiration rate, and Paco2 of SIRS patients using their patient id. Using this visualization doctors can take immediate steps to provide better treatment so that SIRS patients can be saved from entering into the sepsis stage.

Conclusion: SIRS requires immediate treatment in ICU for a month or so. Recovery is achievable but it takes a longer time. It is marked that patients with SIRS have recovered normal health without residual dysfunction. As discussed above, the hospital administration, doctors, and nurses can use the above visualizations to provide timely diagnosis and correct treatment to SIRS patients which can save them from a life-threatening disease.


1. – “How to Improve Sepsis Diagnosis and Outcomes through Innovative Healthcare Analytics



4. htm


1,167 views0 comments

Recent Posts

See All

In the early stages of a visualization project, we often start with two interrelated questions: Where can I find reliable data? What does this data truly represent? Information does not magically appe

bottom of page