CentiAgent · Critical care
CentiAgent Sepsis Dashboard: an illustration
You are the pilot. CentiAgent is your co-pilot.
How a co-pilot dashboard can show a sepsis cohort at a glance: who is affected, how the bedside signs and laboratory values fall against the international thresholds, how the hour-1 bundle unfolds for one patient, and how outcomes are compared. Every threshold carries its source. Use the filters; every chart redraws.
Who has sepsis
Sepsis rate within each age band, by sex. The bars start at zero; hover a bar (or tap it) for the counts.
Sepsis rate by age band and sex
Share of patients meeting Sepsis-3 criteria (%)
Show the numbers
Definition: sepsis = suspected infection with an acute rise in total SOFA score of 2 points or more. Singer M et al. Sepsis-3. JAMA 2016;315(8):801–810.
Organ dysfunction: change in SOFA score
Patients by acute rise in SOFA score (points)
Threshold: SOFA rise ≥2 points with suspected infection (Singer et al. JAMA 2016). Grey bars at 2 or more are patients with organ dysfunction but no suspected infection: they do not meet Sepsis-3.
qSOFA at the bedside
The three qSOFA criteria at recognition. Each dashed line is the published cut-off; the stacked bars count patients, from zero.
Respiratory rate
Patients per 2 breaths/min band
qSOFA: respiratory rate ≥22/min (Singer et al. JAMA 2016).
Systolic blood pressure
Patients per 10 mmHg band (each band includes its upper value)
qSOFA: systolic blood pressure ≤100 mmHg (Singer et al. JAMA 2016).
Patients meeting each qSOFA criterion
Share of each group meeting the criterion (%)
qSOFA criteria: RR ≥22/min, SBP ≤100 mmHg, altered mentation (here GCS <15). Singer et al. JAMA 2016.
qSOFA score
Patients by score (0 to 3), computed from the generated RR, SBP and mentation
Use with care: a qSOFA of 2 or more flags a higher risk of a poor outcome (Singer et al. JAMA 2016), but SSC 2021 recommends against qSOFA as a single screening tool compared with SIRS, NEWS or MEWS (strong recommendation). Evans L et al. Crit Care Med 2021;49(11):e1063–e1143.
Septic shock: pressure and lactate
Septic shock in Sepsis-3 needs both: vasopressors to hold mean arterial pressure at 65 mmHg or more, and serum lactate above 2 mmol/L despite adequate volume.
Mean arterial pressure at recognition
Patients per 5 mmHg band, before any vasopressor
Thresholds: vasopressors to maintain MAP ≥65 mmHg (Sepsis-3 shock criterion); hour-1 bundle: vasopressors if hypotensive during or after fluids to keep MAP ≥65 (Levy MM et al. Crit Care Med 2018;46(6):997–1000); initial MAP target 65 mmHg (SSC 2021, strong).
Serum lactate
Patients per 0.5 mmol/L band (last band: 12 or more)
Thresholds: lactate >2 mmol/L (>18 mg/dL) with vasopressors = septic shock (Singer et al. JAMA 2016); re-measure if the first value is >2; give 30 mL/kg crystalloid for hypotension or lactate ≥4 mmol/L (Levy et al. 2018).
The hour-1 bundle, minute by minute
One fictional patient. Time zero is recognition of sepsis. The shaded zone is the first hour.
Patient S-0417 (fictional) · 68 years · 70 kg · suspected pneumonia · RR 26/min, SBP 92 mmHg, alert: qSOFA 2
Bundle elements (Levy MM, Evans LE, Rhodes A. The Surviving Sepsis Campaign Bundle: 2018 update. Crit Care Med 2018;46(6):997–1000): measure lactate and re-measure if >2 mmol/L; blood cultures before antibiotics; broad-spectrum antibiotics; rapid 30 mL/kg crystalloid for hypotension or lactate ≥4 mmol/L; vasopressors if hypotensive during or after fluids to keep MAP ≥65 mmHg. SSC 2021 restates the timing for antimicrobials: immediately, ideally within 1 hour, for possible septic shock or a high likelihood of sepsis; within 3 hours for possible sepsis without shock if concern persists. SCCM notes that not every element may be completed within the first hour, and the European Society for Emergency Medicine has published concerns about the one-hour target (Freund Y et al. Eur J Emerg Med 2019;26(4):232–233).
Biomarker panel
Box = middle half of patients (25th to 75th percentile), line = median, whiskers = 5th to 95th percentile. Linear axes from zero. Only lactate has a guideline threshold.
Lactate (mmol/L)
Guideline role: >2 mmol/L with vasopressors defines septic shock (Sepsis-3). SSC 2021 suggests measuring lactate in suspected sepsis (weak recommendation).
Limits: also raised by adrenergic drugs, liver failure, metformin and seizures; a normal value does not exclude sepsis.
C-reactive protein (mg/L)
Guideline role: none. CRP is not part of the Sepsis-3 definitions and SSC 2021 sets no CRP threshold.
Limits: a non-specific acute-phase protein, raised after surgery, trauma and in inflammatory disease; it rises and falls more slowly than procalcitonin.
Procalcitonin (ng/mL)
Guideline role: SSC 2021 suggests against using procalcitonin with clinical evaluation to decide when to start antimicrobials (weak), and suggests using it with clinical evaluation to decide when to stop them, where source control is adequate and the duration is unclear (weak).
Limits: raised after major surgery, trauma, burns and cardiogenic shock; cut-offs depend on the assay.
Protein C activity (%)
Guideline role: none. A natural anticoagulant consumed in sepsis-associated coagulopathy; low activity has been associated with worse outcomes in studies, but it is not in Sepsis-3 or SSC 2021.
Limits: reference ranges are laboratory-specific. Recombinant activated protein C was withdrawn in 2011 after PROWESS-SHOCK showed no survival benefit (Ranieri VM et al. N Engl J Med 2012;366:2055–2064).
Groups: ■ no sepsis · ■ sepsis without shock · ■ septic shock. Sources: Singer et al. JAMA 2016; Evans L et al. Intensive Care Med 2021;47(11):1181–1247.
Outcomes by bundle compliance
Patients with sepsis, split by whether every applicable hour-1 element was done within 60 minutes (cultures before antibiotics included). Whiskers are 95% confidence intervals (Wilson).
Simulated in-hospital mortality (%)
Read this as a concept, not an effect estimate. The difference shown is the assumption the generator was given, nothing more. Real comparisons of bundle compliance are observational and confounded by severity, setting and timing; this page cannot and does not estimate the benefit of the bundle.
What the co-pilot adds
In practice, CentiAgent would read the record and the monitors, compute qSOFA, SOFA change and lactate trends, show each against its sourced threshold, time the bundle elements and flag a missing culture or a late antibiotic. The clinician examines the patient, decides and signs.
This page shows the display layer only, on fictional data. A clinical deployment would require validation on local data, clinical governance and the applicable regulatory pathway for software that informs care.
See CentiAgent across 49 specialities · CentiAgent in Critical Care (e-book)
Thresholds and their sources
| Element | Threshold | Source |
|---|---|---|
| Sepsis: organ dysfunction | SOFA rise ≥2 points | Singer M et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA 2016;315(8):801–810 |
| qSOFA: respiratory rate | ≥22/min | Singer et al. JAMA 2016 |
| qSOFA: systolic blood pressure | ≤100 mmHg | Singer et al. JAMA 2016 |
| qSOFA: mentation | Altered (GCS <15) | Singer et al. JAMA 2016 |
| qSOFA score | ≥2 of 3 | Singer et al. JAMA 2016; not as a single screening tool: Evans L et al. Crit Care Med 2021;49(11):e1063–e1143 |
| Septic shock | Vasopressors for MAP ≥65 mmHg and lactate >2 mmol/L | Singer et al. JAMA 2016 |
| Hour-1: lactate | Measure; re-measure if >2 mmol/L | Levy MM, Evans LE, Rhodes A. Crit Care Med 2018;46(6):997–1000 |
| Hour-1: blood cultures | Before antibiotics | Levy et al. 2018 |
| Hour-1: antibiotics | Broad-spectrum; ideally within 1 h for possible shock or high likelihood of sepsis | Levy et al. 2018; Evans et al. 2021 |
| Hour-1: fluids | 30 mL/kg crystalloid for hypotension or lactate ≥4 mmol/L | Levy et al. 2018; within the first 3 h: Evans et al. 2021 (weak) |
| Hour-1: vasopressors | If MAP <65 mmHg during or after fluids | Levy et al. 2018; initial MAP target 65 mmHg: Evans et al. 2021 (strong) |
Guideline texts checked on 1 November 2026. SCCM has since published updated adult guidance; the thresholds above follow the 2016 definitions, the 2018 bundle and the 2021 guidelines named.
Context and method
cases of sepsis and 11 million sepsis-related deaths worldwide in 2017, about 20% of all global deaths. Real-world figure, from the WHO sepsis fact sheet, citing Rudd KE et al. Lancet 2020;395(10219):200–211. This is the only real-world statistic on the page.
How the cohort is built
1,000 fictional adult ICU patients drawn with a seeded generator (seed 20261101), so every visitor sees the same cohort. Sepsis is derived, not assigned: suspected infection plus a SOFA rise of 2 or more. Shock is derived from vasopressor need and lactate above 2 mmol/L. qSOFA is computed from the generated respiratory rate, systolic pressure and mentation.
What it is not
Not real patient data, not a prevalence or mortality estimate for any hospital or country, not a validated model and not a medical device. The distributions are set to look plausible for teaching. Everything runs in your browser; nothing is sent anywhere.