LONE RXANALYTICS

Antibiogram Machine Data interpretation lab

02 / ANTIBIOGRAM MACHINE

A percentage needs a denominator.

Explore invented susceptibility data. Change the cohort, inspect uncertainty, and see why missing tests and small samples must stay visible.

SYNTHETIC · NOT FOR CARE
Try a lesson
Floor / Urine / E. coli · Period BEach agent has its own denominator. Do not add agent test counts to infer unique patients.

Ceftriaxone

n = 120 tested
85% S95% interval 78–90%
Inspect counts and calculation

102 susceptible (S) + 5 intermediate (I) + 13 resistant (R) = 120 tested.

S% = (102 ÷ 120) × 100. Intermediate is not counted as susceptible here.

Cefepime

n = 115 tested
93% S95% interval 87–96%
Inspect counts and calculation

107 susceptible (S) + 3 intermediate (I) + 5 resistant (R) = 115 tested.

S% = (107 ÷ 115) × 100. Intermediate is not counted as susceptible here.

Ciprofloxacin

n = 120 tested
70% S95% interval 61–77%
Inspect counts and calculation

84 susceptible (S) + 8 intermediate (I) + 28 resistant (R) = 120 tested.

S% = (84 ÷ 120) × 100. Intermediate is not counted as susceptible here.

Piperacillin/tazobactam

n = 110 tested
92% S95% interval 85–96%
Inspect counts and calculation

101 susceptible (S) + 4 intermediate (I) + 5 resistant (R) = 110 tested.

S% = (101 ÷ 110) × 100. Intermediate is not counted as susceptible here.

● Period B estimate━ 95% Wilson intervalScale: 0–100% susceptible

THE LESSON

90% can mean different evidence.

18 of 20 and 90 of 100 are both 90%. Their approximate 95% Wilson intervals are 70–97% and 83–94%. More observations narrow sampling uncertainty; they do not remove sampling bias.

THE BOUNDARY

A cohort is not a patient.

Unit, specimen, collection methods, testing selection, and time period all affect interpretation. These invented counts do not represent any hospital, region, patient, or clinical recommendation.

Download synthetic cohort CSV
Data lineage, reporting rules, and references

All rows are deterministic teaching fixtures created for this portfolio. Period A is a separate synthetic comparison, not a real historical trend. S + I + R equals each agent’s tested count; untested isolates are not added to the denominator. The display suppresses n < 30 unless deliberately revealed. Wilson intervals describe binomial sampling uncertainty, not clinical appropriateness. Rounded percentage-point changes are descriptive, not a significance test.

Real antibiograms require laboratory-verified results, appropriate isolate de-duplication, current breakpoints, and consistent reporting. See CLSI M39 ↗ for the reporting framework. This demonstration does not claim full M39 compliance.

Educational software only. Not for diagnosis, treatment selection, dosing, or patient care.