Educational resource · Not a medical device · Infectious diseases review in progress · Does not replace clinical judgement
How the data is selected, how levels of evidence are computed from the type of source cited, and what this atlas does not say. Educational resource, not a medical device.
This section is available in French only. Its content is clinical prose that has not been translated — see the notice above.
SeptiMap is currently developed and maintained by its author. An independent external infectious-disease review is in progress; it is not finished. This transparency is deliberate and is part of the tool's credibility.
SeptiMap is a visual educational tool intended for healthcare professionals and healthcare students. Its purpose is to help understand the anatomical sites, mechanisms and infectious foci associated with the main documented pathogens.
It is not a clinical decision-support device, a diagnostic tool, or an individualised treatment guide. The medical decision belongs exclusively to the healthcare professional.
The texts of this site were written with the help of an artificial-intelligence assistant, then reviewed by the author alone. No germ-organ association has been validated by a third party. The level of evidence is computed by a fixed rule from the type of source cited (meta-analysis or randomised trial: A; guideline or cohort: B; case series, expert opinion or review: C; no source: D) — it is not a GRADE assessment. Source verification covers the existence and consistency of DOIs (Crossref), not their content.
The data are mainly European/French in scope. Prevalences and resistance patterns vary by region.
The English version does not translate the medical content. It separates two things: standardised terminology and clinical prose.
Standardised terms — species names (LPSN), resistance phenotypes (ECDC/CDC, EUCAST), anatomical structures (Terminologia Anatomica), international non-proprietary names of molecules (WHO INN), syndrome and exposure names (MeSH) — are shown in their official international form. Writing MRSA instead of SARM is not translation: it is citing the registry. Every English term on this site comes from a single table, with its registry and reference, checked at every build; a term found in no registry stays in French, and the card shows the Latin name.
Clinical prose — notes, key points, dosing, evidence-level justifications — remains in French, marked “FR” in the English interface, until an English-speaking clinician has reviewed it. A unit or nuance error in machine-translated dosing is exactly the fault this site cannot afford. A physician abroad can therefore start from an organism and find its sites, their level of evidence and their sources — or start from an organ and find the organisms — without reading a word of French; the prose is a complement, flagged as such.
Each anatomical tropism relies on indexed primary medical literature (PubMed) and on official guidelines from learned societies. The bibliography has been fully re-verified: every cited DOI resolves to the announced article (title / journal / year checked against Crossref).
Limitation being addressed: some tropism prevalences are not yet backed by an individual source and are flagged as level of evidence D ("unsourced data"). This traceability work is ongoing; the cards concerned are explicitly marked.
No dose, source or DOI is added "to fill a gap": a reference that cannot be verified is removed rather than kept. When a tropism is not documented by an identifiable source, it is either classed "Not documented" or kept with an explicit level of evidence D.
The level of evidence is computed, never assigned: a fixed rule, applied to every focus, derives it from the type of the sources cited (publication type recorded in the bibliography). It is not a GRADE assessment (GRADE Working Group, BMJ 2004;328:1490): the letter says what is cited, not the methodological quality of each study.
None of the five GRADE judgements (risk of bias, inconsistency, imprecision, indirectness, publication bias) is documented per tropism; the letter only reflects the type of source cited, by the rule below — no appraisal goes into it.
Recalibration (September 2026): the prevalence of an organ tropism almost always comes from observational data; grade A is now reserved for tropisms supported by a verified meta-analysis or randomised trial. Tropisms not backed by an individual source are classed D ("unsourced"). Current distribution of the 249 tropisms: A 9 · B 156 · C 69 · D 15.
One row per pathogen → organ association: tropism level, level of evidence (computed from the type of source cited), route of spread, cited sources and their DOIs. Undocumented tropisms are included, labelled as such — removing them would give a more flattering dataset than the real corpus.
Licence: CC BY 4.0 — chosen by the publisher (10 September 2026). Reuse permitted, including commercial reuse, provided SeptiMap is credited and the source is linked; the notice and the URL are repeated in the exported files, which will circulate without this page.
The 3D anatomical models come from the Human Reference Atlas (HRA) / HuBMAP (NIH / National Library of Medicine), humanatlas.io. They are used under the CC BY 4.0 licence (Creative Commons Attribution).
Some anatomical structures not available in the HRA (ears, sinuses, thyroid) are represented by simplified procedural shapes, labelled in the atlas as “schematic shape — not from a measured anatomical model”. Lymph nodes, jugular veins and the mediastinum have no 3D shape: they appear in the list of foci with the note “no 3D shape available”, and nothing is drawn in their place. A skin focus (face, trunk, limbs) is a tinted area of the silhouette itself, never an added object. The meninges and pleurae do not exist as separate models: when they carry a focus, the atlas draws them as a thin outline of their parent organ and labels them “schematic envelope derived from the [parent organ] — not from a measured anatomical model”. The heart valves come from the measured HRA heart; the rest of the endocardium remains attached to the whole heart. The stomach, adrenal glands, testes and joints come from BodyParts3D (DBCLS, CC BY 4.0), registered into the Visible Human frame by a measured transformation, not adjusted by hand.
In the 3D atlas, colour does not say which organ: it says only the documented tropism level. Three roles, nothing else: the silhouette (neutral grey, translucent), structures not concerned by the displayed agent (desaturated grey, the skeleton fainter), and focus structures — a single tint in three intensities that are exactly the three levels of the legend (high, moderate, rare). No anatomical tint (the liver is not brown, the gallbladder is not green), no hatching, no shadow: those were additional visual languages, and they made it impossible to know what a colour meant.
An organ is named by its label and by the clinical card, never by its tint. Editorial decision of 16 September 2026, following a visual audit of the atlas.
Dr Clément Médeau — MD, general practitioner, La Rochelle (France). Any error found can be reported to: contact@septimap.fr
Data last updated: 2026-11-10
This page is the text version, generated at build time. The 3D anatomical atlas requires JavaScript.