Multiaxial Diagnostics Is Back — Modernised, Digital and Open Source

A design concept with a functional prototype. Not a clinical recommendation — a research tool for further development by the community.

Author: Lukas Geiger (LG). Published as a preprint on Zenodo.

The Problem: What Was Lost in 2013

When the American Psychiatric Association published DSM-5 in 2013, one of the most influential tools in psychiatric diagnostics disappeared: the multiaxial system. Since 1980, it had enabled clinicians to systematically and separately record mental disorders (Axis I), personality disorders (Axis II), medical conditions (Axis III), psychosocial stressors (Axis IV), and level of functioning (Axis V).

The rationale for its abolition: the system was too complex, the separation between Axis I and II clinically untenable, the fifth axis (GAF scale) psychometrically questionable. All of this is true. But with the multiaxial system, something was also lost that no replacement has brought back since: the structured compulsion to think about the whole person — not just their symptoms.

What has happened since: in the specialist literature, a creeping narrowing of diagnostics to individual disorder profiles is being discussed. The biopsychosocial context — housing situation, employment, social isolation, physical illness — ends up in a free-text field at best. Often it ends up nowhere.

The Solution: A 6-Axis Model for 2026

The Multiaxial Diagnostic Expert System is an attempt to bring back the strengths of the old system — without repeating its weaknesses. It extends the original 5-axis model to six axes and, for the first time, integrates three classification systems in a single tool:

Axis	Content	Classification
I	Mental Disorders	DSM-5-TR
II	Personality and Development	DSM-5-TR (AMPD)
III	Medical Condition Factors	ICD-11
IV	Psychosocial and Environmental Stressors	ICD-11 (Z-Codes)
V	Level of Functioning and Disability	ICF
VI	Clinical Global Assessment	Integrative

Three design decisions distinguish it from the old system:

1. Formal coverage analysis. The symptom interface (a subcomponent of Axis I) computes a set-based metric C(S) that systematically identifies unrecognised symptoms. Not: “Did the clinician think of everything?” But: “What has not been covered yet?” — mathematically, not intuitively.

2. Symmetrical Axis I/III architecture. Psychologists and physicians work with identical structural tools. The artificial separation between “mental” and “somatic” is not abolished (it has its justification), but it becomes bridgeable.

3. Gatekeeper logic. A 6-step algorithm following First (2024) implements hierarchical exclusion rules as a state machine: Substance-induced? Medically caused? Psychotic? — before any specific diagnosis is made. This is not AI diagnostics, but formalised clinical logic.

What the Prototype Can Do

The system exists not only as a paper, but as a functional open-source prototype:

Diagnostics interface (Streamlit): Guided process through all 6 axes, PRO/CONTRA evidence evaluation with confidence estimates per diagnosis, CAVE Clinical Alerts for cross-axis risk management
Diagnostic test centre (Flask): 16 validated screening instruments (PHQ-9, GAD-7, PCL-5, AUDIT, C-SSRS, and others), remote access via link sharing, automatic scoring, REST API
Bilingual (DE/EN): 661 internationalisation keys, fully translated
~2,850 lines of code, MIT licence (software) + CC-BY 4.0 (papers)

What the Prototype Cannot Do

Honesty is mandatory: the system is a design concept with a functional prototype — not a validated diagnostic tool. There is no pilot study, no N=X, no empirical validation. The coverage metric is formally defined but not clinically tested. The gatekeeper logic maps published algorithms but has not been tested in a clinical environment.

This is not a shortcoming, but a deliberate positioning: the system is meant to be a research tool that the community can further develop, test, and — hopefully — validate. We provide the architecture. The clinical testing must come from clinicians.

The Real Problem: Integration, Not Diagnostics

The tools for psychiatric diagnostics exist. PHQ-9 for depression, GAD-7 for anxiety, PCL-5 for trauma, AUDIT for alcohol, C-SSRS for suicidality — the list of validated screening instruments is long. The problem is not that they are missing. The problem is that they are scattered, expensive, and not integrated.

In practice, this looks like: a clinician suspects a comorbid anxiety disorder alongside the depression. They need the GAD-7. It sits in another system, costs per administration, and must be scored separately. The results end up in a PDF that is linked to the primary diagnosis nowhere. For the next patient, the same process — different instrument, different system, different costs.

What this tool does differently: All 16 screening instruments in one place, free of charge, with automatic scoring, remote access via link, and a REST API. No login, no subscription, no vendor lock-in. A patient can fill out the PHQ-9 at home; the clinician sees the result immediately — integrated into the overall diagnostic context across all six axes.

This is not a revolution. It is something far more modest: a free building block for improving diagnostic infrastructure. The technology for this exists. The validated instruments exist. What was missing was a place that brings both together — open, integrated, and without a paywall.

Why This Is an Um:bruch Topic

Who gets diagnosed determines what help a person receives — or does not receive. Diagnostics is not a purely medical topic, but a question of access. When screening instruments cost money and integration requires effort, people with resources get better diagnostics than people without. That is not a law of nature — it is an infrastructure problem.

When diagnostics additionally narrows to symptom checklists and ignores the social context, structural problems (poverty, isolation, homelessness) are reinterpreted as individual disorders. The multiaxial system was a counterweight to this. Its disappearance left a gap that remains unfilled to this day.

This project is a proposal for how it could be filled — digitally, transparently, free of charge, and open.

Conflict of Interest

Transparency notice: The author of this blog post, Lukas Geiger, is also the developer of the system presented here. Um:bruch publishes this post because the topic — free access to diagnostic infrastructure — falls within the thematic area of health and participation. Assessment and evaluation remain the responsibility of the readership and the professional community.