Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.brainworkup.org/llms.txt

Use this file to discover all available pages before exploring further.

The neurotyp-luria extension is Luria Voice’s AI-assisted report format. At version 0.1.0 it targets what the project calls “LLM-templated patients” — evaluation reports where narrative sections are drafted or refined by a locally running large language model rather than written entirely by hand. The extension integrates with Ollama, a local inference server, so patient data never leaves your machine.

​
What “LLM-templated patients” means

In standard Luria Voice usage you write every narrative sentence yourself, using the R package functions from cingulate to generate score tables and visualisations. With neurotyp-luria, you can instead provide structured patient data — test scores, demographics, referral question — and let a locally hosted LLM draft the prose narrative for each cognitive domain section. You then review, edit, and approve the output before finalising the report. This workflow is named after Alexander Luria, the Soviet neuropsychologist who pioneered qualitative, hypothesis-driven assessment. The Luria approach emphasises interpretation over rote score reporting — an orientation that aligns well with LLM-assisted narrative drafting, where the clinician’s conceptual framing guides the generation and editing process.
AI-generated narrative text must be reviewed and edited by a licensed clinician before inclusion in a signed evaluation report. neurotyp-luria is a drafting aid, not a substitute for professional clinical judgment.

​
Prerequisites

The neurotyp-luria extension requires a local Ollama installation to generate narrative sections. Install Ollama from ollama.com and pull a model before rendering. 
# Install a model — llama3.2 is a reasonable starting point
ollama pull llama3.2
You do not need Ollama to install or render the template in non-LLM mode. If no Ollama backend is available, the narrative sections render as placeholders.

​
Install the extension

Scaffold a new project from the template.
quarto use template brainworkup/neurotyp-luria

​
Render a report

quarto render template.qmd --to neurotyp-luria-typst
To preview while editing: 
quarto preview template.qmd
To update to the latest release: 
quarto update extension brainworkup/neurotyp-luria

​
Configure the report

Complete the front matter in template.qmd as you would for any other format. 
---
title: NEUROCOGNITIVE EXAMINATION
patient: Nguyen
name: Nguyen, Minh
dob: "1991-03-17"
age: 34
doe: "2025-05-05"
date_of_report: last-modified
case_number: "2025-L-003"
format: neurotyp-luria-typst
---
See Front matter reference for every supported field.

​
Luria methodology in Luria Voice

Alexander Luria’s approach to neuropsychological assessment moves away from fixed batteries and norm-referenced interpretation toward hypothesis-driven, qualitative analysis. A Luria-style report integrates test performance with behavioral observation, developmental history, and a coherent conceptual account of the patient’s neurocognitive profile. neurotyp-luria supports this methodology in two practical ways:
  1. Flexible section composition — include only the domain partials relevant to your hypothesis about this patient, rather than reporting every subtest in a standardised order.
  2. LLM-assisted narrative drafting — use the Ollama integration to generate an initial Luria-style interpretive narrative from structured score data, then refine it to match your clinical formulation.
Start with a conceptual summary of the patient’s presenting concerns and your primary hypothesis before generating narrative sections. This gives the LLM the interpretive frame it needs to produce clinically coherent prose rather than generic summaries.

​
Add cognitive domain sections

Edit _domains_to_include.qmd to include only the domains central to your clinical hypothesis for this patient. 
{{< include _neurocognitive.qmd >}}
{{< include _memory.qmd >}}
{{< include _attention.qmd >}}
See Modular section partials for how the include pattern works and how to write your own domain partial files.
Last modified on May 20, 2026