Yes, a short automated speech test can help distinguish the three main variants of primary progressive aphasia, a dementia that first impairs language. JAMA Neurology reports the model reached 79% overall accuracy with mean cross-validated AUC 0.90 in 214 participants the JAMA Neurology study. Researchers at UCSF and UT Austin built the test from 1-2 minutes of picture-description speech. They measured 40 linguistic and acoustic features and used Lasso modeling to create variant-specific scores.
Medical information disclaimer: This article is for general educational purposes only and does not provide medical advice, diagnosis, or treatment. Always consult a physician or other qualified health professional about symptoms, medications, tests, or treatment decisions.
Table of Contents
- What does the test measure?
- How accurate is it for each type?
- Does speech match brain biology?
- What can families do with this now?
What does the test measure?
The task is simple. The person describes a picture for one to two minutes. Software tracks word choice, grammar, pauses, speech rate, and sound patterns.
Those 40 measures combine into three profile scores, one for each variant. JAMA Neurology reports nonfluent, logopenic, and semantic patterns each produce a distinct profile. A higher score points toward one variant over the others.
How accurate is it for each type?
Sensitivity was 86% for nonfluent PPA, 82% for semantic PPA, and 70% for logopenic PPA. Specificity was 91%, 90%, and 88% in the same order.
Logopenic PPA was hardest to detect. That matters because word-finding pauses overlap with other variants in brief samples. A low logopenic score does not rule out language decline.
Does speech match brain biology?
The same speech scores tracked expected brain areas in voxelwise MRI analyses. They also separated the most common underlying pathologies in an autopsy-confirmed subset the brain-imaging and autopsy analysis.
A separate UCSF reading-task system reached AUC 0.798 for nonfluent versus logopenic PPA. An acoustic-only approach reached 88% balanced accuracy for PPA versus controls but only 73% for logopenic versus nonfluent cases.
What can families do with this now?
The authors state the 1-2 minute test could aid diagnosis and monitoring where specialist speech-language expertise is unavailable. It is a support tool, not a solo diagnosis.
Clinical history, exam, and imaging still guide care. Use results to prompt referral to neurology, neuropsychology, or speech-language therapy when decline persists the UCSF Memory and Aging Center summary.
- Record concerns with dates and examples of speech change
- Bring a brief speech sample topic to the visit, such as picture description
- Ask whether the clinic uses structured language screening
- Ask what follow-up will track change over time
