Tuesday, February 4, 2025

AI in Action: Chat GPT Explains WHY a Clinical Utility Paper is a Clinical Utility Paper

This week, we have a press release from neurology diagnostics firm CN2, about a new clinical utility study of their test, PRECIVITY AD2.   The paper is Monane et al, 2025.

Here's a highlight posting at Linked In.

Here's the open access article at the journal Diagnostics.

I think of a Clinical Utility study for diagostics like this - you run a test, have a result, give to the clinician.  At that point something changes - diagnoses, drug prescriptions, or outcomes.

I found it a bit hard to extract exactly what was going on from the abstract.   Chat GPT read the paper and gave the summary below.   (And at bottom, I clip the authors' original abstract).

###

AI CORNER

###

This paper qualifies as a clinical utility study because it demonstrates how the PrecivityAD2 blood biomarker test influences real-world clinical decision-making for patients with cognitive impairment, following the model you described:

  1. Test Performed: The PrecivityAD2 blood test measures plasma Aβ42/Aβ40 and p-tau217/np-tau217 ratios to generate an Amyloid Probability Score (APS2), indicating the likelihood of Alzheimer’s disease (AD) pathology.

  2. Results Provided to Clinicians: Memory specialists received APS2 scores for 203 cognitively impaired patients.

  3. Impact on Patient Care:

    • Diagnoses Changed: The probability of an AD diagnosis shifted significantly based on the test results. Clinicians’ diagnostic certainty either increased (for positive APS2) or decreased (for negative APS2), with changes noted in 75% of cases.
    • Treatment Altered: Anti-AD medication prescriptions decreased for patients with negative APS2 results (from 74% to 19%) and increased for those with positive results (from 71% to 96%).
    • Further Testing Adjusted: Additional brain amyloid imaging decreased, particularly for APS2-negative patients, reflecting reduced reliance on more invasive or expensive confirmatory tests.

In essence, the study shows that actionable changes in diagnosis, treatment plans, and resource utilization occurred as a direct result of the blood test outcomes, fulfilling the criteria for demonstrating clinical utility.  (I would treat the AI text as a smooth first draft, but only that).

##

This morning, I haven't double checked the AI findings which I would do, if it were part of my regular workstream.

###
Here's the original abstract.


###
Separately, CN2 won a $1.5M grant from MJ Fox Fndtn to study Alzheimer-Parkinson's links.