Ai-assisted interpretation

Ai-assisted interpretation has been studied in 2 papers since 2026: 1 trial and 1 observational study in people; 2 in the past year. The newest story, Oct 10, 2026: “Model predicts acute ischemic stroke risk in type 2 diabetes”.

Intervention

2 papers study it4 findingsUpdated

See all 2 papers on ai-assisted interpretation

At a glance

The findings reach people, from one lab or study. 2 independent labs in 2 countries.

“The intervention group, in comparison with the control group, showed no significant difference in interpretation time per examination (187 vs 172 seconds, respectively; p = .23)”
Artificial Intelligence-Assisted Lung Nodule Evaluation on Low-Dose Chest CT in Asymptomatic Individuals: A Prospective Randomized Controlled TrialAJR. American journal of roentgenology2026doi:10.2214/ajr.26.34552

None of its findings points the other way.

Humans: a single group. Cells, worms or flies and mice: nothing yet.
  • 2 independent labs in 2 countries
  • Funders: AstraZeneca and 3 more
  • Human trial1
  • Humans1

Studies in people

1 study gave ai-assisted interpretation to people and measured what changed.

What changedDesignPeopleSource

no change: Has no effect on interpretation time187 vs 172 seconds, p = .23

up: Increases lung nodule detection rate16.9% vs 10.3%, p = .03

up: Increases follow-up recommendations15.3% vs 7.4%, p = .04

Randomized trial911 peopleBothmean age, 62 yearsOlder adults; real-world clinical workflowsAJR. American journal of roentgenology, Mar 18, 2026

What did studies of ai-assisted interpretation find?

4

Studies of ai-assisted interpretation measured 4 outcomes, among them reader diagnostic correctness, Follow-up recommendations and lung nodule detection rate.

Reader diagnostic correctness

upin humans1

1 study
  1. Ai-assisted interpretation is linked with better reader diagnostic correctness in humans (AI-assisted carotid ultrasound interpretation by senior and junior readers).

    “Generalized estimating equation analyses showed significant improvements in diagnostic correctness among both senior and junior readers, with P values of 0.00235 and 0.000343, respectively.”

    HumansAI-assisted carotid ultrasound interpretation by senior and junior readers

    Model predicts acute ischemic stroke risk in type 2 diabetes

    Academic radiology

Follow-up recommendations

upin humans1

1 study
  1. Ai-assisted interpretation increases follow-up recommendations in humans (real-world clinical workflows) (15.3% vs 7.4%, p = .04).

    “frequency of follow-up LDCT recommendations (15.3% vs 7.4%, respectively; p = .04)”

    Human trialreal-world clinical workflowsn = 911

    Artificial Intelligence-Assisted Lung Nodule Evaluation on Low-Dose Chest CT in Asymptomatic Individuals: A Prospective Randomized Controlled Trial

    AJR. American journal of roentgenology

Lung nodule detection rate

upin humans1

1 study
  1. Ai-assisted interpretation increases lung nodule detection rate in humans (real-world clinical workflows) (16.9% vs 10.3%, p = .03).

    “significantly higher detection rate of Lung-RADS-positive nodules (16.9% vs 10.3%, respectively; p = .03)”

    Human trialreal-world clinical workflowsn = 911

    Artificial Intelligence-Assisted Lung Nodule Evaluation on Low-Dose Chest CT in Asymptomatic Individuals: A Prospective Randomized Controlled Trial

    AJR. American journal of roentgenology

Interpretation time

no changein humans1

1 study
  1. Ai-assisted interpretation showed no clear effect on interpretation time in humans (real-world clinical workflows) (187 vs 172 seconds, p = .23).

    “The intervention group, in comparison with the control group, showed no significant difference in interpretation time per examination (187 vs 172 seconds, respectively; p = .23)”

    Human trialreal-world clinical workflowsn = 911

    Artificial Intelligence-Assisted Lung Nodule Evaluation on Low-Dose Chest CT in Asymptomatic Individuals: A Prospective Randomized Controlled Trial

    AJR. American journal of roentgenology

Latest

Dec 2025: 0Jan 2026: 0Feb 2026: 0Mar 2026: 1Apr 2026: 0May 2026: 0Jun 2026: 0Jul 2026: 0Aug 2026: 0Sep 2026: 0Oct 2026: 1Nov 2026: 02 papers in 12 months