The Rayong study: 218 participants in clinical practice

01

Sources and evidence status

This page draws on a Thai and English abstract and the EasyEF Lunch Symposium CNF 17th presentation of 26 March 2026, which presents the method and summarized results. A full publication and peer-review status for this dataset have not been verified.

The appropriate description is “Rayong study summary,” with its setting stated when quoting figures. The publication status of the 2024 paper must not be transferred to this later dataset; they are separate evidence sources.

Source: Rayong abstract, pp. 1–3; CNF 2026, pp. 54–59

02

Participants and dates

The abstract describes a retrospective study of 218 participants. Slide 55 identifies the two-month period as December 2025–January 2026 at Rayong Hospital. Participants had two different assessment indications, which should be reported separately.

Source: Rayong abstract, pp. 1–2; CNF 2026, p. 55

03

How results were compared

Ultrasound-display videos were recorded with a smartphone camera and classified by EasyEF into poor, fair, and good LV function. Outputs were then compared with clinician assessments. The slides describe registered-nurse operation and confirmation by a cardiologist.

This provides information about operation within a clinical service, but the abstract does not report every relevant detail, such as results by device model or model version. It therefore cannot establish performance in every environment.

Source: Rayong abstract, pp. 1–2; CNF 2026, p. 55

04

Overall and LV-category results

The study team reports overall accuracy of 93.6%, with 100% for poor LV, 96.9% for good LV, and 27.3% for fair LV. These values describe this cohort and are not accuracy guarantees for individual people or other institutions.

The 100% poor LV result does not mean future errors are impossible. Nor should the abstract’s “accuracy” values be relabeled as sensitivity or specificity without the supporting counts and definitions.

Source: Rayong abstract, pp. 1–3; CNF 2026, pp. 57, 59

05

Results by assessment indication

Overall accuracy was reported as 98.7% for pre-chemotherapy assessment and 90.8% for general LV function assessment. These figures show how a single overall value can obscure differences between service populations.

This difference does not establish that the assessment indication itself caused better performance. Interpretation requires LV-category composition, image quality, and further methodological detail. The pre-chemotherapy result does not authorize bypassing standard assessment before treatment.

Source: Rayong abstract, pp. 1–3; CNF 2026, p. 59

06

Interpreting the summary responsibly

CNF page 58 presents a screening and referral scenario based on this cohort. It is a service interpretation by the presenter, not a controlled trial demonstrating referral reduction in every setting. This website therefore does not use that scenario as an outcome guarantee.

The Thai and English abstracts give different p-value thresholds. This summary emphasizes the consistent sample size and accuracy figures instead of choosing a threshold without clarification. Further work needs to evaluate fair LV and multisite settings while retaining clinician decision-making.

Source: Rayong abstract, pp. 1–3; CNF 2026, pp. 58–60; EasyEF project brief, 22 June 2026