Search Results

The new device is a modification to previously reviewed submittals containing the EASI algorithm. The modification is a software-based change that slightly improves existing correlation coefficients. The modification also allows reconstruction to simulate a Mason Likar lead placement.

AI/ML Overview

Acceptance Criteria and Device Performance

Metric	Acceptance Criteria (Explicit or Implied)	Reported Device Performance
Correlation Coefficients	The modified coefficients should not show a decreased correlation in any of the leads compared to traditional 12-Lead ECG and Mason Likar 12-Lead ECG.	The modified coefficients did not show a decreased correlation in any of the leads. Several of the leads showed improvements, with the improvement in the correlation coefficient varying from 0 to 0.10 in one case.
Patient Risk	The modification should not increase patient risk relative to what is on the market today.	"It was therefore concluded that the modification does not increase patient risk relative to what is on the market today. It does improve effectiveness in some of the leads based on the data gathered." This suggests that maintaining or improving effectiveness without increasing risk was the implied acceptance criterion.

Study Information:

Sample size used for the test set and the data provenance: Not explicitly stated but the text mentions "earlier studies on the same patients" when discussing the comparison of correlation coefficients, implying a retrospective analysis of existing data. The number of patients in these "earlier studies" is not specified. The country of origin of the data is not specified.
Number of experts used to establish the ground truth for the test set and the qualifications of those experts: "2 physician over-reads" were involved in comparing the modified coefficients to traditional 12-Lead ECG and Mason Likar 12-Lead ECG. Their specific qualifications are not detailed beyond "physician."
Adjudication method for the test set: Not explicitly stated. The text mentions "2 physician over-reads comparing the modified coefficients," but does not clarify if there was a formal adjudication process if the physicians disagreed.
If a multi reader multi case (MRMC) comparative effectiveness study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance: No, this was not a comparative effectiveness study involving human readers. The study focused on the performance of the algorithm itself in terms of correlation coefficients.
If a standalone (i.e., algorithm only without human-in-the-loop performance) was done: Yes, the study primarily evaluated the standalone performance of the algorithm's modified coefficients by comparing them to established ECG standards. The "2 physician over-reads" serve as a form of validation against human interpretation of established methods, but the core performance metric (correlation coefficient) is algorithmic.
The type of ground truth used (expert consensus, pathology, outcomes data, etc.): The ground truth was established by "traditional 12 Lead ECG and Mason Likar 12 Lead ECG" which were then reviewed by "2 physician over-reads." This implies a form of expert consensus based on established clinical methods.
The sample size for the training set: Not specified. The document only refers to "earlier studies on the same patients" for validation, without detailing a separate training set.
How the ground truth for the training set was established: Not specified, as details about a distinct training set are absent. It can be inferred that the ground truth for any underlying algorithm development would have been established using similar methods (traditional ECGs and expert interpretation) as those used for validation.

Ask a Question

Ask a specific question about this device

Page 1 of 1