User Manual Part 4

H-2
CLINICAL STUDY - SUMMARY OF GDT1000 SEN
SING ACUTE STUDY
Protocol Testing
Four scenarios were tested, including different combinations of sensed atrial
signals (AS), paced atrial signals (AP), sensed ventricular signals (VS), and
paced ve ntricular signals (VP), i.e., AS/VS, AS/VP, AP/VS, and AP/VP. Sensing
algorithm performance was analyzed from patients’ real-time electrograms
(EGM) and electronic signals. Primary analysis was performed by visually
reviewing the EGM and markers of the printed strips for proper sensing, as well
as for instances of undersensing and oversensing.
Additional analysis included tabulating the sensed and paced events stored
in the patient data files from the patient CD-ROM. During this tabulation,
unexpected events were noted. An example of an unexpected event is a
sensed event during an AP/VP testing scenario. The sensed event could be a
real event, such as a PVC, or an oversensed event. T hese unexpected events
were evaluated by viewing the electronic signals stored in the patient data files
and correlating these signals to the printed strips.
Statistical Analysis
The sensitivity, specificity, positive predictive value, rate of oversensing, and
rate of undersensing of the sensing algorithm were analyzed for each chamber.
A true positive ( T P ) is the number of intrinsic/p aced signa ls appropriately
sensed, a false positive (FP) is the number of intrinsic/paced signals from
the opposite chamber oversensed, a false negative (FN) is the number of
intrinsic/paced signals undersensed, and a true negative (TN) is the number
of intrinsic/paced signals from the opposite chamber appropriately not
sensed. The sensing sensitivity was calculated as TP/(TP+FN), specificity
as TN/(TN+FP), positive predictive value (PPV) as TP/(TP+FP), rate of
oversensing as FP/(TP+FP), and rate of undersensing as FN/(TP+FN).
The sensing performance results from the first phase of the study are provided
and compared to the results from the second phase in order to demonstrate
the improvement in the operation of the updated sensing algorithm following
the between-phase changes. Results from the second phase of the study are
the most clinica lly relevant, as they reflect the performance of the final sensing
algorithm implemented in the COGNIS/TELIGEN devices.
The GDT1000 protocol did not pre-specify acceptable sensitivities, specificities,
PPV, rates of oversensing, or rates of undersensing for the RA, RV, and LV
channels.
- DRAFT -