Performance

Esprit Nova™ vs. 3 reviewers – Sleep Architecture Epoch-by-Epoch comparison (N=60)

Esprit Nova - 84%

Scorer 1 - 84%

Scorer 2 - 84%

Scorer 3 - 86%

Clinical Parameters

Table 1

Esprit Nova generated clinical parameter relative error distribution summary (percent of parameter values normalized to TST in % and within 10% error) e.g. 98% of studies have a Total W error less than 10% of SPT (Sleep Period Time).

Table 2

Scorer 1 generated clinical parameter relative error distribution summary (percent of parameter values normalized to TST in % and within 10% error) e.g. 91% of studies have a Total W error less than 10% of SPT (Sleep Period Time).

Table 3

Scorer 2 generated clinical parameter relative error distribution summary (percent of parameter values normalized to TST in % and within 10% error) e.g. 88% of studies have a Total W error less than 10% of SPT (Sleep Period Time).

Table 4

Scorer 3 generated clinical parameter relative error distribution summary (percent of parameter values normalized to TST in % and within 10% error) e.g. 93% of studies have a Total W error less than 10% of SPT (Sleep Period Time).

Esprit Nova™ vs. 3 reviewers – Apnea Hypopnea Index (N=60)

Apnea-hypopnea index vs. patient number.
Esprit Nova (AHI), Scorer 1 (REF1-AHI), Scorer2 (REF2-AHI), Scorer 3(REF3_AHI).

Esprit Nova™ vs. 3 reviewers – Arousal index (N=60)

Arousal index vs. patient number.
Esprit Nova (ARI), Scorer 1 (REF1-ARI), Scorer2 (REF2-ARI), Scorer 3(REF3_ARI). On the ordinate axis the units are events/h.

Best apnea diagnosis using Esprit Nova™

True positives per class and total for the estimation of severity of apnea

EN-Sc1 29-7-1-4 = 41
EN-Sc2 29-9-2-4 = 44
EN-Sc3 27-6-1-3 = 37

Sc1-Sc2 26-8-2-4 = 40
Sc1-Sc3 30-8-2-3 = 43
Sc2-Sc3 27-8-2-3 = 40

The (EN)-Scorer agreement for the classification of the severity of apnea, including detecting severe apnea is comparable to scorers, with best agreement between pairs being achieved by EN-Sc2(44) and EN-Sc1(41).

The agreement EN with two different scorers is higher than between any two of the scorers.

The most challenging aspect of OSA detection is the detection of very mild apnea.

The classification using EN is equivalent with the one from human scorers (29-29-27) ~ (26-30-27)
* Results originating from our pediatric study. Pediatric analysis available in Canada only.

Best microarousal detection using Esprit Nova™

The highest agreement in classification in severity classes is En-Sc1, En-Sc3. This result is illustrated in the graph, where the subject device (blue) scores in between the values scored by other scorers.

EN achieves better classification than the scorers when compared to the other scorers.

* Results originating from our pediatric study. Pediatric analysis available in Canada only.

Best leg movement detection using Esprit Nova™

2x2 Tables
PLM-i

Inter-Scorer Agreement

% (95% C1)
Correct classification 81.7 (71.9 - 91.5)
PPA 90.9 (75.5-97.5)

% (95% C1)
Correct classification 73.3 (62.1 - 84.5)
PPA 78.8 (61.9 - 89.5)

% (95% C1)
Correct classification 71.7 (60.3 - 83.1)
PPA 80.0 (63.7 - 90.1)

Esprit Nova- Scorers Agreement
Scorers

Stats in comparison to Average of Scorers
% (95% C1 )
Correct classification 83.3 (73.9 - 92.8
PPA 100 (87.9 - 100)

EN agreed 46-49-46 times with scorers, whereas inter-scorer agreement was 43-44- 49 on the same data set attesting for the equivalence of EN to the gold standard (in-lab manual scoring).

* Results originating from our pediatric study. Pediatric analysis available in Canada only.

The Esprit Nova™ System

Overview

Comparison

Performance