Rett Syndrome

Rett syndrome severity estimation with the BioStamp nPoint using interactions between heart rate variability and body movement

“…March 18, 2022 with feature popularity scores 0.97 and 0.93, respectively. In the top-10 most popular features for the H+T+N feature combination, two features were HRV-metrics (µ P RSA−DC and σ 2 P N N 50 ), three features were MSNR-features (σ 2 LOOP 4 (3) , σ 2 LOOP 4 (5) and µ T RACE (10) ) and the rest five were MSTE-features.…”

“…Variance (across all surrogate representations) of the number of 4−loops in the network representation of the tuple -(HR signal, activity count signal, DC signal) at the 3 rd coarse-graining scale (σ 2 LOOP 4 (3) ) (ρ = 0.81), ( 4) Variance (across all surrogate representations) of the number of 4−loops in the network representation of the tuple -(HR signal, activity count signal, DC signal) at the 5 th coarse-graining scale (σ 2 LOOP 4 (5) ) (ρ = 0.79), ( 5) Mean (across all surrogate representations) of the trace of the adjacency matrix of the network representation of the tuple -(HR signal, activity count signal, DC signal) at the 10 th coarse-graining scale (µ T RACE (10) ) (ρ = 0.76), ( 6) Variance (across all surrogate representations) of transfer entropy from the activity counts signal to the deceleration capacity signal at the 1 st coarse-graining scale (σ 2 τ (1) Act→DC ) (ρ = 0.47), (7) Variance (across all surrogate representations) of transfer entropy from the activity counts signal to the HR signal at the 1 st coarse-graining scale (σ 2 Mean (across all surrogate representations) of transfer entropy from the activity counts signal to the HR signal at the 4 th coarse-graining scale (µ τ (4)…”

“…The feature weight for (µ P RSA−DC ) was greater than zero for 93% of the time, suggesting an inverse relationship (due to the negative-log relationship between features and odds) between the feature values and the Rett disease severity was the most popular feature with a feature popularity score of 0.94. Finally, when we used the MSNR features alone, (σ 2 LOOP 4 (3) ) was the most popular feature with a feature popularity score of 0.90. Interestingly, the top features from the individual feature set models for HRV-metrics (µ P RSA−DC ), MSTE-features (σ 2…”

“…Finally, we visualized model performance and computed feature popularity scores. and loss of acquired skills resulting in psychomotor regression, development of stereotypical hand movements, and dysautonomia [3]. It leads to the deterioration of the autonomous nervous system, impacting breathing regularity, heart rate (HR), gut motility, and impairs motor planning and locomotion, resulting in significantly impaired mobility, no purposeful hand use, and largely absent verbal communication.…”

“…The feature µ P RSA−DC came out on top with a feature popularity score ρ = 1.00. It was followed by the following 9 features: (1) Variance (across all surrogate representations) of transfer entropy from the activity counts signal to the HR signal at the 8 th coarse-graining scale (σ Act→HR ) (ρ = 0.97), (2) Variance (across all surrogate representations) of transfer entropy from the activity counts signal to the deceleration capacity signal at the 8 th coarse-graining scale (σ 2 τ (8) Act→DC) (ρ = 0.93),(3)…”