410
Y. A. Abdulhameed et al.
properties using wavelet transforms can often reveal statistically significant differences. A similar scenario was observed in the subsequent analysis of cardiovascular
dynamics in malaria and non-malarial states in some of the characteristic frequency
intervals.
The power spectra within individual frequency intervals are usually normalized
by dividing by their total powers. In the present case, this was done for the frequency
rage 0.005–2 Hz.
The differences observed in power spectra between malaria states and non-malaria
were further investigated in order to ascertain whether they could be used to distinguish between groups. Machine learning algorithms were then applied to parameters
extracted by spectral and coherence analysis to classify the FM, NFM and NM states.
26.4.2.1 Spectral Power in Blood Perfusion Fluctuations
The normalized time-averaged wavelet powers of left (BF1) and right (BF2) blood
flow oscillations for the three groups are shown in Fig. 26.1.
For the left ankle and right ankle blood flows, the febrile malaria group had
significantly lower normalized power than NM in the 0.07–0.1 and 0.8–1.4 Hz frequencies associated with the myogenic and cardiac activities respectively, and a
significantly higher normalized power than NM in 0.13–0.16 Hz, the neurogenic
and NO-dependent endothelial frequency intervals. A major shift was seen in the
cardiac oscillatory peak between the febrile malaria and non-malaria groups, with
that of the febrile malaria and non-malaria groups found at ∼1.83, ∼1.02 Hz in the
left ankle blood flow and at ∼1.8, ∼1.04 in the right ankle blood flow respectively
(Fig. 26.1a, b). As was the case for the febrile malaria group, the cardiac peak frequency was slightly higher in non-febrile malaria (found at ∼1.44 Hz, ∼1.45 Hz in
left ankle and right ankle blood flow respectively) than non-malaria. In a similar pattern to that of febrile malaria, non-febrile malaria exhibits a lower normalized power
in the 0.07–0.1 Hz but was not statistically significant, and a higher normalized
power around the neurogenic and NO-dependent endothelial frequency intervals all
of which are statistically significant except at ∼0.02 Hz in left ankle blood flow and
0.14–0.16 Hz in right ankle blood flow, when compared to non-malaria (Fig. 26.1c,
d).
Comparisons of the normalized power between FM and NFM revealed differences
in the cardiac interval, with the difference being significant only in right ankle blood
flow—noting that the frequencies of their cardiac peaks also differ. At frequencies
0.04–0.05 Hz within the neurogenic interval, NFM was found to exhibit a higher
normalized power although this was only statistically significant in the right ankle
blood flow (Fig. 26.1e, f).
The box-plots in Fig. 26.1 compare the normalised power spectral component of
the LDF blood flows within the intervals investigated for different groups. The FM
group was found to have cardiac oscillations of lower power than either the NFM
or NM groups in blood flow recorded from the right ankle, and a similar significant
difference in the febrile malaria group was observed in the left ankle blood flow
Y. A. Abdulhameed et al.
properties using wavelet transforms can often reveal statistically significant differences. A similar scenario was observed in the subsequent analysis of cardiovascular
dynamics in malaria and non-malarial states in some of the characteristic frequency
intervals.
The power spectra within individual frequency intervals are usually normalized
by dividing by their total powers. In the present case, this was done for the frequency
rage 0.005–2 Hz.
The differences observed in power spectra between malaria states and non-malaria
were further investigated in order to ascertain whether they could be used to distinguish between groups. Machine learning algorithms were then applied to parameters
extracted by spectral and coherence analysis to classify the FM, NFM and NM states.
26.4.2.1 Spectral Power in Blood Perfusion Fluctuations
The normalized time-averaged wavelet powers of left (BF1) and right (BF2) blood
flow oscillations for the three groups are shown in Fig. 26.1.
For the left ankle and right ankle blood flows, the febrile malaria group had
significantly lower normalized power than NM in the 0.07–0.1 and 0.8–1.4 Hz frequencies associated with the myogenic and cardiac activities respectively, and a
significantly higher normalized power than NM in 0.13–0.16 Hz, the neurogenic
and NO-dependent endothelial frequency intervals. A major shift was seen in the
cardiac oscillatory peak between the febrile malaria and non-malaria groups, with
that of the febrile malaria and non-malaria groups found at ∼1.83, ∼1.02 Hz in the
left ankle blood flow and at ∼1.8, ∼1.04 in the right ankle blood flow respectively
(Fig. 26.1a, b). As was the case for the febrile malaria group, the cardiac peak frequency was slightly higher in non-febrile malaria (found at ∼1.44 Hz, ∼1.45 Hz in
left ankle and right ankle blood flow respectively) than non-malaria. In a similar pattern to that of febrile malaria, non-febrile malaria exhibits a lower normalized power
in the 0.07–0.1 Hz but was not statistically significant, and a higher normalized
power around the neurogenic and NO-dependent endothelial frequency intervals all
of which are statistically significant except at ∼0.02 Hz in left ankle blood flow and
0.14–0.16 Hz in right ankle blood flow, when compared to non-malaria (Fig. 26.1c,
d).
Comparisons of the normalized power between FM and NFM revealed differences
in the cardiac interval, with the difference being significant only in right ankle blood
flow—noting that the frequencies of their cardiac peaks also differ. At frequencies
0.04–0.05 Hz within the neurogenic interval, NFM was found to exhibit a higher
normalized power although this was only statistically significant in the right ankle
blood flow (Fig. 26.1e, f).
The box-plots in Fig. 26.1 compare the normalised power spectral component of
the LDF blood flows within the intervals investigated for different groups. The FM
group was found to have cardiac oscillations of lower power than either the NFM
or NM groups in blood flow recorded from the right ankle, and a similar significant
difference in the febrile malaria group was observed in the left ankle blood flow