result of natural fluctuations of the corresponding parameter exceeding L B caused by
changes in a physiological state.
2. Increase in T o , on the one hand, improves
reliability for bioindicators reaction detection
as the temporary difference of responses for
two test organisms can exceed T o , and on
the other hand, the probability of false
alarms increases as two independent transitions of bioindicators to the excited state
(especially at low values L B ) can lead to
emergency alarms. The similar can be stated
for parameter T d . When increased T d , both
the fast and slow processes leading to
increase of HR will be taking into account,
i.e. reliability to detect the transition to stress
increase. However, such slow transitions,
most likely, are connected with natural
changes of bioindicators physiological state,
and not associated with the toxicity. In this
case, the temporary delay between the onset
of toxic deposition and alarm in the
BioArgus-W system increases.
Based on the data analysis for some previous
years for each of the main parametrs (SI and
dHR), three values (1000, 5000 and 10,000 s
–3
for SI and 50, 75 and 100% for dHR) were
chosen as a threshold levels. Besides it, the dHR
was calculated for three values of T d : 2, 10 and
30 min (Kinebas et al. 2012).
In processing data on daily cardiac activity of
bioindicators, irregularly excesses of threshold
level L B were detected. Thus, mentioned values
do not exceed L B within a day at all, or exceed it
from time to time. If the dHR values 50, 75 and
100% were chosen as a threshold levels L B , the
average periods of false alarms (T e ) for a 24-hour
interval were calculated for three values of T d : 2,
10 and 30 minutes (Kinebas et al. 2012), based
on the BioArgus-W crayfish HR data analysis for
the time of conventional water quality and the
false alarms probability per day (T e /1440 min)
were calculated. So, if L B = 100% − T e = 9, 4
and 2 min (for T d = 30, 10 and 2 min), if L B =
75% − T e = 18, 8 and 3 min (for T d = 30, 10 and
2 min) and if L B = 50% − T e = 43, 22 and 4 min
(for T d = 30, 10 and 2 min).
Having calculated for each of the installed
system BioArgus-W channel, the probability of
an event when parameters exceeds L B levels
within a day, and also average values for T e , it is
possible to decrease the probability of false alarm
generation. P(A) could be calculated by the followed formula:
P A
ð Þ ¼ P A=B
ð
ÞÁP B
ð Þ;
ð10:2Þ
where P(A)—probability of false alarms; P(B)
—probability of an event when a biomarker
excesses L B level within a day; P(A|B)—probability of a false alarm given event B has occurred.
The probability of P(A|B) is a ratio of time T e
(in minutes), when HR exceeds L B and a total
minutes in one day (1440 min). So, for two
bioindicator P 2 (2) = 0.030−0.015 (T e = 40–20
min, if L B = 50% T d = 30–10 min) and false
alarm may take place 1 time a 30 day.
Results of the toxicological experiments
showed that the response times of biosensors to
high concentration of toxic substance were from
1 till 5–10 min, as at Fig. 8 (Kholodkevich et al.
2008; Kinebas et al. 2012) and the typical dHR
values are 100–150%. Considering these results,
it is possible to recommend the values of the
BioArgus-W setup parameters to be chosen by
the following algorithm:
1. Taking into account technological, regulation
and economic requirements, the maximum
operating time of station (i.e. a temporary
delay between the beginning of influence and
the system alarm – T o.max is chosen as well as
the optimal (or maximum) number of the
predicted false alarms per year. (Usually To =
10–20 min and T d = 10 min as a biosensor
response time is less, than 10 min and the
stress- induced HR high value is conserved
for more, then 10 min after the primary rise).
2. It was shown in Karmazinov et al. (2007) that
the values of the corresponding parameters
(L B , T o for SI and L B , T o , T d for dHR) which
correspond to the required number of false
alarms per year with the assumption:
T o < T o.max .
3. The minimum L B values and the maximum
values of parameters T o and T d have to be
138
S. V. Kholodkevich et al.
changes in a physiological state.
2. Increase in T o , on the one hand, improves
reliability for bioindicators reaction detection
as the temporary difference of responses for
two test organisms can exceed T o , and on
the other hand, the probability of false
alarms increases as two independent transitions of bioindicators to the excited state
(especially at low values L B ) can lead to
emergency alarms. The similar can be stated
for parameter T d . When increased T d , both
the fast and slow processes leading to
increase of HR will be taking into account,
i.e. reliability to detect the transition to stress
increase. However, such slow transitions,
most likely, are connected with natural
changes of bioindicators physiological state,
and not associated with the toxicity. In this
case, the temporary delay between the onset
of toxic deposition and alarm in the
BioArgus-W system increases.
Based on the data analysis for some previous
years for each of the main parametrs (SI and
dHR), three values (1000, 5000 and 10,000 s
–3
for SI and 50, 75 and 100% for dHR) were
chosen as a threshold levels. Besides it, the dHR
was calculated for three values of T d : 2, 10 and
30 min (Kinebas et al. 2012).
In processing data on daily cardiac activity of
bioindicators, irregularly excesses of threshold
level L B were detected. Thus, mentioned values
do not exceed L B within a day at all, or exceed it
from time to time. If the dHR values 50, 75 and
100% were chosen as a threshold levels L B , the
average periods of false alarms (T e ) for a 24-hour
interval were calculated for three values of T d : 2,
10 and 30 minutes (Kinebas et al. 2012), based
on the BioArgus-W crayfish HR data analysis for
the time of conventional water quality and the
false alarms probability per day (T e /1440 min)
were calculated. So, if L B = 100% − T e = 9, 4
and 2 min (for T d = 30, 10 and 2 min), if L B =
75% − T e = 18, 8 and 3 min (for T d = 30, 10 and
2 min) and if L B = 50% − T e = 43, 22 and 4 min
(for T d = 30, 10 and 2 min).
Having calculated for each of the installed
system BioArgus-W channel, the probability of
an event when parameters exceeds L B levels
within a day, and also average values for T e , it is
possible to decrease the probability of false alarm
generation. P(A) could be calculated by the followed formula:
P A
ð Þ ¼ P A=B
ð
ÞÁP B
ð Þ;
ð10:2Þ
where P(A)—probability of false alarms; P(B)
—probability of an event when a biomarker
excesses L B level within a day; P(A|B)—probability of a false alarm given event B has occurred.
The probability of P(A|B) is a ratio of time T e
(in minutes), when HR exceeds L B and a total
minutes in one day (1440 min). So, for two
bioindicator P 2 (2) = 0.030−0.015 (T e = 40–20
min, if L B = 50% T d = 30–10 min) and false
alarm may take place 1 time a 30 day.
Results of the toxicological experiments
showed that the response times of biosensors to
high concentration of toxic substance were from
1 till 5–10 min, as at Fig. 8 (Kholodkevich et al.
2008; Kinebas et al. 2012) and the typical dHR
values are 100–150%. Considering these results,
it is possible to recommend the values of the
BioArgus-W setup parameters to be chosen by
the following algorithm:
1. Taking into account technological, regulation
and economic requirements, the maximum
operating time of station (i.e. a temporary
delay between the beginning of influence and
the system alarm – T o.max is chosen as well as
the optimal (or maximum) number of the
predicted false alarms per year. (Usually To =
10–20 min and T d = 10 min as a biosensor
response time is less, than 10 min and the
stress- induced HR high value is conserved
for more, then 10 min after the primary rise).
2. It was shown in Karmazinov et al. (2007) that
the values of the corresponding parameters
(L B , T o for SI and L B , T o , T d for dHR) which
correspond to the required number of false
alarms per year with the assumption:
T o < T o.max .
3. The minimum L B values and the maximum
values of parameters T o and T d have to be
138
S. V. Kholodkevich et al.