Information Technology Reference
In-Depth Information
•
RSTAT - (F_11): Static resistance from inspiratory pause measured in
cm
O
/L/s;
•
Volume EXP - (F_12): Exhaled tidal volume in litters;
•
Volume INS - (F_13): Tidal volume settings in litters;
•
Volume Minute - (F_14): Exhaled minute volume litters;
•
Ventilation Mode - (F_15): Ventilation mode (CMV, SIMV and CPAP),
•
Process: It is the patient identification;
Table 1 presents a statistical analysis for each field collected from the ventilators. For
each one of the numeric values it was analysed the minimum and maximum value
collected, their average, standard deviation (stDev) and coefficient of variation.
Table 1.
Distribution of variables
Max
Min
Average
stDev
Variation
CDYN
200
0
32.30
21.97
68.02%
CSTAT
126
0
17.80
19.95
112.04%
FIO2
100
22
57.10
15.05
25.95%
Flow
80
0
37.91
15.96
42.10%
RR
29
0
12.21
8.71
71.34%
PEEP
10
3
5.92
1.51
25.58%
PMVA
32
0
12.41
2.83
22.84%
Peak Pressure
85
0
30.87
10.28
33.31%
RDYN
100
0
16.35
7.37
45.08%
RSTAT
51
0
10.40
11.04
106.12%
Volume EXP
1.98
0
0.48
0.12
25.80%
Volume INS
0.71
0
0.42
0.16
39.36%
Volume Minute
30.1
0
9.18
2.05
22.32%
Through Table 1 it is possible to verify the heterogeneity of each variable.
Additionally, two more variables (not quantitative) were considered: Plateau Pressure
(F_8) and the ventilation Mode (F_15). Plateau Pressure has only two possible classes
("<30
cm
O
" and ">=30
cm
O
"), distributed in a proportion of 85.92% for the
class"<30
cm
O
" and the remaining 14.02% for the class "> = 30
cm
O
".
Ventilation Mode (F_15) has three possible classes ("CMV", "SIMV", and “CPAP").
Approximately 65.14% of the records have the "CMV" result, 4.87% of the records
have the SIMV result and 29.94% of the records have the "CPAP" result.
3.3
Data Preparation
Since the present study is to make predictions using data mining techniques
(classification approach) it was necessary to make some transformations in the data
collected. First, some changes were made in the data source in order to transpose
values from the rows to the columns. Then, the records that had at least one null value
were removed. Finally, also were made changes to the target variable, because the
Search WWH ::
Custom Search