Intelligent Adaptive Monitoring for Cardiac Surveillance - Computational Intelligence in Healthcare: Advanced Methodologies - page 325

1 chronicle bigeminy[]() {

2

event(p_wave[?p1], P1); ?p1 in {normal};

3 event(qrs[?w0], R1); ?r1 in {normal};

4 80<R1-P1<120;

5 event(qrs[?r1_1], R_1); ?r1_1 in {abnormal};

6 noevent(qrs[?], (R1+1, R1_1-1));

7 400<R1_1-R1<600;

8 event(p_wave[?p2], P2); ?p2 in {normal};

9

1000<P2-R1_1<2000;

10

event(qrs[?r2], R2); ?r2 in {normal};

11

80<R2-P1<120; 1200<R2-R1_1<2000;

12

event(qrs[?r2_1], R2_1); ?r2_1 in {abnormal};

13

noevent(qrs[?], (R2+1, R2_1-1));

14

400<R2_1-R2<600;

15

event(p_wave[?p3], P3);?p3 in {normal};

16

1000<P3-R2_1<2000;

17
18

event(diastole[?d1], D1);

19

event(systole[?s1], S1);

20

0<S1-D1<100; 0<S1-R1<100;

21

event(diastole[?d2], D2);

22

1000<D2-S1<2000;

23

event(systole[?s1], S1);

24

0<S2-D2<100; 0<S2-R2<100;

25

0 < end-start < 4000;

26

}

Fig. 5. The bigeminy chronicle model.

is associated the task that it can achieve, the objects (types of events) that it can deliver
and features describing the contexts in which it should ensure the best performance.
The pilot receives continuously three kinds of contextual information (cf. Fig. 6):

- the line context is mainly related to the signal quality. Noise type and level, the
event detection rate and the event distribution are used to estimate the signal quality,
- the arrhythmia context is related to the diagnosis state. The chronicle recognition
rate, the types of chronicles recognized so far and their distribution, the expected
patient state, etc. are used to estimate the diagnosis quality,
- the patient context is related to historical medical data associated to the patient, i.e.
disorders, drugs, etc.

From this contextual information the pilot has to decide whether the context has
changed notably and to determine whether a reconfiguration is needed. In this case, the
pilot must select the best algorithms and tune their parameters as well as choose the
right chronicle abstration level and the set of chronicles to be recognized.
To ensure a maximal flexibility and modularity, the pilot uses expert decision rules
to determine when and how to perform reconfigurations. Here follows two examples