RT-87
|
A new bootstrap method for risk
assessment of exposure to Lyme disease
|
|
|
Cesare Furlanello, Stefano
Merler*, Claudio Chemini*.
*ITC-IRST, Trento, *Centro di Ecologia Alpina, Viote Mt. Bondone, Trento, Italy
|
|
Abstract
|
|
Background. Lyme borreliosis, now the
most common vector-borne illness in North America, may involve cardiac manifestations as
atrioventricular block, myopericarditis, and rhythm disturbances. The overall prognosis of
Lyme carditis is good, but temporary cardiac pacing may be required and late dilated
cardiomyopathy may occur. Borreliosis should be suspected in all patients with unexplained
cardiac symptoms (supraventricular tachyarrhythmia and especially in atrioventricular
block of unknown origin in young patients) who have been exposed in regions invaded by the
epidemic. Risk assessment of exposure to bites of infected ticks is thus needed for
prevention and accurate diagnosis of borreliosis.
Methods. We have adapted the new bootstrap .632+ error estimation
procedure to tree-based classifiers integrated on Geographical Information Systems (GIS).
In a first study in Trentino, Italian Alps, the new computational method has been used to
select predictive maps for the risk of tick bites based on data from 438 field samplings
and from thematic digital maps of elevation, type of vegetation and soil, exposure, and
density of roe deer (a key host for adult ticks).
Results. A tree-based model was developed that predicts local risk of
tick presence over cells of size 50 x 50 meters. Overall accuracy of the .632+ model is
81% over positive samples (presence of ticks) and 65% over negative ones. On an additional
data base, high infestation was correctly predicted in 79% of cases. The model indicates
at higher risk habitat locations under 1020 m, limestone substrate soils, and with higher
density of roe deer.
Conclusions. This study shows that risk assessment of Lyme
borreliosis may be implemented on a landscape mesoscale. Besides allowing measures of
prevention, appropriate knowledge of the epidemiological setting of possible exposure to
risk may facilitate recognition of early disseminated disease or specify clinical history
of idiopathic heart failure.
|
Key Words
|
|
Cardiac arrhythmias - epidemiology
borreliosis, Lyme disease, cardiac arrhythmias, risk assessment, bootstrap method, OA
|
