The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008
persons working for public health cannot treat the bursting
events in time and effectively.
The thoughts and methods of collaboration have been
frequently applied to the fields of computer science, geography,
epidemiology and so on in recent years. This paper designed a
collaborative epidemical surveillance and response system,
which is based on internet and mobile network and under C/S
(Client/Server) mode, built up a whole work flow which is fit to
epidemical surveillance and detection, event investigation, and
emergency response, experimented in practical work, and
reached a satisfying result in precision and efficiency.
In the collaborative epidemical surveillance and response
system designed in this paper, the broadly applied spatial and
space-time scan statistic method (Kulldorff, 1997, 2001;
Kulldorff, et al, 2005) was adopted in the disease detection
module. This method analyzes the existed and current disease
data, finds out the disease abnormal clusters in space and time,
and finishes the epidemical space-time surveillance and
detection. Bayesian analysis method was used for the function
of assistant disease diagnosis in emergency response module.
This function could finish the intelligent disease diagnosis
according to the disease symptoms, patient characteristics, and
clinical analysis information, and provides a rapid and scientific
reference for the working people of event response. These
methods above have reached a nice effect in practical
experiments.
2. EPIDEMICAL SURVEILLANCE AND RESPONSE
2.1 Surveillance
judge to the bursting cases, and complete the intelligent disease
diagnosis on the basis of epidemical investigation, clinical
examination of patients and examination results of laboratory.
In addition, the response system should provide introduction of
infectious disease characteristics and direction of fieldwork.
Generally, epidemical response system could include the
following modules: intelligent disease diagnosis, fieldwork
treatment, response recommendation, and so on.
3. ALGORITHMS OF SURVEILLANCE AND
RESPONSE
3.1 Spatial and space-time scan statistics
Spatial and space-time scan statistics methods were proposed
by Kulldorff of Harvard Medical School in 1997 and 2001, and
are widely applied in the fields of medicine, biology and so on.
Log likelihood ratio (LLR) is used as the evaluation norm in
spatial scan statistics (Kulldorff, 1997). Null hypothesis (e.g.
the possibility of persons in the studied regions is identical) is
required before the spatial scan statistics. Firstly, a series of
scan clusters are generated. Generally, these clusters are circles
whose centre points are geographical centres of studied regions.
The radiuses of these scan circles vary from zero to a specified
maximum value. Secondly, LLR values of all scan circles are
calculated. Several (4 to 5) regions with the maximum LLR
values are chosen as the available hotspot regions of disease
bursting. Finally, Monte Carlo hypothesis testing is used for the
statistical significance evaluation of those available hotspot
regions. Regions which have passed the testing are the detected
hotspot regions.
The detailed contents of epidemical surveillance include data
collecting, data analysis, early detection, et al. The collecting
data comprise disease reports, symptom information, medical
cases, etc. Data analysis is classifying and analyzing the
surveillance data statistically using spatial analysis methods,
providing various ways (text, figure, electronic map, etc) for the
epidemical investigation and inquiry, and predicting the
probable hotspot regions of disease bursting. Early detection is
analyzing the statistical data, publishing the detection signals
automatically, and predicting the epidemical hotspot regions
and spreading by the means of statistical models.
GIS and information technique could provide strong help to
epidemical surveillance. Visualization information technique
can help researchers describe the complicated spatial and
attribute data as visual geographical maps, accomplish the
collaborative working mode of text and graph information, and
study the epidemiology visually and çonveniently. Furthermore,
GIS could provide various topological structures and visual
spatial analysis methods for epidemical researches, determine
the geographical characteristics of infectious diseases, analyze
the surveillance and investigation data comprehensively, and
furnish emergency response with helpful suggestions and
directions. Meanwhile, GIS could distinguish multiple data
types and structures, and help the collecting of disease data.
2.2 Response
Epidemical response is to determine the response methods and
resource managements after infectious diseases burst. The
response system can use the diagnosis algorithms, give a quick
Models used in spatial scan statistics include Bernoulli model,
Poisson model, Ordinal model, Exponential model, etc. For
instance, LLR value of Poisson model is calculated as the
following expression:
n z Y Z f n G ~ n z
p{Z)) [p(G)-p(Z)
№G))
n z YY f*G)-n z Y c ~ nz
p(Z)J lrfG)-rtZ)J
where Z is a 3-dimensional vector, including the coordinates of
the centre point and the radius of the scan circle, n z is the real
disease case amount of the scan circle region, /¿(Z) is the
population amount of the scan circle region, n G is the total
disease case amount of the studied regions, and ^(Q) is the
population amount of the studied regions.
The main model of space-time scan statistics is space-time
permutation model. The processing method used in space-time
scan statistics is similar to spatial scan statistics. The scan
cluster used in space-time permutation model is not circle but
cylinder. The height of scan cylinder stands for time value (e.g.
day amount), and the bottom surface of scan cylinder has the