Identification

Name

Althouse Dengue Internet Search Model

Contact people
(1) Benjamin Althouse, balthouse@jhsph.edu
Created

2014-01-22 20:24:06+00:00

Modified

2015-12-04 16:32:06.663575+00:00

Overview

Types
(1) Disease Dynamics
Overview

This model shows the utility of using Google search terms to predict dengue incidence in Singapore and Bangkok. The models predict incidence with a correlation greater than 0.8.

Question

Can you accurately predict a rise in incidence of dengue infections through search queries in order to trigger interventions?

Purpose

Prediction

Scope

Disease and Location

Goals
(1) Baseline awareness
(2) Early warning
Decision support

No

Based on
Methods
(1) Machine learning/Support vector machine (SVM)
(2) Regression/Generalized Boosted Regression
(3) Regression/Negative binomial
(4) Statistics/Akaike information criterion (AIC)
Products
(1) Disease incidence
Documents
(1) Document Benjamin MA, Yng Ng Y, Cummings DAT: Prediction of Dengue incidence using search query surveilla… (2011)
Notes

Application

Diseases
(1) Dengue
Locations
(1) Location Bangkok
Status current
(2) Location Singapore
Status current

Utility and readiness

Readiness

Study

Organizations
(1) Organization Bill and Melinda Gates Foundation
Role Funding support
(2) Organization National Institutes of Health (NIH)
Role Funding support
Compared to model

Yes

Verification & validation notes

Validated with out-of-sample data. Mathematically verified.

Availability

Implementation language

R

Input

Input datasets
(1) Name Google Insights for Search
Category Internet Search Queries
Subject
URL http://www.google.com/insights/search/
Notes
(2) Name Singapore Ministry of Health
Category Established Databases
Subject
URL
Notes