Imperial College London
Portrait Picture

ProfessorThomasChurcher

Faculty of Medicine, School of Public Health

Professor of Infectious Disease Dynamics
 
 
 
//

Contact

 

thomas.churcher

 
 
//

Location

 

G35Medical SchoolSt Mary's Campus

//

Summary

 

Publications

Citation

BibTex format

@unpublished{Esperança:2020:10.1101/2020.04.28.058495,
author = {Esperança, P and Da, D and Lambert, B and Dabiré, R and Churcher, T},
doi = {10.1101/2020.04.28.058495},
publisher = {bioRxiv},
title = {Functional data analysis techniques to improve the generalizability of near-infrared spectral data for monitoring mosquito populations},
url = {http://dx.doi.org/10.1101/2020.04.28.058495},
year = {2020}
}
Download

RIS format (EndNote, RefMan)

TY  - UNPB
AB  - Near infrared spectroscopy is increasingly being used as an economical method to monitormosquito vector populations in support of disease control. Despite this rise in popularity,strong geographical variation in spectra has proven an issue for generalising predictions fromone location to another.  Here, we use a functional data analysis approach—which mod-els spectra as smooth curves rather than as a discrete set of points—to develop a methodthat is robust to geographic heterogeneity. Specifically, we use a penalised generalised linearmodelling framework which includes efficient functional representation of spectra, spectralsmoothing and regularisation.  To ensure better generalisation of model predictions fromone training set to another, we use cross-validation procedures favouring smoother represen-tation of spectra. To illustrate the performance of our approach, we collected spectra forfield-caught specimens ofAnopheles gambiaecomplex mosquitoes – the most epidemiolog-ically important vector species on the planet – in two sites in Burkina Faso. Using thesespectra, we show how models trained on data from one site can successfully classify mor-phologically identical sibling species in another site, over 250km away. Whilst we apply ourframework to species prediction, our unified statistical framework can, alternatively, handleregression analysis (for example, to determine mosquito age) and other types of multinomialclassification (for example, to determine infection status).  To make our methods readilyavailable for field entomologists, we have created an open-source R packagemlevcm.  Alldata used is publicly also available.
AU  - Esperança,P
AU  - Da,D
AU  - Lambert,B
AU  - Dabiré,R
AU  - Churcher,T
DO  - 10.1101/2020.04.28.058495
PB  - bioRxiv
PY  - 2020///
TI  - Functional data analysis techniques to improve the generalizability of near-infrared spectral data for monitoring mosquito populations
UR  - http://dx.doi.org/10.1101/2020.04.28.058495
UR  - http://hdl.handle.net/10044/1/79923
ER  -
Download