Optimal Mapping of Cancers According to Their Properties and Sample Conditions

Authors: Jinfeng Wang*, , Mei-Po Kwan, Co-author, Yilan Liao, Co-auther, Ningxu Zhang, Co-auther, Tonglin Zhang, Co-auther, Chengdong Xu, Co-auther
Topics: Geographic Information Science and Systems, Geographic Information Science and Systems
Keywords: Optimal Mapping,GIS,Spatial analysis
Session Type: Paper
Day: 4/7/2019
Start / End Time: 2:00 PM / 3:40 PM
Room: Tyler, Marriott, Mezzanine Level
Presentation File: No File Uploaded

Background: Cancers’ maps present their burden distributions, imply their determinants, and identify target intervention and control areas. However, researchers often either have to face a large pool of models or simply have been used to one model to mapping various diseases. The accuracy of disease mapping is rarely studied with comparison to alternative mapping models. The problem becomes a barrier in precision medicine.
Methods and Findings:In a disease mapping process, information flows from the property a disease to a sample first, a mapping model next, and a map in the end. Each element in the trinity has many options. Distinct diseases have different statistical properties. A disease can be accurately mapped if and only if the mapping model’s assumption matches properties of the disease under sampling.
Conclusions: A new framework facilitates model choice to optimize disease mapping according to disease mapping trinities. If diseases are independent and identically distributed in a geospace, then sample means are used to estimate risks of the disease in specific areas. If diseases are associated with covariates, then covariate-based models are appropriate . If diseases are spatially autocorrelated, then spatial autocorrelation (SAC)-based models are appropriate . If diseases have spatially stratified heterogeneity (SSH), then SSH-based models are appropriate. If diseases have SACs and SSHs with samples small and biased to populations, then biased sentinel hospital based–area estimation (BSHADE) models are appropriate as they can remedy biases and produce BLUE maps with secondary information. The theory is demonstrated by empirical studies.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login