Dataset Information

Explaining the Neural Network: A Case Study to Model the Incidence of Cervical Cancer

ABSTRACT: Neural networks are frequently applied to medical data. We describe how complex and imbalanced data can be modelled with simple but accurate neural networks that are transparent to the user. In the case of a data set on cervical cancer with 753 observations excluding, missing values, and 32 covariates, with a prevalence of 73 cases (9.69%), we explain how model selection can be applied to the Multi-Layer Perceptron (MLP) by deriving a representation using a General Additive Neural Network. The model achieves an AUROC of 0.621 CI [0.519,0.721] for predicting positive diagnosis with Schiller’s test. This is comparable with the performance obtained by a deep learning network with an AUROC of 0.667 [1]. Instead of using all covariates, the Partial Response Network (PRN) involves just 2 variables, namely the number of years on Hormonal Contraceptives and the number of years using IUD, in a fully explained model. This is consistent with an additive non-linear statistical approach, the Sparse Additive Model [2] which estimates non-linear components in a logistic regression classifier using the backfitting algorithm applied to an ANOVA functional expansion. This paper shows how the PRN, applied to a challenging classification task, can provide insights into the influential variables, in this case correlated with incidence of cervical cancer, so reducing the number of unnecessary variables to be collected for screening. It does so by exploiting the efficiency of sparse statistical models to select features from an ANOVA decomposition of the MLP, in the process deriving a fully interpretable model.

SUBMITTER: Lesot M

PROVIDER: S-EPMC7274310 | biostudies-literature | 2020 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundCervical cancer is a prevalent public health concern and is among the leading causes of death among women globally. Malawi has the second highest cervical cancer prevalence and burden in the world. Due to the cervical cancer burden, the Malawi government scaled up national cancer screening services in 2011, which are free for all women. This paper is the first study to examine the socioeconomic inequality in cervical cancer screening uptake using concentration indices, in Malawi. Furthermore, it decomposes the concentration index to examine how each factor contributes to the level of inequality in the uptake of cervical cancer screening.MethodsThe data used in this paper were obtained from the nationally representative Malawi Population HIV Impact Assessment (MPHIA) household survey, which was conducted in 2015. Concentration curves were constructed to explore whether there was any socioeconomic inequality in cervical cancer screening and, if so, its extent. This was complemented by concentration indices that were computed to quantify the magnitude of socioeconomic inequality. A decomposition analysis was then conducted to examine the factors that explained/were associated with greater socioeconomic inequality in cervical cancer screening. The methodology in this paper followed that of previous studies found in the literature and used the wealth index to measure socioeconomic status.ResultsThe results showed that the concentration curves lay above the line of equality, implying a pro-rich inequality in cervical cancer screening services. Confirming the results from the concentration curves, the overall concentration index was positive and significant (0.142; %95 CI = 0.127, 0.156; p < 0.01). The magnitude was lower in rural areas (0.075; %95CI = 0.059, 0.090; p < 0.01) than in urban areas (0.195; %95CI = 0.162, 0.228 p < 0.001). After undertaking a decomposition of the concentration index, we found that age, education, rural or urban location, and wealth status account for more than 95% of the socioeconomic inequality in cervical cancer uptake.ConclusionDespite the national scale-up of free cancer care at the point of use, cervical cancer screening uptake in Malawi remains pro rich. There is a need to implement parallel demand-side approaches to encourage uptake among poorer groups. These may include self-testing and mobile screening centres, among others.

Dataset Information

Explaining the Neural Network: A Case Study to Model the Incidence of Cervical Cancer

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets