A Hybrid Deep Learning-Based Feature Selection Approach for Supporting Early Detection of Long-Term Behavioral Outcomes in Cancer Survivors (Preprint)

Abstract

The number of cancer survivors is growing, and cancer survivors often suffer from long-term behavioral outcomes due to cancer treatments. There is a need for better computational methods to handle and predict cancer behavioral outcomes so physicians and healthcare providers can implement preventative treatments for cancer survivors.

The aim of this study is to create a new feature selection algorithm to improve the performance of machine learning classifiers to predict long-term behavioral outcomes in cancer survivors.

We devise a hybrid deep learning-based feature selection approach to support early detection of long-term behavioral outcomes in cancer survivors. Within a data-driven, clinical-domain guided framework to select the best set of features among cancer treatments, chronic health conditions, socio-environmental factors, we develop a two-stage feature selection algorithm, i.e., a multi-metric, majority-voting filter and a deep drop-out neural network, to dynamically and automatically select the best set of features for each behavioral outcome. We also conduct an experimental case study on existing study data with 102 survivors of acute lymphoblastic leukemia (ALL) (aged 15 to 39 years old at evaluation and > 5 years post-cancer diagnosis) who were treated in a public hospital of Hong Kong. Finally, we design and implement radial charts to illustrate the significance of the selected features on each behavioral outcome to support clinical professionals’ future treatment and diagnoses.

In this pilot study, we demonstrate that our approach outperforms the traditional statistical and computation methods, including linear and non-linear feature selectors, for the addressed top-priority behavioral outcomes. Our approach holistically has higher F1, precision, and recall scores compared to existing feature selection methods. The models select several significant clinical and socioenvironmental variables as risk factors associated with the development of behavioral problems in young survivors of ALL.

Our novel feature selection algorithm has potential to improve machine learning classifiers’ capability to predict long-term behavioral outcomes in cancer survivors.