Predicting Quality of Life for Breast Cancer Patients

Christos Raspoptsis; Eugenia Mylona; Konstadina Kourou; Georgios C. Manikis; Haridimos Kondylakis; Kostas Marias; Paula Poikonen-Saksela; Panagiotis Simos; Evangelos Karademas; Ketti Mazzocco; Ruth Pat-Horenczyk; Berta Sousa; Dimitris Fotiadis

Abstract

The diagnosis of breast cancer has a significant impact on a patient's quality of life. Several demographic and clinical factors have been reported to affect the quality of life of breast cancer patients. However, few studies have a sufficient sample size for multifactorial assays to be tested. In the present work, we explore a rich set of clinical, psychological, socio-demographic, and lifestyle data from a large multicenter study of breast cancer patients (n = 765), with the aim to predict their global quality of life (QoL) 18 months after the diagnosis and to identify possible QoL-related prognostic factors. For QoL prediction, a set of Machine Learning methods were explored, namely Random Forest (RF), Support Vector Machine (SVM), and K-Nearest Neighbors (KNN). Depending on the model used, prediction accuracy varied between 0.305 and 0.864. Across models, a largely common set of psychological characteristics (optimism, perceived ability to deal with trauma, resilience as a trait, ability to understand the disease), as well as subjective perceptions of personal functionality (physical, social, cognitive function), were identified as key prognostic factors of long-term quality of life after a breast cancer diagnosis. Clinical Relevance- Predicting QoL is critical for decision-making on cancer care. Early detection of protective and obstructive factors associated with patient well-being would help health professionals to tailor preventive psychological programs aimed at enhancing the ability of breast cancer patients to adapt effectively to the disease.

Keywords: nan

Links
[Full text PDF][Bibtex]