Conference Paper

Prediction of Poor Mental Health Following Breast Cancer Diagnosis Using Random Forests 1

Authors:
To read the full-text of this research, you can request a copy directly from the authors.

Abstract

Breast cancer diagnosis has been associated with poor mental health, with significant impairment of quality of life. In order to ensure support for successful adaptation to this illness, it is of paramount importance to identify the most prominent factors affecting well-being that allow for accurate prediction of mental health status across time. Here we exploit a rich set of clinical, psychological, socio-demographic and lifestyle data from a large multicentre study of patients recently diagnosed with breast cancer, in order to classify patients based on their mental health status and further identify potential predictors of such status. For this purpose, a supervised learning pipeline using cross-sectional data was implemented for the formulation of a classification scheme of mental health status 6 months after diagnosis. Model performance in terms of AUC ranged from 0.81± 0.04 to 0.90± 0.03. Several psychological variables, including initial levels of anxiety and depression, emerged as highly predictive of short-term mental health status of women diagnosed with breast cancer.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

... Options regarding the random initialization points and expectation-maximization were kept as default while the time-slice was set to 0.01. A detailed description of the implementation has been presented in a previous publication [14]. ...
Conference Paper
Being diagnosed with breast cancer (BC) can be a traumatic experience for patients who may experience symptoms of depression. In order to facilitate the prevention of such symptoms, it is crucial to understand how and why depressive symptoms emerge and evolve for each individual, from diagnosis through treatment and recovery. In the present work, data from a multicentric study of 706 BC patients followed for 12 months are analyzed. First, a trajectory-based unsupervised clustering based on K-means is performed to capture the dynamic patterns of change in patients' depressive symptoms after BC diagnosis and to identify distinct trajectory clusters. Then a supervised learning approach was employed to build a classification model of depression progression and to identify potential predictors. Patients were clustered into 4 groups: stable low, stable high, improving, and worsening depressive symptoms. In a nested cross-validation pipeline, the performance of the Support Vector Machine model for discriminating between "good" and "poor" progression was 0.78±0.05 in terms of AUC. Several psychological variables emerged as highly predictive of the evolution of depressive symptoms with the most important ones being negative affectivity and anxious preoccupation. Clinical Relevance-The findings of the present study may help clinicians tailor individualized psychological interventions aiming at alleviating the burden of these symptoms in women with breast cancer and improving their overall well-being.
Article
Mental disorders are major problems and influence the thought of suicide, so it needs to be diagnosed and treated earlier. Detecting the symptoms of mental disorders is highly significant as they are considered a life-threatening task. Hence, this paper plans to implement new mental depression recognition by analyzing the tweet media text to secure the lives of the affected individuals. In the standard dataset, the text tweet data is collected. The gathered text data are used in the bags of n-grams and glove embedding for extracting the text features. Here, the extracted features are considered for the weighted feature selection, where the weight optimization takes place using the Awareness Probability-based Crow Black Widow Search algorithm. Finally, the weighted features are used for the recognition phase using the Convolutional Neural Network-Ensemble learning, where the final layer of CNN is replaced by the “Support Vector Machine (SVM), Random Forest (RF), Long Short Term Memory (LSTM) and Artificial Neural Network (ANN)” for classification. Here, parameter optimization occurs in the classifiers to improve the evaluation of the designed method. The experimental outcomes show the efficient performance of the developed method when compared with the other baseline approaches.
Article
Full-text available
Background: The status of the data-driven management of cancer care as well as the challenges, opportunities, and recommendations aimed at accelerating the rate of progress in this field are topics of great interest. Two international workshops, one conducted in June 2019 in Cordoba, Spain, and one in October 2019 in Athens, Greece, were organized by four Horizon 2020 (H2020) European Union (EU)–funded projects: BOUNCE, CATCH ITN, DESIREE, and MyPal. The issues covered included patient engagement, knowledge and data-driven decision support systems, patient journey, rehabilitation, personalized diagnosis, trust, assessment of guidelines, and interoperability of information and communication technology (ICT) platforms. A series of recommendations was provided as the complex landscape of data-driven technical innovation in cancer care was portrayed. Objective: This study aims to provide information on the current state of the art of technology and data-driven innovations for the management of cancer care through the work of four EU H2020–funded projects. Methods: Two international workshops on ICT in the management of cancer care were held, and several topics were identified through discussion among the participants. A focus group was formulated after the second workshop, in which the status of technological and data-driven cancer management as well as the challenges, opportunities, and recommendations in this area were collected and analyzed. Results: Technical and data-driven innovations provide promising tools for the management of cancer care. However, several challenges must be successfully addressed, such as patient engagement, interoperability of ICT-based systems, knowledge management, and trust. This paper analyzes these challenges, which can be opportunities for further research and practical implementation and can provide practical recommendations for future work. Conclusions: Technology and data-driven innovations are becoming an integral part of cancer care management. In this process, specific challenges need to be addressed, such as increasing trust and engaging the whole stakeholder ecosystem, to fully benefit from these innovations.
Article
Full-text available
Purpose There is limited high-quality evidence on quality of life, anxiety, and depressive symptoms in breast cancer survivors and women with no history of cancer. We aimed to address this by comparing patient-reported outcomes between breast cancer survivors and women with no history of breast cancer. Methods Breast cancer survivors and women with no prior cancer were selected from the UK Clinical Practice Research Datalink GOLD primary care database, which includes population-based primary care electronic health record data. Breast cancer survivors and controls were frequency matched by age and primary care practice. Outcomes were assessed with validated instruments via postal questionnaire. Linear and logistic regression models were fitted to estimate adjusted associations between breast cancer survivorship and outcomes. Results A total of 356 breast cancer survivors (8.1 years post diagnosis) and 252 women with no prior cancer participated in the study. Compared with non-cancer controls, breast cancer survivors had poorer QoL in the domains of cognitive problems (adjusted β (aβ) = 1.4, p = 0.01), sexual function (aβ = 1.7, p = 0.02) and fatigue (aβ = 1.3, p = 0.01), but no difference in negative feelings, positive feelings, pain, or social avoidance. Breast cancer survivors had higher odds of borderline-probable anxiety (score ≥ 8) (adjusted OR = 1.47, 95%CI:1.15–1.87), but no differences in depression. Advanced stage at diagnosis and chemotherapy treatment were associated with poorer QoL. Conclusions Compared with women with no history of cancer, breast cancer survivors report more problems with cognition, sexual function, fatigue, and anxiety, particularly where their cancer was advanced and/or treated with chemotherapy. Implications for Cancer Survivors Breast cancer survivors with more advanced disease and/or treated with chemotherapy should be closely monitored and, when possible, offered evidence-based intervention for fatigue, cognitive dysfunction, and sexual problems.
Article
Full-text available
Background Breast cancer has distinct causes, prognoses, and outcomes and effects in patients at premenopausal and postmenopausal ages. We sought to assess the global burden and trends in breast cancer by menopausal status. Methods We did a population-based analysis of global breast cancer incidence and mortality among premenopausal and postmenopausal women. Menopausal status was defined using age as a proxy, whereby breast cancer cases or deaths at age 50 years or older were regarded as postmenopausal. Age-standardised breast cancer incidence and mortality in 2018 were calculated using GLOBOCAN data. Incidence trends for 1998–2012 were assessed in 44 populations from 41 countries using the Cancer in Five Continents plus database, by calculating the annual average percent change. Findings Approximately 645 000 premenopausal and 1·4 million postmenopausal breast cancer cases were diagnosed worldwide in 2018, with more than 130 000 and 490 000 deaths occurring in each menopausal group, respectively. Proportionally, countries with a low UNDP human development index (HDI) faced a greater burden of premenopausal breast cancer for both new cases and deaths compared with higher income countries. Countries with a very high HDI had the highest premenopausal and postmenopausal breast cancer incidence (30·6 and 253·6 cases per 100 000, respectively), whereas countries with low and medium HDI had the highest premenopausal and postmenopausal mortality, respectively (8·5 and 53·3 deaths per 100 000, respectively). When examining breast cancer trends, we noted significantly increasing age-standardised incidence rates (ASIRs) for premenopausal breast cancer in 20 of 44 populations and significantly increasing ASIRs for postmenopausal breast cancer in 24 of 44 populations. The growth exclusively at premenopausal ages largely occurred in high-income countries, whereas the increasing postmenopausal breast cancer burden was most notable in countries under transition. Interpretation We provide evidence of a rising burden of both premenopausal and postmenopausal breast cancer worldwide. Although early diagnosis and access to treatment remain crucial in low-income and middle-income countries, primary prevention efforts seeking to decrease exposure to known breast cancer risk factors are warranted in all world regions to curb the future breast cancer burden. Funding None.
Chapter
Full-text available
In the current study, a model-based system for predicting resilience in silico, as part of personalizing precision medicine, to better understand the needs for improved therapeutic protocols of each patient is proposed. The computational environment, which is currently under implementation within the BOUNCE EU project (“Predicting Effective Adaptation to Breast Cancer to Help Women to BOUNCE Back”), will help clinicians and care-givers to predict the patient’s resilience trajectory throughout cancer continuum. The overall proposed system architecture contributes to clinical outcomes and patient well-being by taking into consideration biological, social, environmental, occupational and lifestyle factors for resilience prediction in women with breast cancer. Supervised, semisupervised and unsupervised learning algorithms are adopted with the inherent ability to represent the time-varying behaviour of the underlying system which allows for a better representation of spatiotemporal input-output dependencies. The in silico resilience prediction approach accommodates numerous diverse factors contributing to multi-scale models of cancer, in order to better specify clinically useful aspects of recovery, treatment and intervention.
Article
Full-text available
Purpose: Unmanaged distress has been shown to adversely affect survival and quality of life in breast cancer survivors. Fortunately, distress can be managed and even prevented with appropriate evidence-based interventions. Therefore, the objective of this systematic review was to synthesize the published literature around predictors of distress in female breast cancer survivors to help guide targeted intervention to prevent distress. Methods: Relevant studies were located by searching MEDLINE, Embase, PsycINFO, and CINAHL databases. Significance and directionality of associations for commonly assessed candidate predictors (n ≥ 5) and predictors shown to be significant (p ≤ 0.05) by at least two studies were summarized descriptively. Predictors were evaluated based on the proportion of studies that showed a significant and positive association with the presence of distress. Results: Forty-two studies met the target criteria and were included in the review. Breast cancer and treatment-related predictors were more advanced cancer at diagnosis, treatment with chemotherapy, longer primary treatment duration, more recent transition into survivorship, and breast cancer recurrence. Manageable treatment-related symptoms associated with distress included menopausal/vasomotor symptoms, pain, fatigue, and sleep disturbance. Sociodemographic characteristics that increased the risk of distress were younger age, non-Caucasian ethnicity, being unmarried, and lower socioeconomic status. Comorbidities, history of mental health problems, and perceived functioning limitations were also associated. Modifiable predictors of distress were lower physical activity, lower social support, and cigarette smoking. Conclusions: This review established a set of evidence-based predictors that can be used to help identify women at higher risk of experiencing distress following completion of primary breast cancer treatment.
Article
Full-text available
Objectives: Breast cancer (BC) can be a traumatic and stressful experience for women but there are wide-ranging differences in the ways in which women respond and adapt to BC. This systematic review examines which sociodemographic, disease-related, and psychosocial factors near diagnosis predict later psychological adjustment to BC. Methods: Database searches were conducted in nine different health-related databases from 2000 to December 2015 using relevant search terms. Full-text, peer-reviewed articles in English that analyzed potential predictors of psychological adjustment in longitudinal studies were considered for inclusion. Results: Of 1780 abstracts 41 studies fulfilled inclusion criteria. Consistent sociodemographic and disease-related variables predictors of adjustment were income, fatigue, cancer stage, and physical functioning. Psychosocial factors, particularly optimism and trait-anxiety, as well as perceived social support, coping strategies, and initial levels of psychological functioning were found to be predictive of later depressive and anxiety symptoms, psychological distress, and quality of life for women with BC, in predictable ways. Other psychosocial variables, such as cognitive and body image factors, predicted psychological adjustment but were explored only by a few studies. Conclusions: The majority of studies showed a significant relationship between psychosocial factors and psychological adjustment. These results point to specific sociodemographic, disease-related, and psychosocial factors that can help to identify women at the time of diagnosis who are at risk for long-term psychological challenges so they can be referred for psychological support that targets their specific needs and can improve their quality of life and mood, and decrease indicators of anxiety, depression and psychological distress.
Article
Full-text available
Major depression is an important complication of cancer. However, reliable data are lacking for the prevalence of depression in patients with cancer in different primary sites, the association of depression with demographic and clinical variables within cancer groupings, and the proportion of depressed patients with cancer receiving potentially effective treatment for depression. We investigated these questions with data from a large representative clinical sample. We analysed data from patients with breast, lung, colorectal, genitourinary, or gynaecological cancer who had participated in routine screening for depression in cancer clinics in Scotland, UK between May 12, 2008, and Aug 24, 2011. Depression screening was done in two stages (first, Hospital Anxiety and Depression Scale; then, major depression section of the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition). Data for depression status were linked with demographic and clinical data obtained from the Scottish National Cancer Registry. We analysed data for 21 151 patients. The prevalence of major depression was highest in patients with lung cancer (13·1%, 95% CI 11·9-14·2%), followed by gynaecological cancer (10·9%, 9·8-12·1), breast cancer (9·3%, 8·7-10·0), colorectal cancer (7·0%, 6·1-8·0), and genitourinary cancer (5·6%, 4·5-6·7). Within these cancer groupings, a diagnosis of major depression was more likely in patients who were younger, had worse social deprivation scores, and, for lung cancer and colorectal cancer, female patients. 1130 (73%) of 1538 patients with depression and complete patient-reported treatment data were not receiving potentially effective treatment. Major depression is common in patients attending cancer clinics and most goes untreated. A pressing need exists to improve the management of major depression for patients attending specialist cancer services. Cancer Research UK and Chief Scientist Office of the Scottish Government.
Article
Full-text available
It is important to identify predictors of psychological health among breast cancer patients that can be relatively easily identified by medical care providers. This article investigates the role of one class of such potential predictors: easily identified demographics that have potential social and/or practical implications. Specifically, we examined whether income, marital status, presence of children in the home, education, travel distance, age and rurality interact with time to predict psychological health over the first year post diagnosis. Two hundred and twenty five breast cancer patients receiving radiation treatment completed four surveys over the course of 13 months that included measures of both their physical health and depressive symptoms. The results revealed that women who were not married had children living in the home or had to travel long distances to receive radiation treatment reported higher levels of depressive symptoms across the entire study. Women with lower incomes reported increased depressive symptoms, but only after the completion of treatment. Younger women reported elevated depressive symptoms during initial treatment, but this effect dissipated after the completion of treatment. The current results suggest that demographic patient characteristics may indeed be useful in identifying both when and for whom depressive symptoms are particularly likely to be problematic.
Article
Full-text available
An approach to the construction of classifiers from imbalanced datasets is described. A dataset is imbalanced if the classification categories are not approximately equally represented. Often real-world data sets are predominately composed of "normal" examples with only a small percentage of "abnormal" or "interesting" examples. It is also the case that the cost of misclassifying an abnormal (interesting) example as a normal example is often much higher than the cost of the reverse error. Under-sampling of the majority (normal) class has been proposed as a good means of increasing the sensitivity of a classifier to the minority class. This paper shows that a combination of our method of over-sampling the minority (abnormal) class and under-sampling the majority (normal) class can achieve better classifier performance (in ROC space) than only under-sampling the majority class. This paper also shows that a combination of our method of over-sampling the minority class and under-sampling the majority class can achieve better classifier performance (in ROC space) than varying the loss ratios in Ripper or class priors in Naive Bayes. Our method of over-sampling the minority class involves creating synthetic minority class examples. Experiments are performed using C4.5, Ripper and a Naive Bayes classifier. The method is evaluated using the area under the Receiver Operating Characteristic curve (AUC) and the ROC convex hull strategy.
Article
Full-text available
A system approach termed P4 medicine has recently been proposed in the field of oncology. This approach has been advanced as an extension of what is usually called 'personalized' or 'genomic medicine'. P4 medicine creates effective predictive, personalized, preventive and participatory models to treat patients. In order to give more relevance to the behavioural component that impinges on the way individuals act to prevent, cope and react to illnesses, how they decide between different therapeutic options and interact with physicians and adhere to treatment, we propose that P4 medicine should be transformed into P5 medicine. The fifth P represents the psycho-cognitive aspects to be considered in order to empower the patient, increase his/her quality of life and transform him/her from a passive recipient into an active decision-maker in the treatment process.
Article
Full-text available
The aim of this study was to determine optimal cutoff scores for the Hospital Anxiety and Depression Scale (HADS) when used in evaluating cancer patients in acute care. A total of 689 cancer patients were assessed during their first days of in-patient treatment, using the structured clinical interview for DSM and the HADS. Statistical analysis was performed using ROC curves. A total of 222 patients (32%) had a mental disorder. The area under the curve was the best in the total scale of the HADS, namely 0.73. With a score of > or =13, it is possible to detect 76% of the cases with a specificity of .60, whereas 95% of the cases can be detected with a score of > or =6 (specificity 0.21). With scores of > or =16 and > or =22, recommended by the test authors for primary care, only 59 and 30% of the comorbid cancer patients are indicated. Lower HADS cutoff scores when preferable when evaluating cancer patients than are recommended for use in primary care. When using HADS in clinical practice and epidemiological studies, it is important to decide whether, for the task at hand, high detection rates of affected patients or low misclassification rates are more important.
Article
Displaying resilience following a diagnosis of breast cancer is crucial for successful adaptation to illness, well-being, and health outcomes. Several theoretical and computational models have been proposed toward understanding the complex process of illness adaptation, involving a large variety of patient sociodemographic, lifestyle, medical, and psychological characteristics. To date, conventional multivariate statistical methods have been used extensively to model resilience. In the present work we describe a computational pipeline designed to identify the most prominent predictors of mental health outcomes following breast cancer diagnosis. A machine learning framework was developed and tested on the baseline data (recorded immediately post diagnosis) from an ongoing prospective, multinational study. This fully annotated dataset includes socio-demographic, lifestyle, medical and self-reported psychological characteristics of women recently diagnosed with breast cancer (N = 609). Nine different feature selection and cross-validated classification schemes were compared on their performance in classifying patients into low vs high depression symptom severity. Best-performing approaches involved a meta-estimator combined with a Support Vector Machines (SVMs) classification algorithm, exhibiting balanced accuracy of 0.825, and a fair balance between sensitivity (90%) and specificity (74%). These models consistently identified a set of psychological traits (optimism, perceived ability to cope with trauma, resilience as trait, ability to comprehend the illness), and subjective perceptions of personal functionality (physical, social, cognitive) as key factors accounting for concurrent depression symptoms. A comprehensive supervised learning pipeline is proposed for the identification of predictors of depression symptoms which could severely impede adaptation to illness.
Article
Objectives: Most research addressing needs and concerns of young patients with breast cancer (≤40 years) is retrospective. The HOHO European protocol is a prospective multicenter cohort study of young women with newly diagnosed breast cancer, about fertility, psychosocial and quality of life concerns. Here we report the baseline data and focus on predictors of fertility concerns. Materials and methods: Patient surveys and medical record review were used. The baseline survey included sociodemographic, medical and treatment data as well as questions on fertility concerns and preservation strategies. Subscales from the CAncer Rehabilitation Evaluation System-Short Form (CARES-SF) were administered to measure specific quality of life aspects. Uni- and multivariable modeling were used to investigate predictors of greater fertility concern. Results: Among 297 eligible respondents, 67% discussed fertility issues before starting therapy, 64% were concerned about becoming infertile after treatment, and 15% decided not to follow prescribed therapies. Fifty-four percent of women wished future children before diagnosis; of these, 71% still desired biologic children afterwards. In multivariable analysis, not having children was the only patient characteristic significantly associated with fertility concerns at diagnosis. Twenty-seven percent used fertility preservation strategies. Women who received chemotherapy reported greater physical (p = 0.021) and sexual difficulties (p = 0.039) than women who did not. Women who were married or had a partner reported less psychosocial problems than single women (p = 0.039). Conclusions: Young women with newly diagnosed breast cancer have several concerns, including, but not limited to, fertility. The HOHO European study provides valuable information to develop targeted interventions.
Article
Social support is theoretically expected to be positively associated with posttraumatic growth (PTG) and subjective well-being, and stress is expected to be positively associated with PTG and negatively associated with subjective well-being among breast cancer (BC) survivors. However, empirical evidence is mixed, predominantly cross-sectional, and few studies have examined the unique effects of these predictors on positive changes in psychological experiences post cancer diagnosis and systemic treatment. This study examined both general and BC-specific social support and stress as predictors of change in PTG and subjective well-being among BC survivors. Women (N = 173, Mage = 55.40, SD = 10.99) who had recently finished treatment completed demographic and treatment measures at baseline (T1); general and cancer-specific social support and stress, PTG and subjective well-being at 3 months (T2); and PTG and subjective well-being again at 6 months (T3). Longitudinal predictors of change in PTG and subjective well-being were examined using hierarchical multiple regression. The BC-specific social support (β = .12) and stress (cancer worry; β = .10) predicted increasing levels of PTG. Improvements in subjective well-being were predicted by higher levels of general social support (β = .21) and lower levels of general stress (β = -.59). There are distinct predictors of change in PTG and subjective well-being among BC survivors, supporting the distinction between the trauma-specific process of PTG and well-being. Copyright © 2013 John Wiley & Sons, Ltd.
The convergence properties of a nearest neighbor rule that uses an editing procedure to reduce the number of preclassified samples and to improve the performance of the rule are developed. Editing of the preclassified samples using the three-nearest neighbor rule followed by classification using the single-nearest neighbor rule with the remaining preclassified samples appears to produce a decision procedure whose risk approaches the Bayes' risk quite closely in many problems with only a few preclassified samples. The asymptotic risk of the nearest neighbor rules and the nearest neighbor rules using edited preclassified samples is calculated for several problems.
Developing the BOUNCE Psychological Ontology
  • H Kondylakis
  • E Alekos
  • K Marias
  • M Tsiknakis
  • N Papadakis
H. Kondylakis, E. Alekos, K. Marias, M. Tsiknakis, and N. Papadakis, "Developing the BOUNCE Psychological Ontology," ISWC, 2020.
Resilience Indices for Breast Cancer Management
  • H Kondylakis
  • P Simos
  • E Karademas
  • K Marias
  • P Poikonen-Saksela
H. Kondylakis, P. Simos, E. Karademas, K. Marias, and P. Poikonen-Saksela, "Resilience Indices for Breast Cancer Management," IEEE Int. Conf. Biomed. Heal. Informatics, 2021.
Computational models for predicting resilience levels of women with breast cancer
  • K Kourou
K. Kourou et al., "Computational models for predicting resilience levels of women with breast cancer," in IFMBE Proceedings, Sep. 2020, vol. 76, pp. 518-525.
Developing the BOUNCE Psychological Ontology
  • kondylakis
Resilience Indices for Breast Cancer Management
  • kondylakis