Conference PaperPDF Available

Computational Modeling of Psychological Resilience Trajectories During Breast Cancer Treatment

Authors:
Computational modeling of psychological resilience
trajectories during breast cancer treatment
Georgios C. Manikis
Computational Biomedicine Laboratory
FORTH-ICS
Heraklion, Greece
gmanikis@ics.forth.gr
Haridimos Kondylakis
Computational Biomedicine Laboratory
FORTH-ICS
Heraklion, Greece
kondylak@ics.forth.gr
Konstantina Kourou
Institute of Molecular Biology and
Biotechnology
FORTH-IMBB
Ioannina, Greece
konstadina.kourou@gmail.com
Evangelos Karademas
Computational Biomedicine Laboratory
FORTH-ICS & University of Crete
Heraklion, Greece
karademas@uoc.gr
Paula Poikonen-Saksela
Helsinki University Hospital
Comprehensive Cancer Center
Helsinki, Finland
paula.poikonen-saksela@hus.fi
Kostas Marias
Computational Biomedicine Laboratory
FORTH-ICS & Hellenic Mediterranean
University
Heraklion, Greece
kmarias@gmail.com
Dimitrios G. Katehakis
Computational Biomedicine Laboratory
FORTH-ICS
Heraklion, Greece
katehaki@ics.forth.gr
Ruth Pat-Horenczyk
Hebrew University School of Social
Work, Jerusalem, Israel
ruth.pat-horenczyk@mail.huji.ac.il
Lefteris Koumakis
Computational Biomedicine Laboratory
FORTH-ICS
Heraklion, Greece
koumakis@ics.forth.gr
Dimitrios I. Fotiadis
Institute of Molecular Biology and
Biotechnology
FORTH-IMBB, Ioannina, Greece
dimitris.fotiadis30@gmail.com
Panagiotis Simos
Computational Biomedicine Laboratory
FORTH-ICS & University of Crete
Heraklion, Greece
akis.simos@gmail.com
Angelina Kouroubali
Computational Biomedicine Laboratory
FORTH-ICS
Heraklion, Greece
kouroump@ics.forth.gr
Manolis Tsiknakis
Computational Biomedicine Laboratory
FORTH-ICS & Hellenic Mediterranean
University
Heraklion, Greece
tsiknaki@ics.forth.gr
Abstract— Coping w ith breast cancer and its consequences
has now become a major socioeconomic challenge. The
BOUNCE EU H2020 project aims at building a quantitative
mathematical model of factors associated with optimal
adjustment capacity to cancer. This paper gives an overview of
the project targets and on the algorithmic methods focusing on
modeling the psychological resilience trajectories during breast
cancer treatment.
Keywords— Computational Modeling, Machine Learning,
Resilience, Cancer
I. INTRODUCTION
Breast cancer is the most common cancer in women
world-wide accounting for 28% of the total cancer cases in
the WHO European Region, while the incidence of breast
cancer in the eastern and southern European countries is still
rising1. Still, there is a very high survival rate among breast
cancer patients2. Therefore and in order to ensuring a better
and faster recovery and higher quality of life, coping with
breast cancer and its consequences becomes a major
socioeconomic challenge calling for novel strategies for
1 http://www.euro.who.int/en/health-topics/noncommunicable-
diseases/cancer/news/news/2012/2/early-detection-of-common-
cancers/breast-cancer
2 https://www.cancerresearchuk.org/health-professional/cancer-
statistics/statistics-by-cancer-type/breast-cancer/survival
understanding, predicting, increasing the resilience of
patients to all the stressful challenges and experiences [1],
[2] and providing appropriate recommendations [3], [4], [5].
In this context, an interdisciplinary consortium of experts
was formed in 2017 in response to a HORIZON 2020 call,
for Personalized Medicine research and innovation solutions,
in order to work towards predicting effective adaptation to
breast cancer to help women to BOUNCE Back. The broad
and general objective of the BOUNCE project3 is to build a
quantitative mathematical model of factors associated with
optimal adjustment capacity to cancer. Special emphasis is
given to modifiable factors associated with optimal disease
outcomes. At the core of the project is a prospective multi-
centre clinical pilot at four major oncology centres: European
Institute of Oncology in Italy, Helsinki University Hospital
in Finland, the Rabin Medical Center, Shaare Zedek Medical
Center coordinated by the Hebrew University of Jerusalem in
Israel and Champalimaud Foundation in Portugal. The study
is currently recruiting over 660 breast cancer patients with
stage I-III histologically confirmed diagnosis. There will be
seven assessment waves over a period of 18 months:
baseline, which will occur after the first visit to the
oncologist, Month 3, 6, 9, 12, 15, and 18.
3 https://www.bounce-project.eu
423
2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)
2471-7819/19/$31.00 ©2019 IEEE
DOI 10.1109/BIBE.2019.00082
Fig. 1. Schematic representation of trajectories of psychosocial and
functional illness outcomes during the course of illness. T1 represents
measurements at the time of cancer diagnosis. Within BOUNCE Tn
represents each subsequent outcome measurements up to 18 months after
diagnosis. Each line indicates one patient or subgroup of patients
displaying similar adaptation to illness. Additional stressors may emerge
at various time points that may affect the adaptation process in varying
degrees.
The primary end-point of the BOUNCE clinical pilot is
to identify factors and processes that may predict both
interim and long-term patient resilience, their physical
wellbeing and psychological outcomes of cancer and cancer
treatment. The design of the study will allow researchers to
apply statistical models to understand complex interactions
of variables both within and across measured domains (e.g.,
from personality traits to health outcomes, through health-
related beliefs and behaviour, with socio-demographic and
cultural variables as moderating conditions) in order to gain
an enhanced understanding of the dynamic process of
adaptation to breast cancer, and resilience-as-a-process.
The rest of the paper is structured as follows: In section
2, we define resilience whereas in Section 3 we present the
algorithms that are going to be eventually used for achieving
individualized models of resilience. Finally, Section 4
concludes this paper and presents the next steps to be
implemented until the end of the project.
II. DEFINING RESILIENCE
Being resilient does not mean that a person doesn't
experience difficulty or distress [6][7]. Emotional pain and
sadness are part of the common experience of people who
have suffered major difficult life events and trauma [8]. This
is also true for breast cancer patients for whom resilience is a
major asset for achieving better adaptation to illness and
higher quality of life [9][10].
Some understand resilience to be a predisposition or an
existing potential, before even facing an adverse situation
[11]. In this sense, resilience as a capacity or trait is the
integration of the internal and external resources (e.g.,
optimism, humor, social support) available to the individual
facing adversity which may influence/increase the
effectiveness of the coping process. Resilience-as-trait
measurement is subject to significant drawbacks, for instance
people understand themselves in different ways. However,
instead of looking at resilience through the “eyes” of the
patients--which is subject to report bias and it is also likely to
be affected by illness representations, coping strategies etc.--
we also look at resilience by modelling the person’s affective
and behavioural responses to the disease and to subsequent
negative events (i.e. stressors) (Fig. 1). In the present work
we focus on analysis techniques addressing the evolution of
resilience-as-process which is inferred from the observation
of positive adaptation to illness and better outcomes, despite
any negative events [11], such as initial diagnosis,
subsequent therapy side-effects, negative test results etc. The
outcomes at all time points T1-Tn (see Fig. 1) could be on a
single dimension (i.e. Quality of Life (QoL) and mental
health/affective state and functionality) or complementary
outcomes considered separately (i.e. QoL, mental
health/affective state, functionality and physical health).
III. USING ADAPTIVE ALGORITHMS TO ACHIEVE
INDIVIDUALIZED MODELS OF RESILIENCE
Modelling of psychological resilience requires building
theoretically plausible, clinically useful, and computationally
sound schemes describing: (i) the predominant mechanisms
involved in the process of psychological adaptation to cancer
and (ii) the most powerful longitudinal predictors of long-
term psychosocial and functional outcomes following
treatment for breast cancer. BOUNCE aspires to go further
than conventional multivariate statistical methods permit and
develop a prediction tool that can be used at any point during
breast cancer diagnosis and treatment to identify patients at
risk for poor psychosocial and functional outcomes—i.e.,
patients who, at a given point in time, demonstrate poor
psychological resilience. Given the inherent complexity of
the longitudinal data, BOUNCE will develop and evaluate a
Machine Learning (ML) framework to identify subgroups of
patients that display distinct psychosocial profiles (at specific
time points and over time) in adapting to breast cancer.
Firstly, patient clustering is applied to group patients at a
given time point according to a specific ‘level’ of adaptation
to illness. In this approach resilience is defined according to
the observation of affective and functional status. Secondly,
longitudinal data are exploited through a clustering
methodology aiming to distinguish patient profiles according
to possible transitions from one resilience category to
another due to changes in specific factors. Finally, all clinical
predictive outcomes will be entered into a decision-level
fusion model to investigate whether the ensemble of the
decisions further improves prediction of resilience at a
specific time point. The BOUNCE trajectory predictor will
exploit effectively factors measured in the multicenter
clinical study. This set of factors consists of: (i) patient-
reported outcomes (i.e. mental health, distress level, health-
and overall QoL, and functionality), (ii) illness-related self-
regulation variables (i.e. self-rated health etc.), (iii)
potentially stressful events taking place during the follow up
period, (iv) moderators and facilitators (i.e. self-efficacy,
resilience, social support etc.) and (v) lifestyle factors (i.e.
health habits etc.).
All aforementioned data are collected using standardized
questionnaires and are available through the Noona4 tool.
Then data are exported in batches and are continuously
integrated using a novel architecture [12] and data
infrastructure [13] combining a data lake for staging the
available information and an ontology for the integration and
their harmonization and a research supporting tool for
facilitating data exploration and visualization [14], [15].
A. Unsupervised Machine Learning Techniques
BOUNCE unsupervised learning analysis framework will
be coupled with machine learning and conventional
4 http://www.noona.com/
424
statistical approaches with the aim to group individual
trajectories into distinct clusters and reveal intra-individual
changes and inter-individual differences in the examined
patients. To this extend, clustering techniques will be also
employed to analyze data under a cross-sectional perspective
using corresponding data acquired from distinct time points
during the critical 18-month period following cancer
diagnosis.
Within the BOUNCE cross sectional unsupervised
learning framework, symptom clusters will be generated
based on current mental health and illness-related distress,
QoL, and functional level. Cut-off thresholds will be set to
define discrete levels of each condition (e.g. high, moderate,
mild and low level of each condition) and clustering
performance will be validated both externally, using current
patients’ condition, and internally based on validity indices
such as Silhouette, Dunn, etc. Descriptive statistics and
frequency distributions will assess potential statistical
differences between the clinical, psychosocial, and
behavioural characteristics grouped into the distinct clusters.
Illustrative representation of the reported clusters and
distributions of all examined data will be provided via
clustergrams, radar plots, cluster-change membership, and
alluvial diagrams. Robust and sparse k-means clustering
(RSKC) [12] will be the model of choice since it is able to: a)
deal with missing data and outliers, and b) assign weights to
the examined data as they might have varying effects on
clustering and not contribute equally in determining the
clusters.
The aim of longitudinal clustering is to establish patient
profiles across time-points using unsupervised learning
(clustering) techniques. Cluster-change profile of each
person over time will be also determined revealing the
patient's’ potential shifting from one adaptation/resilience
category to another due to changes in medical and/or
psychological and behavioural factors. Within this analysis
framework persons who belong to cluster A at T1, Cluster B
at T2 and Cluster A again at T3 will be grouped together.
Determining the shifting from one category to another during
the trajectory of the disease will enable the estimation of the
exact number of patients that belong to each category.
Moreover, an accurate estimation for the possibility of
shifting from low to high resilience and vice versa across
different time points will be achieved. The overall process of
adaptation to illness and the resilience level will be grouped
into certain clusters of patients’ characteristics and
behaviour; thus, empowering the prediction of final
outcomes according to the categories that are most
informative across time. Latent growth curve modelling
(LGCM) and growth mixture modelling (GMM) have
frequently been employed to handle data on disease progress
of oncological patients after treatment, showing interesting
results in trajectory clustering and identifying behavioural
risk factors as predictors of trajectory groups [17]. A
machine learning based resilience clustering approach will be
also applied providing distinct clusters not only on the basis
of individual trajectory similarities across time but on the
trajectory shape as well. K-Means for longitudinal data using
shape-respecting distance has recently been investigated
demonstrating higher performance when compared to
traditional longitudinal clustering techniques [18].
Within this analysis, the factors and/or their interactions
that can accurately predict final and intermediate outcomes
will be identified. The medical and psychological/
behavioural factors that will be considered in the supervised
analysis framework will be assessed at: i) previous time
interval, ii) at baseline and previous time interval and iii)
across different time intervals within the 18-month follow up
period. Towards this direction, the development of the final
BOUNCE predictive tool will be achieved by exploiting
factors over time.
B. Supervised Machine Learning Techniques
Supervised machine learning and conventional statistical
methods will be adopted to further exploit and model the
heterogeneous multiscale data within BOUNCE both
longitudinally and at single time intervals. This type of
analysis will enable the prediction of intermediate and final
outcomes related to illness adaptation and resilience. The
different patients’ profiles in terms of the scores in the
medical and psychological/behavioural variables assessed
during the follow-up period will be considered for prediction
purposes (Fig.2).
More specifically, concerning the cross-sectional studies
within the BOUNCE supervised learning analysis
framework, several predictive models (extreme gradient
boosting-XGBoost, generalized linear models-GLM, random
forests-RF, weighted random support vector machine
clusters analysis-WRSVMC, etc.) will be trained, validated
and tested under a nested cross-validation schema and model
comparison will be reported using exactly the same input
data during all iterations. Feature ranking/selection
mechanisms will be implemented during model training. A
probabilistic outcome will be provided related to resilience
status and all medical, psychological, and behavioral factors
will be ranked according to their importance in the predictive
performance. Model performance will be quantified using
several statistical metrics including accuracy, sensitivity or
recall, specificity, precision, F1-score and area under the
curve (AUC).
BOUNCE longitudinal analysis will include, among
other models, a novel semi-parametric marginal approach
(boosted multivariate trees for longitudinal data-boostmtree)
as reported in [15] to identify all related interactions between
BOUNCE medical, psychological, and behavioural factors
and time semi-nonparametrically. The most important factors
and factor-time interactions will be identified using
permutation variable importance techniques. Growth-based
trajectory modelling will be used to classify patients
according to their adaptation/resilience level at final (i.e., at
18 months) and intermediate outcomes (i.e., at 6, 12…
months). Several regression models will fit BOUNCE
longitudinal data simultaneously and patient-specific
probability of group membership will be assigned.
Additionally, group-based trajectories will be estimated for
each group of patients over time and goodness of fit accuracy
will be assessed using C-statistics and Bayesian information
criterion (BIC).
C. Design of the Predictor Model Aggregation
Within BOUNCE, the prediction tool includes a limited
number of biomedical factors and self-ratings, that will
emerge (and be validated) as significant predictors of
outcomes in addition to anxiety, depression, and distress
which are measured early in the course of the disease.
Focusing on end-point outcomes instead of trajectories the
predictive outcome will be continuous or categorical. To this
425
Fig. 2. BOUNCE cross-sectional predictive modelling framework for
resilience status.
end, good outcomes at Tn regardless of prior status imply
indirectly high resilience, whereas poor outcomes imply low
resilience as has been defined in BOUNCE.
The fusion model within BOUNCE is implemented by
utilizing ensemble methods which combine the predictions
of several base estimators. The base estimators are built with
given learning algorithms, such as generalized linear models,
random forests, extreme gradient boosting, support vector
machines-SVM, etc., in order to improve the performance
compared to a single estimator. Voting classifier [20] is
applied for combining conceptually different machine
learning classifiers and use a majority vote (“hard” vote) or
the average predicted probabilities (“soft” vote) to predict the
class labels. Such a classifier can be useful for a set of
equally well performing supervised or unsupervised models
in order to balance out their individual weaknesses. With the
ensemble vote classifier different training sets are considered
for building the predictive models (Fig.3). As mentioned
above, different classification algorithms are adopted for
predicting the class label of new samples. Voting classifier
will be applied for making the final and more robust
prediction of the end-point outcomes.
Fig. 3. The process followed by an ensemble voting classifier for
combining the different predictions and make the final prediction in terms
of voting. Different training sets are considered for each classifier in order
to build the predictive models before their fusion.
In sum, BOUNCE develops and deploys advanced
computational tools to validate indices of patients’ capacity
to bounce back during the highly stressful treatment and
recovery period following diagnosis of breast cancer.
Elements of a dynamic, predictive model of patient outcomes
are incorporated in building a decision-support system to be
used in routine clinical practice providing oncologists and
other health professionals with concrete, personalized
recommendations regarding optimal psychosocial support
strategies.
IV. CONCLUSIONS
This paper presents an overview of the BOUNCE project
for developing an individualized model of resilience. This
model will be constructed using unsupervised and supervised
machine learning techniques that will be combined through
model aggregation techniques, in order to gain an enhanced
understanding of the dynamic process of adaptation to breast
cancer. Currently the project is collecting data, whereas the
data infrastructure and the research support tool for data
visualization and exploration are already available. The next
step is the implementation of the model execution engine and
the identification of the specific models to be used.
ACKNOWLEDGMENT
Work presented in the paper is part of the BOUNCE project
and has received funding from the European Union’s
Horizon 2020 research and innovation programme under
grant agreement No 777167. Any opinions, results,
conclusions, and recommendations expressed in this
material are those of the authors and do not necessarily
reflect the views of BOUNCE or the European Commission.
REFERENCES
[1] G. Ristevska-Dimitrovska, I. Filov, D. Rajchanovska, P. Stefanovski,
B. Dejanova, “Resilience and Quality of Lif e in Breast Cancer
Patients,” Open Access Maced. J. Med. Sci. 3, pp.727–731, 2015.
[2] H. Kondylakis, et al. "iManageCancer: Developing a platform for
Empowering patients and strengthening self-management in cancer
diseases." 2017 IEEE 30th International Symposium on Computer-
Based Medical Systems (CBMS). IEEE, pp. 755-760, 2017.
[3] H. Kondylakis, et al. "Semantically-enabled Personal Medical
Information Recommender." International Semantic Web Conference
(Posters & Demos). 2015.
[4] M. Stratigi, H. Kondylakis, K. Stefanidis. "Fairness in group
recommendations in the health domain." 2017 IEEE 33rd
International Conference on Data Engineering (ICDE). IEEE, 2017.
[5] H. Kondylakis, et al. "Patient Empowerment through Personal
Medical Recommendations." MedInfo 216, pp. 1117, 2015.
[6] American Psychological Association. “The Road to Resilience”.
Retrieved from http://www.apa.org/helpcenter/road-resilience.aspx,
visited August 2019.
[7] A.S. Masten, “Ordinary magic: Resilience processes in development,”
American Psychologist, 56, pp. 227–238, 2001.
[8] G.A. Bonanno, “Loss, trauma, and human resi lience: have we
underestimated th e human capacity to thrive after extremely aversive
events?” American psychologist, 59(1), pp. 20, 2004.
[9] B. Izydorczyk, A. Kwapniewska, S. Lizinczyk, K. Sitnik-Warchulska,
“Psychological Resilience as a Protective Factor for the Body Image
in Post-Mastectomy Women with Breast Cancer,” International
Journal of Environmental Research and Public Health, 15, pp. 1181,
2018.
[10] S.E. Markovitz, W. Schrooten, A. Arntz, M.L. Peters, “Resilience as a
predictor for emotional response to the diagnosis and surgery
in breast cancer patients,” Psycho-Oncology, 24, pp. 1639-1645,
2015.
[11] S.M. Southwick, G.A. Bonanno, A.S. Masten, C. Panter-Brick, R.
Yehuda, “Resilience definitions, theory, and challenges:
Interdisciplinary perspectives,”. European journal of
Psychotraumatology, 5, 25338, 2014.
[12] K. Kourou, H. Kondylakis, L. Koumakis, G.C. Manikis, K. Marias,
M. Tsiknakis, P.G. Simos, E. Karademas, D.I. Fotiadis, “A Reference
Architecture for Predicting Resilience Levels of Women with Breast
426
Cancer,” IEEE International Conference on Biomedical and Health
Informatics (BHI'19) , 2019.
[13] H. Kondylakis, L. Koumakis, D. Katehakis, A. Kouroubali, K.
Marias, M. Tsiknakis, P. Simos, E. Karademas, “Developing a Data
Infrastructure for Enabling Breast Cancer Women to BOUNCE
Back”, IEEE CBMS, pp. 652-657, 2019.
[14] K. Kourou, H. Kondylakis, L. Koumakis, G.C. Manikis, K. Marias,
M. Tsiknakis, P.G. Simos, E. Karademas, D.I. Fotiadis,
“Computational Models for Predicting Resilience Levels of Women
with Breast Cancer,” MEDICON, 2019.
[15] Koumakis, L., Kondylakis, H., Chatzimina, M., Iatraki, G.,
Argyropaidas, P., Kazantzaki, E., Tsiknakis, M., Kiefer, S. and
Marias, K., “Designing smart analytical data services for a personal
health framework”. pHealth, 224, pp.123-128, 2016.
[16] Y. Kondo, M. Salibian-Barrera and R. Zamar. “RSKC: An R Package
for a Robust and Sparse K-Means Clustering Algorithm”, Journal of
Statistical Software, 72(5), 1 – 26, 2016.
[17] Y. Yang, J. Cameron, C. Bedi, and G. Humphris, "Fear of cancer
recurrence trajectory during radiation treatment and follow-up into
survivorship of patients with breast cancer," BMC cancer, vol. 18, p.
1002, 2018.
[18] Genolini, C., Ecochard, R., Benghezal, M., Driss, T., Andrieu, S., &
Subtil, F., “kmlShape: an efficient method to cluster longitudinal data
(time-series) according to their shapes”. Plos one, 11(6), e0150738,
2016.
[19] A. Pande, L. Li, J. Rajeswaran, J. Ehrlinger, U. B. Kogalur, E. H.
Blackstone, et al., “Boosted multivariate trees for longitudinal data.
Machine learning”, vol. 106, pp. 277-305, 2017
[20] D. Ruta and B., “Gabrys: Classifier selection for majority voting.
Information fusion”, vol. 6, pp. 63-81, 2005
427
... In the era of Big Data, Machine Learning (ML) algorithms have emerged as an appealing alternative to conventional statistical approaches for improving prediction accuracy, thanks to their ability to efficiently handle a large amount of heterogeneous data and complex interactions [10,11]. Random Forests (RF), in particular, is efficient in handling highly nonlinear data and a large number of features, agile in terms of noise in data, and simpler to tune. ...
Conference Paper
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.
... The patients' ability to build resilience to emotional and physical pains of cancer and its' treatment is identi ed as one of the key assets to surviving the disease and improving quality of life [21]. Resilience factor exists for every individual, however, across the life course, the experience of resilience will vary [22] and those that can cope and adapt during setbacks and adversities are labeled resilient, while those that are not able to adapt to the challenging life events fade off, and succumb to adversities [23]. In Ghana little is known about factors that in uence resilience among BC women who received chemotherapy for breast cancer. ...
Preprint
Full-text available
Background: Chemotherapy is the most common treatment option for breast cancer in Ghana. Chemotherapy brings about unpleasant and traumatic experiences due to the effects of drugs on patients. Despite contexts specific variations, recent literature show improvement in breast cancer survival rates globally. Methods: This study aimed to explore factors that influenced resilience among women who completed chemotherapy for breast cancer in the Greater Accra Metropolis of Ghana. A qualitative exploratory descriptive design was adopted to study fifteen breast cancer survivors. A purposive sampling technique was used to select the participants. Individual in-depth interviews were conducted in English using a semi-structured interview guide. Data collection and analysis occurred concurrently. All interviews were audio-recorded with consent from the participants. Using a thematic and content analysis approach, all interviews were transcribed verbatim, coded and themes and subthemes generated. Results: Two major themes and eight subthemes were generated. Thus, 1. The Burden associated with breast cancer diagnosis and treatment; with subthemes; Physical, Psychological, Social, Financial burdens. 2. Personal traits; and the subthemes; Hope, Optimism, Self-esteem, Confidence. Conclusions: The study concluded that chemotherapy is associated with biopsychosocial pain and economic burden. In the absence of organized support systems, surviving breast cancer treatment is influenced by individuals’ personality and emotional orientation to the disease.
Article
This paper analyzes the clinical and technological characteristics of cancer mobile apps that enable patients to securely record, manage, and share their information. It proposes an integrated approach towards developing quality mobile health apps for cancer and discusses issues relevant to the enhancement of patient experience and acceptance, adherence to treatment, and effective support of coordinated care. Regulation, standardization, and interoperability together with the existence of useful, accurate, and reliable tools for active patient engagement are important aids towards efficient cancer disease management. Improving quality of life and well-being, in a secure and safe manner that respects the patients' privacy, seems to be a key challenge, together with the exploitation of the outcomes of recent research projects, relevant to end user digital engagement towards an integrated approach for the introduction of trustworthy, interoperable, usable, adaptable, and quality mobile health apps.
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
Background: Fear of cancer recurrence (FCR) has been shown to be higher in patients treated with external beam radiotherapy (RT) compared to those untreated. However, little is known about the dynamics of patient's FCR during and after RT. The aim of this study was to examine FCR levels in a longitudinal panel design with breast cancer patients receiving RT. Methods: Consecutive newly-diagnosed breast cancer patients (n = 94) attending a single cancer centre were invited to complete a 7-item FCR scale (FCR7) that was collected weekly by paper instrument and at a follow-up phone call 6-8 weeks after completion of RT. Descriptive statistics, and Latent Growth Curve Modelling (LGCM) were utilised to analyse the data. Results: Women who were younger, single/separated, had chemotherapy, had extra boost radiation treatment, taking Herceptin and treated by 4-field technique reported higher recurrence fear at baseline. There was strong evidence of substantial variation in the trajectory of FCR (z = - 3.54, p < .0001). The average trajectory of FCR over RT was negative (unstandardized estimate = - 0.59) and associated with FCR follow-up level (standardised estimate = 0.36, z = 3.05, p < .002), independent of baseline recurrence fears. Conclusion: Patients vary in their trajectory of recurrence fears over RT which predicts FCR approximately 2 months following treatment. Review appointments by therapy radiographers presents an opportunity to intervene in FCR trajectories. Trial registration: ClinicalTrials.gov: NCT02599506 . Prospectively registered on 11th March 2015.
Article
Full-text available
European statistics confirm a rise in breast cancer among contemporary women. Those suffering from cancer and undergoing a surgery (mastectomy) are undoubtedly considered to be in difficult situations. The range of the numerous negative and/or positive emotions, thoughts, and behaviours depend on many psychological factors such as psychological resilience. The authors are currently drawing a report on their own studies where they are trying to determine factors that protect body image resilience in women suffering from breast cancer after mastectomies. The research group consisted of 120 women after a short (up to 2 years) or a long (over 2 years) duration having elapsed since their mastectomy. The results of the research groups show that psychological resilience is a significant protecting factor for the body image that prevents the excessive development of negative self-esteem in post-mastectomy women. Female patients ought to be provided aid in the short time immediately after the procedure and afterwards, when they are less capable of tolerating negative emotions. In order to significantly improve the general body image resilience to emotional and cognitive distortions in post-mastectomy women who experienced breast cancer, it is recommended that psychological interventions (from psychoeducation to psychological assistance and specialist psychotherapy) are conducted systematically throughout the course of treatment.
Article
Full-text available
Information in the healthcare domain and in particular personal health record information is heterogeneous by nature. Clinical, lifestyle, environmental data and personal preferences are stored and managed within such platforms. As a result, significant information from such diverse data is difficult to be delivered, especially to non-IT users like patients, physicians or managers. Another issue related to the management and analysis is the volume, which increases more and more making the need for efficient data visualization and analysis methods mandatory. The objective of this work is to present the architectural design for seamless integration and intelligent analysis of distributed and heterogeneous clinical information in the PHR context, as a result of a requirements elicitation process in iManageCancer project. This systemic approach aims to assist health-care professionals to orient themselves in the disperse information space and enhance their decision-making capabilities, to encourage patients to have an active role by managing their health information and interacting with health-care professionals.
Article
Full-text available
Machine learning methods provide a powerful approach for analyzing longitudinal data in which repeated measurements are observed for a subject over time. We boost multivariate trees to fit a novel flexible semi-nonparametric marginal model for longitudinal data. In this model, features are assumed to be nonparametric, while feature-time interactions are modeled semi-nonparametrically utilizing P-splines with estimated smoothing parameter. In order to avoid overfitting, we describe a relatively simple in sample cross-validation method which can be used to estimate the optimal boosting iteration and which has the surprising added benefit of stabilizing certain parameter estimates. Our new multivariate tree boosting method is shown to be highly flexible, robust to covariance misspecification and unbalanced designs, and resistant to overfitting in high dimensions. Feature selection can be used to identify important features and feature-time interactions. An application to longitudinal data of forced 1-second lung expiratory volume (FEV1) for lung transplant patients identifies an important feature-time interaction and illustrates the ease with which our method can find complex relationships in longitudinal data.
Article
Full-text available
Witten and Tibshirani (2010) proposed an algorithim to simultaneously find clusters and select clustering variables, called sparse K-means (SK-means). SK-means is particularly useful when the dataset has a large fraction of noise variables (that is, variables without useful information to separate the clusters). SK-means works very well on clean and complete data but cannot handle outliers nor missing data. To remedy these problems we introduce a new robust and sparse K-means clustering algorithm implemented in the R package RSKC. We demonstrate the use of our package on four datasets. We also conduct a Monte Carlo study to compare the performances of RSK-means and SK-means regarding the selection of important variables and identification of clusters. Our simulation study shows that RSK-means performs well on clean data and better than SK-means and other competitors on outlier-contaminated data.
Article
Full-text available
Background: Many studies have shown that a relationship exists between quality of life (QoL) and resilience in breast cancer patients, but few studies present information on the nature of this relationship of resilience on QoL. Our aim was to examine the relationship between resilience and quality of life in breast cancer patients. Methods: QoL was measured in 218 consequent breast cancer patients, with EORTC - QLQ Core 30 questionnaire, and EORTC QLQ-BR23. The resilience was measured with Connor Davidson Resilience Scale. Results: The global quality of life was positively correlated with the levels of resilience (R = 0.39 p < 0.001). All functional scales (physical, role, emotional, cognitive and social functioning was in a positive correlation with resilience. The symptoms severity (fatigue, nausea and vomitus, pain, dyspnea, insomnia, appetite loss, constipation, diarrhea, financial difficulties) was in negative correlation with resilience. Less resilient breast cancer patients reported worse body image and future perspective and suffered from more severe adverse effects of systemic therapy, and arm/breast symptoms. Conclusion: Our findings show that psychological resilience affects different aspects of health-related quality of life. More resilient patients have significantly better quality of life in almost all aspects of QoL.
Conference Paper
During the last decade, the number of users who look for health-related information has impressively increased. On the other hand, health professionals have less and less time to recommend useful sources of such information online to their patients. To this direction, we target at streamlining the process of providing useful online information to patients by their caregivers and improving as such the opportunities that patients have to inform themselves online about diseases and possible treatments. Using our system, relevant and high quality information is delivered to patients based on their profile, as represented in their personal healthcare record data, facilitating an easy interaction by minimizing the necessary manual effort. Specifically, in this paper, we propose a model for group recommendations following the collaborative filtering approach. Since in collaborative filtering is crucial to identify the correct set of similar users for a user in question, in addition to the traditional ratings, we pay particular attention on how to exploit health-related information for computing similarities between users. Our special focus is on providing valuable suggestions to a caregiver who is responsible for a group of users. We interpret valuable suggestions as suggestions that are both highly related and fair to the users of the group. In this line, we propose an algorithm for identifying the top-z most valuable recommendations, and present its implementation in MapReduce.