Risk Stratification for Cardiotoxicity in Breast Cancer Patients: Predicting Early Decline of LVEF After Treatment.

This study introduces AI-based models in prediction and risk assessment of early cardiac dysfunction in older breast cancer patients, as a side-effect of their cancer treatment. Using only features extracted during the baseline evaluation of each patient the proposed methodology could predict a decline in LVEF values in 4 different follow-up intervals during the first year after treatment initiation (i.e. months 3-12), with a mean accuracy of 66.67% and up to 73.55%. Selected baseline predictive factors were ranked according to their prevalence in the evaluation experiments, replicating the importance of various cardiac disorders at baseline, LVEF value and a higher age, which are all previously reported, while introducing Diabetes as an important risk factor.Clinical Relevance- Healthcare providers can better assess cardiovascular health status and risk of cardiotoxicity in the cancer treatment continuum. This will enable timely intervention and close monitoring on high risk patients while saving resources for low risk patients.

A Machine Learning Model for the prediction of the progression of carotid arterial stenoses.

Atherosclerotic carotid plaque development results in a steady narrowing of the artery lumen, which may eventually trigger catastrophic plaque rupture leading to thromboembolism and stroke. The primary cause of ischemic stroke in the EU is carotid artery disease, which increases the demand for tools for risk stratification and patient management in carotid artery disease. Additionally, advancements in cardiovascular modeling over the past few years have made it possible to build accurate three-dimensional models of patient-specific primary carotid arteries. Computational models then incorporate the aforementioned 3D models to estimate either the development of atherosclerotic plaque or a number of flow-related parameters that are linked to risk assessment. This work presents an attempt to provide a carotid artery stenosis prognostic model, utilizing non-imaging and imaging data, as well as simulated hemodynamic data. The overall methodology was trained and tested on a dataset of 41 cases with 23 carotid arteries with stable stenosis and 18 carotids with increasing stenosis degree. The highest accuracy of 71% was achieved using a neural network classifier. The novel aspect of our work is the definition of the problem that is solved, as well as the amount of simulated data that are used as input for the prognostic model.Clinical Relevance-A prognostic model for the prediction of the trajectory of carotid artery atherosclerosis is proposed, which can support physicians in critical treatment decisions.

Emotional Models for the Estimation of Arousal and Pleasure in Older Adults During Balance Rehabilitation Training.

Emotional computing has been previously applied to assess physiological behavior in a wide variety of tasks and activities. This study extends for the first time the use of emotional computing in the field of balance rehabilitation training. A proof-of-concept study was conducted to assess arousal and pleasure response to a range of physical exercises from the OTAGO and HOLOBALANCE balance rehabilitation programs with varying levels of difficulty and physical demand. Eleven participants were enrolled and performed a set of exercises wearing an ECG sensor, reporting arousal and pleasure at the end of each session. A dataset of 264 unique sessions was collected and used to extract heart rate variability (HRV) features from the measured RR intervals and automatically assess user arousal and pleasure, evaluating different classification algorithms. The results suggested that assessment of both emotions is feasible, reaching an accuracy of 72% and 74% for arousal and pleasure estimation, resnectively. Clinical Relevance- Arousal and pleasure are clinically useful indicators of patient's experience and engagement while performing balance rehabilitation exercises with novel sensing technologies and monitoring platforms.

Automated Assessment of Balance Rehabilitation Exercises With a Data-Driven Scoring Model: Algorithm Development and Validation Study.

BACKGROUND: Balance rehabilitation programs represent the most common treatments for balance disorders. Nonetheless, lack of resources and lack of highly expert physiotherapists are barriers for patients to undergo individualized rehabilitation sessions. Therefore, balance rehabilitation programs are often transferred to the home environment, with a considerable risk of the patient misperforming the exercises or failing to follow the program at all. Holobalance is a persuasive coaching system with the capacity to offer full-scale rehabilitation services at home. Holobalance involves several modules, from rehabilitation program management to augmented reality coach presentation. OBJECTIVE: The aim of this study was to design, implement, test, and evaluate a scoring model for the accurate assessment of balance rehabilitation exercises, based on data-driven techniques. METHODS: The data-driven scoring module is based on an extensive data set (approximately 1300 rehabilitation exercise sessions) collected during the Holobalance pilot study. It can be used as a training and testing data set for training machine learning (ML) models, which can infer the scoring components of all physical rehabilitation exercises. In that direction, for creating the data set, 2 independent experts monitored (in the clinic) 19 patients performing 1313 balance rehabilitation exercises and scored their performance based on a predefined scoring rubric. On the collected data, preprocessing, data cleansing, and normalization techniques were applied before deploying feature selection techniques. Finally, a wide set of ML algorithms, like random forests and neural networks, were used to identify the most suitable model for each scoring component. RESULTS: The results of the trained model improved the performance of the scoring module in terms of more accurate assessment of a performed exercise, when compared with a rule-based scoring model deployed at an early phase of the system (k-statistic value of 15.9% for sitting exercises, 20.8% for standing exercises, and 26.8% for walking exercises). Finally, the resulting performance of the model resembled the threshold of the interobserver variability, enabling trustworthy usage of the scoring module in the closed-loop chain of the Holobalance coaching system. CONCLUSIONS: The proposed set of ML models can effectively score the balance rehabilitation exercises of the Holobalance system. The models had similar accuracy in terms of Cohen kappa analysis, with interobserver variability, enabling the scoring module to infer the score of an exercise based on the collected signals from sensing devices. More specifically, for sitting exercises, the scoring model had high classification accuracy, ranging from 0.86 to 0.90. Similarly, for standing exercises, the classification accuracy ranged from 0.85 to 0.92, while for walking exercises, it ranged from 0.81 to 0.90. TRIAL REGISTRATION: ClinicalTrials.gov NCT04053829; https://clinicaltrials.gov/ct2/show/NCT04053829.

A Review of Virtual Coaching Systems in Healthcare: Closing the Loop With Real-Time Feedback.

This review focuses on virtual coaching systems that were designed to enhance healthcare interventions, combining the available sensing and system-user interaction technologies. In total, more than 1,200 research papers have been retrieved and evaluated for the purposes of this review, which were obtained from three online databases (i.e.,PubMed, Scopus and IEEE Xplore) using an extensive set of search keywords. After applying exclusion criteria, the remaining 41 research papers were used to evaluate the status of virtual coaching systems over the past 10 years and assess current and future trends in this field. The results suggest that in home coaching systems were mainly focused in promoting physical activity and a healthier lifestyle, while a wider range of medical domains was considered in systems that were evaluated in lab environment. In home patient monitoring with IoT devices and sensors was mostly limited to activity trackers, pedometers and heart rate monitoring. Real-time evaluations and personalized patient feedback was also found to be rather lacking in home coaching systems and this is the most alarming find of this analysis. Feasibility studies in controlled environment and an ongoing active research on Horizon 2020 funded projects, show that the future trends in this field are aiming to close the loop with automated patient monitoring, real-time evaluations and more precise interventions.

Prognostic factors of Rapid symptoms progression in patients with newly diagnosed parkinson's disease.

Tracking symptoms progression in the early stages of Parkinson's disease (PD) is a laborious endeavor as the disease can be expressed with vastly different phenotypes, forcing clinicians to follow a multi-parametric approach in patient evaluation, looking for not only motor symptomatology but also non-motor complications, including cognitive decline, sleep problems and mood disturbances. Being neurodegenerative in nature, PD is expected to inflict a continuous degradation in patients' condition over time. The rate of symptoms progression, however, is found to be even more chaotic than the vastly different phenotypes that can be expressed in the initial stages of PD. In this work, an analysis of baseline PD characteristics is performed using machine learning techniques, to identify prognostic factors for early rapid progression of PD symptoms. Using open data from the Parkinson's Progression Markers Initiative (PPMI) study, an extensive set of baseline patient evaluation outcomes is examined to isolate determinants of rapid progression within the first two and four years of PD. The rate of symptoms progression is estimated by tracking the change of the Movement Disorder Society-Unified Parkinson's Disease Rating Scale (MDS-UPDRS) total score over the corresponding follow-up period. Patients are ranked according to their progression rates and those who expressed the highest rates of MDS-UPDRS total score increase per year of follow-up period are assigned into the rapid progression class, using 5- and 10-quantiles partition. Classification performance against the rapid progression class was evaluated in a per quantile partition analysis scheme and in quantile-independent approach, respectively. The results shown a more accurate patient discrimination with quantile partitioning, however, a much more compact subset of baseline factors is extracted in the latter, making a more suitable for actual interventions in practice. Classification accuracy improved in all cases when using the longer 4-year follow-up period to estimate PD progression, suggesting that a prolonged patient evaluation can provide better outcomes in identifying rapid progression phenotype. Non-motor symptoms are found to be the main determinants of rapid symptoms progression in both follow-up periods, with autonomic dysfunction, mood impairment, anxiety, REM sleep behavior disorders, cognitive decline and memory impairment being alarming signs at baseline evaluation, along with rigidity symptoms, certain laboratory blood test results and genetic mutations.

Automatic Absence Seizures Detection in EEG signals: An Unsupervised Module.

Absence seizures are expressed with distinctive spike-and-wave complexes in the electroencephalogram (EEG), which can be used to automatically distinguish them from other types of seizures and interictal activity. Considering the chaotic nature of the EEG signal, it is very unlikely that such continuous, repetitive patterns with strict periodic behavior would occur naturally under normal conditions. Searching for spectral activity in the range of 2.5-4.5 Hz and assessing the presence of synchronous, repeated patterns across multiple EEG channels in an unsupervised manner, the proposed methodology provides high absence seizure detection sensitivity of 93.94% with a low false detection rate of 0.168 FD/h using the open TUSZ dataset.

Decision Support for Medication Change of Parkinson's Disease Patients.

BACKGROUND AND OBJECTIVE: Parkinson's disease (PD) is a degenerative disorder of the central nervous system for which currently there is no cure. Its treatment requires long-term, interdisciplinary disease management, and usage of typical medications, including levodopa, dopamine agonists, and enzymes, such as MAO-B inhibitors. The key goal of disease management is to prolong patients' independence and keep their quality of life. Due to the different combinations of motor and non-motor symptoms from which PD patients suffer, in addition to existing comorbidities, the change of medications and their combinations is difficult and patient-specific. To help physicians, we developed two decision support models for PD management, which suggest how to change the medication treatment. METHODS: The models were developed using DEX methodology, which integrates the qualitative multi-criteria decision modelling with rule-based expert systems. The two DEX models differ in the way the decision rules were defined. In the first model, the decision rules are based on the interviews with neurologists (DEX expert model), and in the second model, they are formed from a database of past medication change decisions (DEX data model). We assessed both models on the Parkinson's Progression Markers Initiative (PPMI) and on a questionnaire answered by 17 neurologists from 4 European countries using accuracy measure and the Jaccard index. RESULTS: Both models include 15 sub-models that address possible medication treatment changes based on the given patients' current state. In particular, the models incorporate current state changes in patients' motor symptoms (dyskinesia intensity, dyskinesia duration, OFF duration), mental problems (impulsivity, cognition, hallucinations and paranoia), epidemiologic data (patient's age, activity level) and comorbidities (cardiovascular problems, hypertension and low blood pressure). The highest accuracy of the developed sub-models for 15 medication treatment changes ranges from 69.31 to 99.06 %. CONCLUSIONS: Results show that the DEX expert model is superior to the DEX data model. The results indicate that the constructed models are sufficiently adequate and thus fit for the purpose of making "second-opinion" suggestions to decision support users.

An Unsupervised Methodology for the Detection of Epileptic Seizures Using EEG Signals: A Multi-Dataset Evaluation.

Although the electroencephalogram (EEG) is the most commonly used means to monitor epileptic patients, public EEG datasets are very scarce making it difficult to develop and validate seizure detection algorithms. In this work an unsupervised seizure detection methodology is used to isolate ictal EEG segments without requiring any apriori information or human intervention. Seizures are detected using four simple seizure detection conditions that are activated when rhythmical activity from different brain areas is simultaneously concentrated in the alpha (8-13 Hz), theta (4-7 Hz) or delta (1-3 Hz) frequency range. Then, only a small proportion of the EEG segments that are most likely to contain ictal activity is selected and presented to the physician for the final evaluation. In this way, large volumes of EEG signals can be annotated in a fraction of the time and effort that would be otherwise required. Using EEG data from 33 sessions from the Temple University Hospital (TUH) EEG Corpus, our unsupervised methodology reached, on average, 84.92% seizure detection sensitivity with 3.46 false detections per hour of EEG signals.

Predicting rapid progression of Parkinson's Disease at baseline patients evaluation.

The rate of Parkinson's Disease (PD) progression in the initial post-diagnosis years can vary significantly. In this work, a methodology for the extraction of the most informative features for predicting rapid progression of the disease is proposed, using public data from the Parkinson's Progression Markers Initiative (PPMI) and machine learning techniques. The aim is to determine if a patient is at risk of expressing rapid progression of PD symptoms from the baseline evaluation and as close to diagnosis as possible. By examining the records of 409 patients from the PPMI dataset, the features with the best predictive value at baseline patient evaluation are found to be sleep problems, daytime sleepiness and fatigue, motor symptoms at legs, cognition impairment, early axial and facial symptoms and in the most rapidly advanced cases speech issues, loss of smell and affected leg muscle reflexes.

PD_Manager: an mHealth platform for Parkinson's disease patient management.

PD_Manager is a mobile health platform designed to cover most of the aspects regarding the management of Parkinson's disease (PD) in a holistic approach. Patients are unobtrusively monitored using commercial wrist and insole sensors paired with a smartphone, to automatically estimate the severity of most of the PD motor symptoms. Besides motor symptoms monitoring, the patient's mobile application also provides various non-motor self-evaluation tests for assessing cognition, mood and nutrition to motivate them in becoming more active in managing their disease. All data from the mobile application and the sensors is transferred to a cloud infrastructure to allow easy access for clinicians and further processing. Clinicians can access this information using a separate mobile application that is specifically designed for their respective needs to provide faster and more accurate assessment of PD symptoms that facilitate patient evaluation. Machine learning techniques are used to estimate symptoms and disease progression trends to further enhance the provided information. The platform is also complemented with a decision support system (DSS) that notifies clinicians for the detection of new symptoms or the worsening of existing ones. As patient's symptoms are progressing, the DSS can also provide specific suggestions regarding appropriate medication changes.