VisCARS: Knowledge Graph-based Context-Aware Recommender System for Time-Series Data Visualization and Monitoring Dashboards.

Data visualization recommendation aims to assist the user in creating visualizations from a given dataset. The process of creating appropriate visualizations requires expert knowledge of the available data model as well as the dashboard application that is used. To relieve the user from requiring this knowledge and from the manual process of creating numerous visualizations or dashboards, we present a context-aware visualization recommender system (VisCARS) for monitoring applications that automatically recommends a personalized dashboard to the user, based on the system they are monitoring and the task they are trying to achieve. Through a knowledge graph-based approach, expert knowledge about the data and the application is included as contextual features to improve the recommendation process. A dashboard ontology is presented that describes key components in a dashboard ecosystem in order to semantically annotate all the knowledge in the graph. The recommender system leverages knowledge graph embedding and comparison techniques in combination with a context-aware collaborative filtering approach to derive recommendations based on the context, i.e., the state of the monitored system, and the end-user preferences. The proposed methodology is implemented and integrated in a dynamic dashboard solution. The resulting recommender system is evaluated on a smart healthcare use-case through a quantitative performance and scalability analysis as well as a qualitative user study. The results highlight the performance of the proposed solution compared to the state-of-the-art and its potential for time-critical monitoring applications.

Acoustic and prosodic speech features reflect physiological stress but not isolated negative affect: a multi-paradigm study on psychosocial stressors.

Heterogeneity in speech under stress has been a recurring issue in stress research, potentially due to varied stress induction paradigms. This study investigated speech features in semi-guided speech following two distinct psychosocial stress paradigms (Cyberball and MIST) and their respective control conditions. Only negative affect increased during Cyberball, while self-reported stress, skin conductance response rate, and negative affect increased during MIST. Fundamental frequency (F0), speech rate, and jitter significantly changed during MIST, but not Cyberball; HNR and shimmer showed no expected changes. The results indicate that observed speech features are robust in semi-guided speech and sensitive to stressors eliciting additional physiological stress responses, not solely decreases in negative affect. These differences between stressors may explain literature heterogeneity. Our findings support the potential of speech as a stress level biomarker, especially when stress elicits physiological reactions, similar to other biomarkers. This highlights its promise as a tool for measuring stress in everyday settings, considering its affordability, non-intrusiveness, and ease of collection. Future research should test these results' robustness and specificity in naturalistic settings, such as freely spoken speech and noisy environments while exploring and validating a broader range of informative speech features in the context of stress.

Unlocking the potential of smartphone and ambient sensors for ADL detection.

The detection of Activities of Daily Living (ADL) holds significant importance in a range of applications, including elderly care and health monitoring. Our research focuses on the relevance of ADL detection in elderly care, highlighting the importance of accurate and unobtrusive monitoring. In this paper, we present a novel approach that that leverages smartphone data as the primary source for detecting ADLs. Additionally, we investigate the possibilities offered by ambient sensors installed in smart home environments to complement the smartphone data and optimize the ADL detection. Our approach uses a Long Short-Term Memory (LSTM) model. One of the key contributions of our work is defining ADL detection as a multilabeling problem, allowing us to detect different activities that occur simultaneously. This is particularly valuable since in real-world scenarios, individuals can perform multiple activities concurrently, such as cooking while watching TV. We also made use of unlabeled data to further enhance the accuracy of our model. Performance is evaluated on a real-world collected dataset, strengthening reliability of our findings. We also made the dataset openly available for further research and analysis. Results show that utilizing smartphone data alone already yields satisfactory results, above 50% true positive rate and balanced accuracy for all activities, providing a convenient and non-intrusive method for ADL detection. However, by incorporating ambient sensors, as an additional data source, one can improve the balanced accuracy of the ADL detection by 7% and 8% of balanced accuracy and true positive rate respectively, on average.

Patients with chronic cluster headache may show reduced activity energy expenditure on ambulatory wrist actigraphy recordings during daytime attacks.

OBJECTIVE: To investigate the changes in activity energy expenditure (AEE) throughout daytime cluster headache (CH) attacks in patients with chronic CH and to evaluate the usefulness of actigraphy as a digital biomarker of CH attacks. BACKGROUND: CH is a primary headache disorder characterized by attacks of severe to very severe unilateral pain (orbital, supraorbital, temporal, or in any combination of these sites), with ipsilateral cranial autonomic symptoms and/or a sense of restlessness or agitation. We hypothesized increased AEE from hyperactivity during attacks measured by actigraphy. METHODS: An observational study including patients with chronic CH was conducted. During 21 days, patients wore an actigraphy device on the nondominant wrist and recorded CH attack-related data in a dedicated smartphone application. Accelerometer data were used for the calculation of AEE before and during daytime CH attacks that occurred in ambulatory settings, and without restrictions on acute and preventive headache treatment. We compared the activity and movements during the pre-ictal, ictal, and postictal phases with data from wrist-worn actigraphy with time-concordant intervals during non-headache periods. RESULTS: Four patients provided 34 attacks, of which 15 attacks met the eligibility criteria for further analysis. In contrast with the initial hypothesis of increased energy expenditure during CH attacks, a decrease in movement was observed during the pre-ictal phase (30 min before onset to onset) and during the headache phase. A significant decrease (p < .01) in the proportion of high-intensity movement during headache attacks, of which the majority were oxygen-treated, was observed. This trend was less present for low-intensity movements. CONCLUSION: The unexpected decrease in AEE during the pre-ictal and headache phase of daytime CH attacks in patients with chronic CH under acute and preventive treatment in ambulatory settings has important implications for future research on wrist actigraphy in CH.

Ecologically valid speech collection in behavioral research: The Ghent Semi-spontaneous Speech Paradigm (GSSP).

This paper introduces the Ghent Semi-spontaneous Speech Paradigm (GSSP), a new method for collecting unscripted speech data for affective-behavioral research in both experimental and real-world settings through the description of peer-rated pictures with a consistent affective load. The GSSP was designed to meet five criteria: (1) allow flexible speech recording durations, (2) provide a straightforward and non-interfering task, (3) allow for experimental control, (4) favor spontaneous speech for its prosodic richness, and (5) require minimal human interference to enable scalability. The validity of the GSSP was evaluated through an online task, in which this paradigm was implemented alongside a fixed-text read-aloud task. The results indicate that participants were able to describe images with an adequate duration, and acoustic analysis demonstrated a trend for most features in line with the targeted speech styles (i.e., unscripted spontaneous speech versus scripted read-aloud speech). A speech style classification model using acoustic features achieved a balanced accuracy of 83% on within-dataset validation, indicating separability between the GSSP and read-aloud speech task. Furthermore, when validating this model on an external dataset that contains interview and read-aloud speech, a balanced accuracy score of 70% is obtained, indicating an acoustic correspondence between the GSSP speech and spontaneous interviewee speech. The GSSP is of special interest for behavioral and speech researchers looking to capture spontaneous speech, both in longitudinal ambulatory behavioral studies and laboratory studies. To facilitate future research on speech styles, acoustics, and affective states, the task implementation code, the collected dataset, and analysis notebooks are available.

Tryp: a dataset of microscopy images of unstained thick blood smears for trypanosome detection.

Trypanosomiasis, a neglected tropical disease (NTD), challenges communities in sub-Saharan Africa and Latin America. The World Health Organization underscores the need for practical, field-adaptable diagnostics and rapid screening tools to address the negative impact of NTDs. While artificial intelligence has shown promising results in disease screening, the lack of curated datasets impedes progress. In response to this challenge, we developed the Tryp dataset, comprising microscopy images of unstained thick blood smears containing the Trypanosoma brucei brucei parasite. The Tryp dataset provides bounding box annotations for tightly enclosed regions containing the parasite for 3,085 positive images, and 93 images collected from negative blood samples. The Tryp dataset represents the largest of its kind. Furthermore, we provide a benchmark on three leading deep learning-based object detection techniques that demonstrate the feasibility of AI for this task. Overall, the availability of the Tryp dataset is expected to facilitate research advancements in diagnostic screening for this disease, which may lead to improved healthcare outcomes for the communities impacted.

From Lab to Real World: Assessing the Effectiveness of Human Activity Recognition and Optimization through Personalization.

Human activity recognition (HAR) algorithms today are designed and evaluated on data collected in controlled settings, providing limited insights into their performance in real-world situations with noisy and missing sensor data and natural human activities. We present a real-world HAR open dataset compiled from a wristband equipped with a triaxial accelerometer. During data collection, participants had autonomy in their daily life activities, and the process remained unobserved and uncontrolled. A general convolutional neural network model was trained on this dataset, achieving a mean balanced accuracy (MBA) of 80%. Personalizing the general model through transfer learning can yield comparable and even superior results using fewer data, with the MBA improving to 85%. To emphasize the issue of insufficient real-world training data, we conducted training of the model using the public MHEALTH dataset, resulting in 100% MBA. However, upon evaluating the MHEALTH-trained model on our real-world dataset, the MBA drops to 62%. After personalizing the model with real-world data, an improvement of 17% in the MBA is achieved. This paper showcases the potential of transfer learning to make HAR models trained in different contexts (lab vs. real-world) and on different participants perform well for new individuals with limited real-world labeled data available.

Generalizable calibrated machine learning models for real-time atrial fibrillation risk prediction in ICU patients.

BACKGROUND: Atrial Fibrillation (AF) is the most common arrhythmia in the intensive care unit (ICU) and is associated with increased morbidity and mortality. Identification of patients at risk for AF is not routinely performed as AF prediction models are almost solely developed for the general population or for particular ICU populations. However, early AF risk identification could help to take targeted preemptive actions and possibly reduce morbidity and mortality. Predictive models need to be validated across hospitals with different standards of care and convey their predictions in a clinically useful manner. Therefore, we designed AF risk models for ICU patients using uncertainty quantification to provide a risk score and evaluated them on multiple ICU datasets. METHODS: Three CatBoost models, utilizing feature windows comprising data 1.5-13.5, 6-18, or 12-24 hours before AF occurrence, were built using 2-repeat-10-fold cross-validation on AmsterdamUMCdb, the first freely available European ICU database. Furthermore, AF Patients were matched with no-AF patients for training. Transferability was validated using a direct and a recalibration evaluation on two independent external datasets, MIMIC-IV and GUH. The calibration of the predicted probability, used as an AF risk score, was measured using the Expected Calibration Error (ECE) and the presented Expected Signed Calibration Error (ESCE). Additionally, all models were evaluated across time during the ICU stay. RESULTS: The model performance reached Areas Under the Curve (AUCs) of 0.81 at internal validation. Direct external validation showed partial generalizability with AUCs reaching 0.77. However, recalibration resulted in performances matching or exceeding that of the internal validation. All models furthermore showed calibration capabilities demonstrating adequate risk prediction competence. CONCLUSION: Ultimately, recalibrating models reduces the challenge of generalization to unseen datasets. Moreover, utilizing the patient-matching methodology together with the assessment of uncertainty calibration can serve as a step toward the development of clinical AF prediction models.

Pathogen-based target attainment of optimized continuous infusion dosing regimens of piperacillin-tazobactam and meropenem in surgical ICU patients: a prospective single center observational study.

BACKGROUND: Several studies have indicated that commonly used piperacillin-tazobactam (TZP) and meropenem (MEM) dosing regimens lead to suboptimal plasma concentrations for a range of pharmacokinetic/pharmacodynamic (PK/PD) targets in intensive care unit (ICU) patients. These targets are often based on a hypothetical worst-case scenario, possibly overestimating the percentage of suboptimal concentrations. We aimed to evaluate the pathogen-based clinically relevant target attainment (CRTA) and therapeutic range attainment (TRA) of optimized continuous infusion dosing regimens of TZP and MEM in surgical ICU patients. METHODS: A single center prospective observational study was conducted between March 2016 and April 2019. Free plasma concentrations were calculated by correcting total plasma concentrations, determined on remnants of blood gas samples by ultra-performance liquid chromatography with tandem mass spectrometry, for their protein binding. Break points (BP) of identified pathogens were derived from epidemiological cut-off values. CRTA was defined as a corrected measured total serum concentration above the BP and calculated for increasing BP multiplications up to 6 × BP. The upper limit of the therapeutic range was set at 157.2 mg/L for TZP and 45 mg/L for MEM. As a worst-case scenario, a BP of 16 mg/L for TZP and 2 mg/L for MEM was used. RESULTS: 781 unique patients were included with 1036 distinctive beta-lactam antimicrobial prescriptions (731 TZP, 305 MEM) for 1003 unique infections/prophylactic regimens (750 TZP, 323 MEM). 2810 samples were available (1892 TZP, 918 MEM). The median corrected plasma concentration for TZP was 86.4 mg/L [IQR 56.2-148] and 16.2 mg/L [10.2-25.5] for MEM. CRTA and TRA was consistently higher for the pathogen-based scenario than for the worst-case scenario, but nonetheless, a substantial proportion of samples did not attain commonly used PK/PD targets. CONCLUSION: Despite these pathogen-based data demonstrating that CRTA and TRA is higher than in the often-used theoretical worst-case scenario, a substantial proportion of samples did not attain commonly used PK/PD targets when using optimised continuous infusion dosing regimens. Therefore, more dosing optimization research seems warranted. At the same time, a 'pathogen-based analysis' approach might prove to be more sensible than a worst-case scenario approach when evaluating target attainment and linked clinical outcomes.

Real-Time Estimation and Monitoring of COVID-19 Aerosol Transmission Risk in Office Buildings.

A healthy and safe indoor environment is an important part of containing the coronavirus disease 2019 (COVID-19) pandemic. Therefore, this work presents a real-time Internet of things (IoT) software architecture to automatically calculate and visualize a COVID-19 aerosol transmission risk estimation. This risk estimation is based on indoor climate sensor data, such as carbon dioxide (CO2) and temperature, which is fed into Streaming MASSIF, a semantic stream processing platform, to perform the computations. The results are visualized on a dynamic dashboard that automatically suggests appropriate visualizations based on the semantics of the data. To evaluate the complete architecture, the indoor climate during the student examination periods of January 2020 (pre-COVID) and January 2021 (mid-COVID) was analyzed. When compared to each other, we observe that the COVID-19 measures in 2021 resulted in a safer indoor environment.

Parameter Efficient Neural Networks With Singular Value Decomposed Kernels.

Traditionally, neural networks are viewed from the perspective of connected neuron layers represented as matrix multiplications. We propose to compose these weight matrices from a set of orthogonal basis matrices by approaching them as elements of the real matrices vector space under addition and multiplication. Making use of the Kronecker product for vectors, this composition is unified with the singular value decomposition (SVD) of the weight matrix. The orthogonal components of this SVD are trained with a descent curve on the Stiefel manifold using the Cayley transform. Next, update equations for the singular values and initialization routines are derived. Finally, acceleration for stochastic gradient descent optimization using this formulation is discussed. Our proposed method allows more parameter-efficient representations of weight matrices in neural networks. These decomposed weight matrices achieve maximal performance in both standard and more complicated neural architectures. Furthermore, the more parameter-efficient decomposed layers are shown to be less dependent on optimization and better conditioned. As a tradeoff, training time is increased up to a factor of 2. These observations are consequently attributed to the properties of the method and choice of optimization over the manifold of orthogonal matrices.

Acoustic speech features in social comparison: how stress impacts the way you sound.

The use of speech as a digital biomarker to detect stress levels is increasingly gaining attention. Yet, heterogeneous effects of stress on specific acoustic speech features have been observed, possibly due to previous studies' use of different stress labels/categories and the lack of solid stress induction paradigms or validation of experienced stress. Here, we deployed a controlled, within-subject psychosocial stress induction experiment in which participants received both neutral (control condition) and negative (negative condition) comparative feedback after solving a challenging cognitive task. This study is the first to use a (non-actor) within-participant design that verifies a successful stress induction using both self-report (i.e., decreased reported valence) and physiological measures (i.e., increased heart rate acceleration using event-related cardiac responses during feedback exposure). Analyses of acoustic speech features showed a significant increase in Fundamental Frequency (F0) and Harmonics-to-Noise Ratio (HNR), and a significant decrease in shimmer during the negative feedback condition. Our results using read-out-loud speech comply with earlier research, yet we are the first to validate these results in a well-controlled but ecologically-valid setting to guarantee the generalization of our findings to real-life settings. Further research should aim to replicate these results in a free speech setting to test the robustness of our findings for real-world settings and should include semantics to also take into account what you say and not only how you say it.

Assessing the added value of context during stress detection from wearable data.

BACKGROUND: Insomnia, eating disorders, heart problems and even strokes are just some of the illnesses that reveal the negative impact of stress overload on health and well-being. Early detection of stress is therefore of utmost importance. Whereas the gold-standard for detecting stress is by means of questionnaires, more recent work uses wearable sensors to find continuous and qualitative physical markers of stress. As some physiological stress responses, e.g. increased heart rate or sweating and chills, might also occur when doing sports, a more profound approach is needed for stress detection than purely considering physiological data. METHODS: In this paper, we analyse the added value of context information during stress detection from wearable data. We do so by comparing the performance of models trained purely on physiological data and models trained on physiological and context data. We consider the user's activity and hours of sleep as context information, where we compare the influence of user-given context versus machine learning derived context. RESULTS: Context-aware models reach higher accuracy and lower standard deviations in comparison to the baseline (physiological) models. We also observe higher accuracy and improved weighted F1 score when incorporating machine learning predicted, instead of user-given, activities as context information. CONCLUSIONS: In this paper we show that considering context information when performing stress detection from wearables leads to better performance. We also show that it is possible to move away from human labeling and rely only on the wearables for both physiology and context.

Development and evaluation of uncertainty quantifying machine learning models to predict piperacillin plasma concentrations in critically ill patients.

BACKGROUND: Beta-lactam antimicrobial concentrations are frequently suboptimal in critically ill patients. Population pharmacokinetic (PopPK) modeling is the golden standard to predict drug concentrations. However, currently available PopPK models often lack predictive accuracy, making them less suited to guide dosing regimen adaptations. Furthermore, many currently developed models for clinical applications often lack uncertainty quantification. We, therefore, aimed to develop machine learning (ML) models for the prediction of piperacillin plasma concentrations while also providing uncertainty quantification with the aim of clinical practice. METHODS: Blood samples for piperacillin analysis were prospectively collected from critically ill patients receiving continuous infusion of piperacillin/tazobactam. Interpretable ML models for the prediction of piperacillin concentrations were designed using CatBoost and Gaussian processes. Distribution-based Uncertainty Quantification was added to the CatBoost model using a proposed Quantile Ensemble method, useable for any model optimizing a quantile function. These models are subsequently evaluated using the distribution coverage error, a proposed interpretable uncertainty quantification calibration metric. Development and internal evaluation of the ML models were performed on the Ghent University Hospital database (752 piperacillin concentrations from 282 patients). Ensuing, ML models were compared with a published PopPK model on a database from the University Medical Centre of Groningen where a different dosing regimen is used (46 piperacillin concentrations from 15 patients.). RESULTS: The best performing model was the Catboost model with an RMSE and [Formula: see text] of 31.94-0.64 and 33.53-0.60 for internal evaluation with and without previous concentration. Furthermore, the results prove the added value of the proposed Quantile Ensemble model in providing clinically useful individualized uncertainty predictions and show the limits of homoscedastic methods like Gaussian Processes in clinical applications. CONCLUSIONS: Our results show that ML models can consistently estimate piperacillin concentrations with acceptable and high predictive accuracy when identical dosing regimens as in the training data are used while providing highly relevant uncertainty predictions. However, generalization capabilities to other dosing schemes are limited. Notwithstanding, incorporating ML models in therapeutic drug monitoring programs seems definitely promising and the current work provides a basis for validating the model in clinical practice.

Affordable artificial intelligence-based digital pathology for neglected tropical diseases: A proof-of-concept for the detection of soil-transmitted helminths and Schistosoma mansoni eggs in Kato-Katz stool thick smears.

BACKGROUND: With the World Health Organization's (WHO) publication of the 2021-2030 neglected tropical diseases (NTDs) roadmap, the current gap in global diagnostics became painfully apparent. Improving existing diagnostic standards with state-of-the-art technology and artificial intelligence has the potential to close this gap. METHODOLOGY/PRINCIPAL FINDINGS: We prototyped an artificial intelligence-based digital pathology (AI-DP) device to explore automated scanning and detection of helminth eggs in stool prepared with the Kato-Katz (KK) technique, the current diagnostic standard for diagnosing soil-transmitted helminths (STHs; Ascaris lumbricoides, Trichuris trichiura and hookworms) and Schistosoma mansoni (SCH) infections. First, we embedded a prototype whole slide imaging scanner into field studies in Cambodia, Ethiopia, Kenya and Tanzania. With the scanner, over 300 KK stool thick smears were scanned, resulting in total of 7,780 field-of-view (FOV) images containing 16,990 annotated helminth eggs (Ascaris: 8,600; Trichuris: 4,083; hookworms: 3,623; SCH: 684). Around 90% of the annotated eggs were used to train a deep learning-based object detection model. From an unseen test set of 752 FOV images containing 1,671 manually verified STH and SCH eggs (the remaining 10% of annotated eggs), our trained object detection model extracted and classified helminth eggs from co-infected FOV images in KK stool thick smears, achieving a weighted average precision (± standard deviation) of 94.9% ± 0.8% and a weighted average recall of 96.1% ± 2.1% across all four helminth egg species. CONCLUSIONS/SIGNIFICANCE: We present a proof-of-concept for an AI-DP device for automated scanning and detection of helminth eggs in KK stool thick smears. We identified obstacles that need to be addressed before the diagnostic performance can be evaluated against the target product profiles for both STH and SCH. Given that these obstacles are primarily associated with the required hardware and scanning methodology, opposed to the feasibility of AI-based results, we are hopeful that this research can support the 2030 NTDs road map and eventually other poverty-related diseases for which microscopy is the diagnostic standard.

mBrain: towards the continuous follow-up and headache classification of primary headache disorder patients.

BACKGROUND: The diagnosis of headache disorders relies on the correct classification of individual headache attacks. Currently, this is mainly done by clinicians in a clinical setting, which is dependent on subjective self-reported input from patients. Existing classification apps also rely on self-reported information and lack validation. Therefore, the exploratory mBrain study investigates moving to continuous, semi-autonomous and objective follow-up and classification based on both self-reported and objective physiological and contextual data. METHODS: The data collection set-up of the observational, longitudinal mBrain study involved physiological data from the Empatica E4 wearable, data-driven machine learning (ML) algorithms detecting activity, stress and sleep events from the wearables' data modalities, and a custom-made application to interact with these events and keep a diary of contextual and headache-specific data. A knowledge-based classification system for individual headache attacks was designed, focusing on migraine, cluster headache (CH) and tension-type headache (TTH) attacks, by using the classification criteria of ICHD-3. To show how headache and physiological data can be linked, a basic knowledge-based system for headache trigger detection is presented. RESULTS: In two waves, 14 migraine and 4 CH patients participated (mean duration 22.3 days). 133 headache attacks were registered (98 by migraine, 35 by CH patients). Strictly applying ICHD-3 criteria leads to 8/98 migraine without aura and 0/35 CH classifications. Adapted versions yield 28/98 migraine without aura and 17/35 CH classifications, with 12/18 participants having mostly diagnosis classifications when episodic TTH classifications (57/98 and 32/35) are ignored. CONCLUSIONS: Strictly applying the ICHD-3 criteria on individual attacks does not yield good classification results. Adapted versions yield better results, with the mostly classified phenotype (migraine without aura vs. CH) matching the diagnosis for 12/18 patients. The absolute number of migraine without aura and CH classifications is, however, rather low. Example cases can be identified where activity and stress events explain patient-reported headache triggers. Continuous improvement of the data collection protocol, ML algorithms, and headache classification criteria (including the investigation of integrating physiological data), will further improve future headache follow-up, classification and trigger detection. Trial registration This trial was retrospectively registered with number NCT04949204 on 24 June 2021 at www. CLINICALTRIALS: gov .

Artificial Intelligence in Infection Management in the ICU.

This article is one of ten reviews selected from the Annual Update in Intensive Care and Emergency Medicine 2022. Other selected articles can be found online at https://www.biomedcentral.com/collections/annualupdate2022 . Further information about the Annual Update in Intensive Care and Emergency Medicine is available from https://link.springer.com/bookseries/8901 .

A Semi-Supervised Approach with Monotonic Constraints for Improved Remaining Useful Life Estimation.

Remaining useful life is of great value in the industry and is a key component of Prognostics and Health Management (PHM) in the context of the Predictive Maintenance (PdM) strategy. Accurate estimation of the remaining useful life (RUL) is helpful for optimizing maintenance schedules, obtaining insights into the component degradation, and avoiding unexpected breakdowns. This paper presents a methodology for creating health index models with monotonicity in a semi-supervised approach. The health indexes are then used for enhancing remaining useful life estimation models. The methodology is evaluated on two bearing datasets. Results demonstrate the advantage of using the monotonic health index for obtaining insights into the bearing degradation and for remaining useful life estimation.

Overly optimistic prediction results on imbalanced data: a case study of flaws and benefits when applying over-sampling.

Information extracted from electrohysterography recordings could potentially prove to be an interesting additional source of information to estimate the risk on preterm birth. Recently, a large number of studies have reported near-perfect results to distinguish between recordings of patients that will deliver term or preterm using a public resource, called the Term/Preterm Electrohysterogram database. However, we argue that these results are overly optimistic due to a methodological flaw being made. In this work, we focus on one specific type of methodological flaw: applying over-sampling before partitioning the data into mutually exclusive training and testing sets. We show how this causes the results to be biased using two artificial datasets and reproduce results of studies in which this flaw was identified. Moreover, we evaluate the actual impact of over-sampling on predictive performance, when applied prior to data partitioning, using the same methodologies of related studies, to provide a realistic view of these methodologies' generalization capabilities. We make our research reproducible by providing all the code under an open license.

Clinical information extraction for preterm birth risk prediction.

This paper contributes to the pursuit of leveraging unstructured medical notes to structured clinical decision making. In particular, we present a pipeline for clinical information extraction from medical notes related to preterm birth, and discuss the main challenges as well as its potential for clinical practice. A large collection of medical notes, created by staff during hospitalizations of patients who were at risk of delivering preterm, was gathered and analyzed. Based on an annotated collection of notes, we trained and evaluated information extraction components to discover clinical entities such as symptoms, events, anatomical sites and procedures, as well as attributes linked to these clinical entities. In a retrospective study, we show that these are highly informative for clinical decision support models that are trained to predict whether delivery is likely to occur within specific time windows, in combination with structured information from electronic health records.