Remote Patient Monitoring and Machine Learning in Acute Exacerbations of Chronic Obstructive Pulmonary Disease: Dual Systematic Literature Review and Narrative Synthesis.

BACKGROUND: Acute exacerbations of chronic obstructive pulmonary disease (AECOPD) are associated with high mortality, morbidity, and poor quality of life and constitute a substantial burden to patients and health care systems. New approaches to prevent or reduce the severity of AECOPD are urgently needed. Internationally, this has prompted increased interest in the potential of remote patient monitoring (RPM) and digital medicine. RPM refers to the direct transmission of patient-reported outcomes, physiological, and functional data, including heart rate, weight, blood pressure, oxygen saturation, physical activity, and lung function (spirometry), directly to health care professionals through automation, web-based data entry, or phone-based data entry. Machine learning has the potential to enhance RPM in chronic obstructive pulmonary disease by increasing the accuracy and precision of AECOPD prediction systems. OBJECTIVE: This study aimed to conduct a dual systematic review. The first review focuses on randomized controlled trials where RPM was used as an intervention to treat or improve AECOPD. The second review examines studies that combined machine learning with RPM to predict AECOPD. We review the evidence and concepts behind RPM and machine learning and discuss the strengths, limitations, and clinical use of available systems. We have generated a list of recommendations needed to deliver patient and health care system benefits. METHODS: A comprehensive search strategy, encompassing the Scopus and Web of Science databases, was used to identify relevant studies. A total of 2 independent reviewers (HMGG and CM) conducted study selection, data extraction, and quality assessment, with discrepancies resolved through consensus. Data synthesis involved evidence assessment using a Critical Appraisal Skills Programme checklist and a narrative synthesis. Reporting followed PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. RESULTS: These narrative syntheses suggest that 57% (16/28) of the randomized controlled trials for RPM interventions fail to achieve the required level of evidence for better outcomes in AECOPD. However, the integration of machine learning into RPM demonstrates promise for increasing the predictive accuracy of AECOPD and, therefore, early intervention. CONCLUSIONS: This review suggests a transition toward the integration of machine learning into RPM for predicting AECOPD. We discuss particular RPM indices that have the potential to improve AECOPD prediction and highlight research gaps concerning patient factors and the maintained adoption of RPM. Furthermore, we emphasize the importance of a more comprehensive examination of patient and health care burdens associated with RPM, along with the development of practical solutions.

Exacerbation predictive modelling using real-world data from the myCOPD app.

Background: Acute exacerbations of COPD (AECOPD) are episodes of breathlessness, cough and sputum which are associated with the risk of hospitalisation, progressive lung function decline and death. They are often missed or diagnosed late. Accurate timely intervention can improve these poor outcomes. Digital tools can be used to capture symptoms and other clinical data in COPD. This study aims to apply machine learning to the largest available real-world digital dataset to develop AECOPD Prediction tools which could be used to support early intervention and improve clinical outcomes. Objective: To create and validate a machine learning predictive model that forecasts exacerbations of COPD 1-8 days in advance. The model is based on routine patient-entered data from myCOPD self-management app. Method: Adaptations of the AdaBoost algorithm were employed as machine learning approaches. The dataset included 506 patients users between 2017 and 2021. 55,066 app records were available for stable COPD event labels and 1263 records of AECOPD event labels. The data used for training the model included COPD assessment test (CAT) scores, symptom scores, smoking history, and previous exacerbation frequency. All exacerbation records used in the model were confined to the 1-8 days preceding a self-reported exacerbation event. Results: TheEasyEnsemble Classifier resulted in a Sensitivity of 67.0 % and a Specificity of 65 % with a positive predictive value (PPV) of 5.0 % and a negative predictive value (NPV) of 98.9 %. An AdaBoost model with a cost-sensitive decision tree resulted in a a Sensitivity of 35.0 % and a Specificity of 89.0 % with a PPV of 7.08 % and NPV of 98.3 %. Conclusion: This preliminary analysis demonstrates that machine learning approaches to real-world data from a widely deployed digital therapeutic has the potential to predict AECOPD and can be used to confidently exclude the risk of exacerbations of COPD within the next 8 days.

Detecting dynamical changes in vital signs using switching Kalman filter.

Vital signs contain valuable information about patients' health status during their stay in general wards, when the deterioration process begins. The use of methods to predict and detect regime changes such as switching models can help to understand how vital sign dynamics are altered in disease conditions. However, time series of vital signs are remarkably non-stationary in these scenarios. The objective of this study is to quantify the potential bias of switching models in the presence of non-stationarities, when the inputs are spectral, symbolic and entropy indices. To distinguish stationary from non-stationary periods, a test was used to verify the stability of the mean and variance over short periods. Then, we compared the results from a switching Kalman filter (SKF) model trained using indices obtained over stationary periods with a model trained solely over non-stationary periods. It was observed that indices measured over stationary and non-stationary periods were significantly different. The results of switching models were highly dependent on the indices that were used as inputs. The multi-scale entropy (MSE) approach presented the highest correlation values between non-stationary and stationary switches, an average correlation coefficient of 38%.

Scrutinizing human MHC polymorphism: Supertype analysis using Poisson-Boltzmann electrostatics and clustering.

Peptide-binding MHC proteins are thought the most variable across the human population; the extreme MHC polymorphism observed is functionally important and results from constrained divergent evolution. MHCs have vital functions in immunology and homeostasis: cell surface MHC class I molecules report cell status to CD8+ T cells, NKT cells and NK cells, thus playing key roles in pathogen defence, as well as mediating smell recognition, mate choice, Adverse Drug Reactions, and transplantation rejection. MHC peptide specificity falls into several supertypes exhibiting commonality of binding. It seems likely that other supertypes exist relevant to other functions. Since comprehensive experimental characterization is intractable, structure-based bioinformatics is the only viable solution. We modelled functional MHC proteins by homology and used calculated Poisson-Boltzmann electrostatics projected from the top surface of the MHC as multi-dimensional descriptors, analysing them using state-of-the-art dimensionality reduction techniques and clustering algorithms. We were able to recover the 3 MHC loci as separate clusters and identify clear sub-groups within them, vindicating unequivocally our choice of both data representation and clustering strategy. We expect this approach to make a profound contribution to the study of MHC polymorphism and its functional consequences, and, by extension, other burgeoning structural systems, such as GPCRs.

Regularity analysis of nocturnal oximetry recordings to assist in the diagnosis of sleep apnoea syndrome.

The relationship between sleep apnoea-hypopnoea syndrome (SAHS) severity and the regularity of nocturnal oxygen saturation (SaO2) recordings was analysed. Three different methods were proposed to quantify regularity: approximate entropy (AEn), sample entropy (SEn) and kernel entropy (KEn). A total of 240 subjects suspected of suffering from SAHS took part in the study. They were randomly divided into a training set (96 subjects) and a test set (144 subjects) for the adjustment and assessment of the proposed methods, respectively. According to the measurements provided by AEn, SEn and KEn, higher irregularity of oximetry signals is associated with SAHS-positive patients. Receiver operating characteristic (ROC) and Pearson correlation analyses showed that KEn was the most reliable predictor of SAHS. It provided an area under the ROC curve of 0.91 in two-class classification of subjects as SAHS-negative or SAHS-positive. Moreover, KEn measurements from oximetry data exhibited a linear dependence on the apnoea-hypopnoea index, as shown by a correlation coefficient of 0.87. Therefore, these measurements could be used for the development of simplified diagnostic techniques in order to reduce the demand for polysomnographies. Furthermore, KEn represents a convincing alternative to AEn and SEn for the diagnostic analysis of noisy biomedical signals.

Age-associated changes in long-chain fatty acid profile during healthy aging promote pro-inflammatory monocyte polarization via PPARγ.

Differences in lipid metabolism associate with age-related disease development and lifespan. Inflammation is a common link between metabolic dysregulation and aging. Saturated fatty acids (FAs) initiate pro-inflammatory signalling from many cells including monocytes; however, no existing studies have quantified age-associated changes in individual FAs in relation to inflammatory phenotype. Therefore, we have determined the plasma concentrations of distinct FAs by gas chromatography in 26 healthy younger individuals (age < 30 years) and 21 healthy FA individuals (age > 50 years). Linear mixed models were used to explore the association between circulating FAs, age and cytokines. We showed that plasma saturated, poly- and mono-unsaturated FAs increase with age. Circulating TNF-α and IL-6 concentrations increased with age, whereas IL-10 and TGF-ß1 concentrations decreased. Oxidation of MitoSOX Red was higher in leucocytes from FA adults, and plasma oxidized glutathione concentrations were higher. There was significant colinearity between plasma saturated FAs, indicative of their metabolic relationships. Higher levels of the saturated FAs C18:0 and C24:0 were associated with lower TGF-ß1 concentrations, and higher C16:0 were associated with higher TNF-α concentrations. We further examined effects of the aging FA profile on monocyte polarization and metabolism in THP1 monocytes. Monocytes preincubated with C16:0 increased secretion of pro-inflammatory cytokines in response to phorbol myristate acetate-induced differentiation through ceramide-dependent inhibition of PPARγ activity. Conversely, C18:1 primed a pro-resolving macrophage which was PPARγ dependent and ceramide dependent and which required oxidative phosphorylation. These data suggest that a midlife adult FA profile impairs the switch from proinflammatory to lower energy, requiring anti-inflammatory macrophages through metabolic reprogramming.

Early warnings of heart rate deterioration.

Hospitals can experience difficulty in detecting and responding to early signs of patient deterioration leading to late intensive care referrals, excess mortality and morbidity, and increased hospital costs. Our study aims to explore potential indicators of physiological deterioration by the analysis of vital-signs. The dataset used comprises heart rate (HR) measurements from MIMIC II waveform database, taken from six patients admitted to the Intensive Care Unit (ICU) and diagnosed with severe sepsis. Different indicators were considered: 1) generic early warning indicators used in ecosystems analysis (autocorrelation at-1-lag (ACF1), standard deviation (SD), skewness, kurtosis and heteroskedasticity) and 2) entropy analysis (kernel entropy and multi scale entropy). Our preliminary findings suggest that when a critical transition is approaching, the equilibrium state changes what is visible in the ACF1 and SD values, but also by the analysis of the entropy. Entropy allows to characterize the complexity of the time series during the hospital stay and can be used as an indicator of regime shifts in a patient's condition. One of the main problems is its dependency of the scale used. Our results demonstrate that different entropy scales should be used depending of the level of entropy verified.

PIP-DB: the Protein Isoelectric Point database.

UNLABELLED: A protein's isoelectric point or pI corresponds to the solution pH at which its net surface charge is zero. Since the early days of solution biochemistry, the pI has been recorded and reported, and thus literature reports of pI abound. The Protein Isoelectric Point database (PIP-DB) has collected and collated these data to provide an increasingly comprehensive database for comparison and benchmarking purposes. A web application has been developed to warehouse this database and provide public access to this unique resource. PIP-DB is a web-enabled SQL database with an HTML GUI front-end. PIP-DB is fully searchable across a range of properties. AVAILABILITY AND IMPLEMENTATION: The PIP-DB database and documentation are available at http://www.pip-db.org.

Visualization of molecular fingerprints.

A visualization plot of a data set of molecular data is a useful tool for gaining insight into a set of molecules. In chemoinformatics, most visualization plots are of molecular descriptors, and the statistical model most often used to produce a visualization is principal component analysis (PCA). This paper takes PCA, together with four other statistical models (NeuroScale, GTM, LTM, and LTM-LIN), and evaluates their ability to produce clustering in visualizations not of molecular descriptors but of molecular fingerprints. Two different tasks are addressed: understanding structural information (particularly combinatorial libraries) and relating structure to activity. The quality of the visualizations is compared both subjectively (by visual inspection) and objectively (with global distance comparisons and local k-nearest-neighbor predictors). On the data sets used to evaluate clustering by structure, LTM is found to perform significantly better than the other models. In particular, the clusters in LTM visualization space are consistent with the relationships between the core scaffolds that define the combinatorial sublibraries. On the data sets used to evaluate clustering by activity, LTM again gives the best performance but by a smaller margin. The results of this paper demonstrate the value of using both a nonlinear projection map and a Bernoulli noise model for modeling binary data.

Increased diversity of libraries from libraries: chemoinformatic analysis of bis-diazacyclic libraries.

Combinatorial libraries continue to play a key role in drug discovery. To increase structural diversity, several experimental methods have been developed. However, limited efforts have been performed so far to quantify the diversity of the broadly used diversity-oriented synthetic libraries. Herein, we report a comprehensive characterization of 15 bis-diazacyclic combinatorial libraries obtained through libraries from libraries, which is a diversity-oriented synthetic approach. Using MACCS keys, radial and different pharmacophoric fingerprints as well as six molecular properties, it was demonstrated the increased structural and property diversity of the libraries from libraries over the individual libraries. Comparison of the libraries to existing drugs, NCI diversity, and the Molecular Libraries Small Molecule Repository revealed the structural uniqueness of the combinatorial libraries (mean similarity <0.5 for any fingerprint representation). In particular, bis-cyclic thiourea libraries were the most structurally dissimilar to drugs retaining drug-like character in property space. This study represents the first comprehensive quantification of the diversity of libraries from libraries providing a solid quantitative approach to compare and contrast the diversity of diversity-oriented synthetic libraries with existing drugs or any other compound collection.

Analysis of nocturnal oxygen saturation recordings using kernel entropy to assist in sleep apnea-hypopnea diagnosis.

In this study, a new entropy measure known as kernel entropy (KerEnt), which quantifies the irregularity in a series, was applied to nocturnal oxygen saturation (SaO(2)) recordings. A total of 96 subjects suspected of suffering from sleep apnea-hypopnea syndrome (SAHS) took part in the study: 32 SAHS-negative and 64 SAHS-positive subjects. Their SaO(2) signals were separately processed by means of KerEnt. Our results show that a higher degree of irregularity is associated to SAHS-positive subjects. Statistical analysis revealed significant differences between the KerEnt values of SAHS-negative and SAHS-positive groups. The diagnostic utility of this parameter was studied by means of receiver operating characteristic (ROC) analysis. A classification accuracy of 81.25% (81.25% sensitivity and 81.25% specificity) was achieved. Repeated apneas during sleep increase irregularity in SaO(2) data. This effect can be measured by KerEnt in order to detect SAHS. This non-linear measure can provide useful information for the development of alternative diagnostic techniques in order to reduce the demand for conventional polysomnography (PSG).

Data visualization during the early stages of drug discovery.

Multidimensional compound optimization is a new paradigm in the drug discovery process, yielding efficiencies during early stages and reducing attrition in the later stages of drug development. The success of this strategy relies heavily on understanding this multidimensional data and extracting useful information from it. This paper demonstrates how principled visualization algorithms can be used to understand and explore a large data set created in the early stages of drug discovery. The experiments presented are performed on a real-world data set comprising biological activity data and some whole-molecular physicochemical properties. Data visualization is a popular way of presenting complex data in a simpler form. We have applied powerful principled visualization methods, such as generative topographic mapping (GTM) and hierarchical GTM (HGTM), to help the domain experts (screening scientists, chemists, biologists, etc.) understand and draw meaningful decisions. We also benchmark these principled methods against relatively better known visualization approaches, principal component analysis (PCA), Sammon's mapping, and self-organizing maps (SOMs), to demonstrate their enhanced power to help the user visualize the large multidimensional data sets one has to deal with during the early stages of the drug discovery process. The results reported clearly show that the GTM and HGTM algorithms allow the user to cluster active compounds for different targets and understand them better than the benchmarks. An interactive software tool supporting these visualization algorithms was provided to the domain experts. The tool facilitates the domain experts by exploration of the projection obtained from the visualization algorithms providing facilities such as parallel coordinate plots, magnification factors, directional curvatures, and integration with industry standard software.

Classifying ovarian tumors using Bayesian Multi-Layer Perceptrons and Automatic Relevance Determination: a multi-center study.

Ovarian masses are common and a good pre-surgical assessment of their nature is important for adequate treatment. Bayesian Multi-Layer Perceptrons (MLPs) using the evidence procedure were used to predict whether tumors are malignant or not. Automatic Relevance Determination (ARD) is used to select the most relevant of the 40+ available variables. Cross-validation is used to select an optimal combination of input set and number of hidden neurons. The data set consists of 1066 tumors collected at nine centers across Europe. Results indicate good performance of the models with AUC values of 0.93-0.94 on independent data. A comparison with a Bayesian perceptron model shows that the present problem is to a large extent linearly separable. The analyses further show that the number of hidden neurons specified in the ARD analyses for input selection may influence model performance.

MILVA: an interactive tool for the exploration of multidimensional microarray data.

MOTIVATION: Clustering techniques such as k-means and hierarchical clustering are commonly used to analyze DNA microarray derived gene expression data. However, the interactions between processes underlying the cell activity suggest that the complexity of the microarray data structure may not be fully represented with discrete clustering methods. RESULTS: A newly developed software tool called MILVA (microarray latent visualization and analysis) is presented here to investigate microarray data without separating gene expression profiles into discrete classes. The underpinning of the MILVA software is the two-dimensional topographic representation of multidimensional microarray data. On this basis, the interactive MILVA functions allow a continuous exploration of microarray data driven by the direct supervision of the biologist in detecting activity patterns of co-regulated genes. AVAILABILITY: The MILVA software is freely available. The software and the related documentation can be downloaded from http://www.ncrg.aston.ac.uk/Projects/milva. User 'surrey' as username and '3245' as password to login. The software is currently available for Windows platform only.

Nonlinear prediction of quantitative structure-activity relationships.

Predicting the log of the partition coefficient P is a long-standing benchmark problem in Quantitative Structure-Activity Relationships (QSAR). In this paper we show that a relatively simple molecular representation (using 14 variables) can be combined with leading edge machine learning algorithms to predict logP on new compounds more accurately than existing benchmark algorithms which use complex molecular representations.

Efficient training of RBF networks for classification.

Radial Basis Function networks with linear outputs are often used in regression problems because they can be substantially faster to train than Multi-layer Perceptrons. For classification problems, the use of linear outputs is less appropriate as the outputs are not guaranteed to represent probabilities. We show how RBFs with logistic and softmax outputs can be trained efficiently using the Fisher scoring algorithm. This approach can be used with any model which consists of a generalised linear output function applied to a model which is linear in its parameters. We compare this approach with standard non-linear optimisation algorithms on a number of datasets.

Outlier detection in scatterometer data: neural network approaches.

Satellite-borne scatterometers are used to measure backscattered micro-wave radiation from the ocean surface. This data may be used to infer surface wind vectors where no direct measurements exist. Inherent in this data are outliers owing to aberrations on the water surface and measurement errors within the equipment. We present two techniques for identifying outliers using neural networks; the outliers may then be removed to improve models derived from the data. Firstly the generative topographic mapping (GTM) is used to create a probability density model; data with low probability under the model may be classed as outliers. In the second part of the paper, a sensor model with input-dependent noise is used and outliers are identified based on their probability under this model.GTM was successfully modified to incorporate prior knowledge of the shape of the observation manifold; however, GTM could not learn the double skinned nature of the observation manifold. To learn this double skinned manifold necessitated the use of a sensor model which imposes strong constraints on the mapping. The results using GTM with a fixed noise level suggested the noise level may vary as a function of wind speed. This was confirmed by experiments using a sensor model with input-dependent noise, where the variation in noise is most sensitive to the wind speed input. Both models successfully identified gross outliers with the largest differences between models occurring at low wind speeds.