In Vitro Glucose Measurement from NIR and MIR Spectroscopy: Comprehensive Benchmark of Machine Learning and Filtering Chemometrics.

The quantitative analysis of glucose using spectroscopy is a topic of great significance and interest in science and industry. One conundrum in this area is deploying appropriate preprocessing and regression tools. To contribute to addressing this challenge, in this study, we conducted a comprehensive and novel comparative analysis of various machine learning and preprocessing filtering techniques applied to near-infrared, mid-infrared, and a combination of near-infrared and mid-infrared spectroscopy for glucose assay. Our objective was to evaluate the effectiveness of these techniques in accurately predicting glucose levels and to determine which approach was most optimal. Our investigation involved the acquisition of spectral data from samples of glucose solutions using the three aforementioned spectroscopy techniques. The data was subjected to several preprocessing filtering methods, including convolutional moving average, Savitzky-Golay, multiplicative scatter correction, and normalisation. We then applied representative machine learning algorithms from three categories: linear modelling, traditional nonlinear modelling, and artificial neural networks. The evaluation results revealed that linear models exhibited higher predictive accuracy than nonlinear models, whereas artificial neural network models demonstrated comparable performance. Additionally, the comparative analysis of various filtering methods demonstrated that the convolutional moving average and Savitzky-Golay filters yielded the most precise outcomes overall. In conclusion, our study provides valuable insights into the efficacy of different machine learning techniques for glucose measurement and highlights the importance of applying appropriate filtering methods in enhancing predictive accuracy. These findings have important implications for the development of new and improved glucose quantification technologies.

Blood Glucose Level Time Series Forecasting: Nested Deep Ensemble Learning Lag Fusion.

Blood glucose level prediction is a critical aspect of diabetes management. It enables individuals to make informed decisions about their insulin dosing, diet, and physical activity. This, in turn, improves their quality of life and reduces the risk of chronic and acute complications. One conundrum in developing time-series forecasting models for blood glucose level prediction is to determine an appropriate length for look-back windows. On the one hand, studying short histories foists the risk of information incompletion. On the other hand, analysing long histories might induce information redundancy due to the data shift phenomenon. Additionally, optimal lag lengths are inconsistent across individuals because of the domain shift occurrence. Therefore, in bespoke analysis, either optimal lag values should be found for each individual separately or a globally suboptimal lag value should be used for all. The former approach degenerates the analysis's congruency and imposes extra perplexity. With the latter, the fine-tunned lag is not necessarily the optimum option for all individuals. To cope with this challenge, this work suggests an interconnected lag fusion framework based on nested meta-learning analysis that improves the accuracy and precision of predictions for personalised blood glucose level forecasting. The proposed framework is leveraged to generate blood glucose prediction models for patients with type 1 diabetes by scrutinising two well-established publicly available Ohio type 1 diabetes datasets. The models developed undergo vigorous evaluation and statistical analysis from mathematical and clinical perspectives. The results achieved underpin the efficacy of the proposed method in blood glucose level time-series prediction analysis.

Causality analysis in type 1 diabetes mellitus with application to blood glucose level prediction.

Effective control of blood glucose level (BGL) is the key factor in the management of type 1 diabetes mellitus (T1D). BGL prediction is an important tool to help maximise the time BGL is in the target range and thus minimise both acute and chronic diabetes-related complications. To predict future BGL, histories of variables known to affect BGL, such as carbohydrate intake, injected bolus insulin, and physical activity, are utilised. Due to these identified cause and effect relationships, T1D management can be examined via the causality context. In this respect, this work initially investigates these relations and quantifies the causality strengths of each variable with BGL using the convergent cross mapping method (CCM). Then, considering the extended CCM, the causality strengths of each variable for different lags are quantified. After that, the optimal time lag for each variable is determined according to the quantified causality effects. Subsequently, the feasibility of leveraging causality information as prior knowledge for BGL prediction is investigated by proposing two approaches. In the first approach, causality strengths are used as weights for relevant affecting variables. In the second approach, the optimal causal lags and the corresponding causality strengths are considered the shifts and weights for the variables, respectively. Overall, the evaluation criteria and statistical analysis used for comparing results show the effectiveness of using causality analysis in T1D management.

Interpretable Machine Learning for Inpatient COVID-19 Mortality Risk Assessments: Diabetes Mellitus Exclusive Interplay.

People with diabetes mellitus (DM) are at elevated risk of in-hospital mortality from coronavirus disease-2019 (COVID-19). This vulnerability has spurred efforts to pinpoint distinctive characteristics of COVID-19 patients with DM. In this context, the present article develops ML models equipped with interpretation modules for inpatient mortality risk assessments of COVID-19 patients with DM. To this end, a cohort of 156 hospitalised COVID-19 patients with pre-existing DM is studied. For creating risk assessment platforms, this work explores a pool of historical, on-admission, and during-admission data that are DM-related or, according to preliminary investigations, are exclusively attributed to the COVID-19 susceptibility of DM patients. First, a set of careful pre-modelling steps are executed on the clinical data, including cleaning, pre-processing, subdivision, and feature elimination. Subsequently, standard machine learning (ML) modelling analysis is performed on the cured data. Initially, a classifier is tasked with forecasting COVID-19 fatality from selected features. The model undergoes thorough evaluation analysis. The results achieved substantiate the efficacy of the undertaken data curation and modelling steps. Afterwards, SHapley Additive exPlanations (SHAP) technique is assigned to interpret the generated mortality risk prediction model by rating the predictors' global and local influence on the model's outputs. These interpretations advance the comprehensibility of the analysis by explaining the formation of outcomes and, in this way, foster the adoption of the proposed methodologies. Next, a clustering algorithm demarcates patients into four separate groups based on their SHAP values, providing a practical risk stratification method. Finally, a re-evaluation analysis is performed to verify the robustness of the proposed framework.

COVID-19 mortality risk assessments for individuals with and without diabetes mellitus: Machine learning models integrated with interpretation framework.

This research develops machine learning models equipped with interpretation modules for mortality risk prediction and stratification in cohorts of hospitalised coronavirus disease-2019 (COVID-19) patients with and without diabetes mellitus (DM). To this end, routinely collected clinical data from 156 COVID-19 patients with DM and 349 COVID-19 patients without DM were scrutinised. First, a random forest classifier forecasted in-hospital COVID-19 fatality utilising admission data for each cohort. For the DM cohort, the model predicted mortality risk with the accuracy of 82%, area under the receiver operating characteristic curve (AUC) of 80%, sensitivity of 80%, and specificity of 56%. For the non-DM cohort, the achieved accuracy, AUC, sensitivity, and specificity were 80%, 84%, 91%, and 56%, respectively. The models were then interpreted using SHapley Additive exPlanations (SHAP), which explained predictors' global and local influences on model outputs. Finally, the k-means algorithm was applied to cluster patients on their SHAP values. The algorithm demarcated patients into three clusters. Average mortality rates within the generated clusters were 8%, 20%, and 76% for the DM cohort, 2.7%, 28%, and 41.9% for the non-DM cohort, providing a functional method of risk stratification.

Signal fragmentation based feature vector generation in a model agnostic framework with application to glucose quantification using absorption spectroscopy.

This paper proposes feature vector generation based on signal fragmentation equipped with a model interpretation module to enhance glucose quantification from absorption spectroscopy signals. For this purpose, near-infrared (NIR) and mid-infrared (MIR) spectra collected from experimental samples of varying glucose concentrations are scrutinised. Initially, a given spectrum is optimally dissected into several fragments. A base-learner then studies the obtained fragments individually to estimate the reference glucose concentration from each fragment. Subsequently, the resultant estimates from all fragments are stacked, forming a feature vector for the original spectrum. Afterwards, a meta-learner studies the generated feature vector to yield a final estimation of the reference glucose concentration pertaining to the entire original spectrum. The reliability of the proposed approach is reviewed under a set of circumstances encompassing modelling upon NIR or MIR signals alone and combinations of NIR and MIR signals at different fusion levels. In addition, the compatibility of the proposed approach with an underlying preprocessing technique in spectroscopy is assessed. The results obtained substantiate the utility of incorporating the designed feature vector generator into standard benchmarked modelling procedures under all considered scenarios. Finally, to promote the transparency and adoption of the propositions, SHapley additive exPlanations (SHAP) is leveraged to interpret the quantification outcomes.

Blood Glucose Level Prediction: Advanced Deep-Ensemble Learning Approach.

Optimal and sustainable control of blood glucose levels (BGLs) is the aim of type-1 diabetes management. The automated prediction of BGL using machine learning (ML) algorithms is considered as a promising tool that can support this aim. In this context, this paper proposes new advanced ML architectures to predict BGL leveraging deep learning and ensemble learning. The deep-ensemble models are developed with novel meta-learning approaches, where the feasibility of changing the dimension of a univariate time series forecasting task is investigated. The models are evaluated regression-wise and clinical-wise. The performance of the proposed ensemble models are compared with benchmark non-ensemble models. The results show the superior performance of the developed ensemble models over developed non-ensemble benchmark models and also show the efficacy of the proposed meta-learning approaches.

Classification before regression for improving the accuracy of glucose quantification using absorption spectroscopy.

This work contributes to the improvement of glucose quantification using near-infrared (NIR), mid-infrared (MIR), and combination of NIR and MIR absorbance spectroscopy by classifying the spectral data prior to the application of regression models. Both manual and automated classification are presented based on three homogeneous classes defined following the clinical definition of the glycaemic ranges (hypoglycaemia, euglycaemia, and hyperglycaemia). For the manual classification, partial least squares and principal component regressions are applied to each class separately and shown to lead to improved quantification results compared to when applying the same regression models for the whole dataset. For the automatic classification, linear discriminant analysis coupled with principal component analysis is deployed, and regressions are applied to each class separately. The results obtained are shown to outperform those of regressions for the entire dataset.

Classification of breast lesions in ultrasonography using sparse logistic regression and morphology-based texture features.

PURPOSE: This work proposes a new reliable computer-aided diagnostic (CAD) system for the diagnosis of breast cancer from breast ultrasound (BUS) images. The system can be useful to reduce the number of biopsies and pathological tests, which are invasive, costly, and often unnecessary. METHODS: The proposed CAD system classifies breast tumors into benign and malignant classes using morphological and textural features extracted from breast ultrasound (BUS) images. The images are first preprocessed to enhance the edges and filter the speckles. The tumor is then segmented semiautomatically using the watershed method. Having the tumor contour, a set of 855 features including 21 shape-based, 810 contour-based, and 24 textural features are extracted from each tumor. Then, a Bayesian Automatic Relevance Detection (ARD) mechanism is used for computing the discrimination power of different features and dimensionality reduction. Finally, a logistic regression classifier computed the posterior probabilities of malignant vs benign tumors using the reduced set of features. RESULTS: A dataset of 104 BUS images of breast tumors, including 72 benign and 32 malignant tumors, was used for evaluation using an eightfold cross-validation. The algorithm outperformed six state-of-the-art methods for BUS image classification with large margins by achieving 97.12% accuracy, 93.75% sensitivity, and 98.61% specificity rates. CONCLUSIONS: Using ARD, the proposed CAD system selects five new features for breast tumor classification and outperforms state-of-the-art, making a reliable and complementary tool to help clinicians diagnose breast cancer.