A Multimorbidity Analysis of Hospitalized Patients With COVID-19 in Northwest Italy: Longitudinal Study Using Evolutionary Machine Learning and Health Administrative Data.

BACKGROUND: Multimorbidity is a significant public health concern, characterized by the coexistence and interaction of multiple preexisting medical conditions. This complex condition has been associated with an increased risk of COVID-19. Individuals with multimorbidity who contract COVID-19 often face a significant reduction in life expectancy. The postpandemic period has also highlighted an increase in frailty, emphasizing the importance of integrating existing multimorbidity details into epidemiological risk assessments. Managing clinical data that include medical histories presents significant challenges, particularly due to the sparsity of data arising from the rarity of multimorbidity conditions. Also, the complex enumeration of combinatorial multimorbidity features introduces challenges associated with combinatorial explosions. OBJECTIVE: This study aims to assess the severity of COVID-19 in individuals with multiple medical conditions, considering their demographic characteristics such as age and sex. We propose an evolutionary machine learning model designed to handle sparsity, analyzing preexisting multimorbidity profiles of patients hospitalized with COVID-19 based on their medical history. Our objective is to identify the optimal set of multimorbidity feature combinations strongly associated with COVID-19 severity. We also apply the Apriori algorithm to these evolutionarily derived predictive feature combinations to identify those with high support. METHODS: We used data from 3 administrative sources in Piedmont, Italy, involving 12,793 individuals aged 45-74 years who tested positive for COVID-19 between February and May 2020. From their 5-year pre-COVID-19 medical histories, we extracted multimorbidity features, including drug prescriptions, disease diagnoses, sex, and age. Focusing on COVID-19 hospitalization, we segmented the data into 4 cohorts based on age and sex. Addressing data imbalance through random resampling, we compared various machine learning algorithms to identify the optimal classification model for our evolutionary approach. Using 5-fold cross-validation, we evaluated each model's performance. Our evolutionary algorithm, utilizing a deep learning classifier, generated prediction-based fitness scores to pinpoint multimorbidity combinations associated with COVID-19 hospitalization risk. Eventually, the Apriori algorithm was applied to identify frequent combinations with high support. RESULTS: We identified multimorbidity predictors associated with COVID-19 hospitalization, indicating more severe COVID-19 outcomes. Frequently occurring morbidity features in the final evolved combinations were age>53, R03BA (glucocorticoid inhalants), and N03AX (other antiepileptics) in cohort 1; A10BA (biguanide or metformin) and N02BE (anilides) in cohort 2; N02AX (other opioids) and M04AA (preparations inhibiting uric acid production) in cohort 3; and G04CA (Alpha-adrenoreceptor antagonists) in cohort 4. CONCLUSIONS: When combined with other multimorbidity features, even less prevalent medical conditions show associations with the outcome. This study provides insights beyond COVID-19, demonstrating how repurposed administrative data can be adapted and contribute to enhanced risk assessment for vulnerable populations.

Multimorbidity in middle-aged women and COVID-19: binary data clustering for unsupervised binning of rare multimorbidity features and predictive modeling.

BACKGROUND: Multimorbidity is typically associated with deficient health-related quality of life in mid-life, and the likelihood of developing multimorbidity in women is elevated. We address the issue of data sparsity in non-prevalent features by clustering the binary data of various rare medical conditions in a cohort of middle-aged women. This study aims to enhance understanding of how multimorbidity affects COVID-19 severity by clustering rare medical conditions and combining them with prevalent features for predictive modeling. The insights gained can guide the development of targeted interventions and improved management strategies for individuals with multiple health conditions. METHODS: The study focuses on a cohort of 4477 female patients, (aged 45-60) in Piedmont, Italy, and utilizes their multimorbidity data prior to the COVID-19 pandemic from their medical history from 2015 to 2019. The COVID-19 severity is determined by the hospitalization status of the patients from February to May 2020. Each patient profile in the dataset is depicted as a binary vector, where each feature denotes the presence or absence of a specific multimorbidity condition. By clustering the sparse medical data, newly engineered features are generated as a bin of features, and they are combined with the prevalent features for COVID-19 severity predictive modeling. RESULTS: From sparse data consisting of 174 input features, we have created a low-dimensional feature matrix of 17 features. Machine Learning algorithms are applied to the reduced sparsity-free data to predict the Covid-19 hospital admission outcome. The performance obtained for the corresponding models are as follows: Logistic Regression (accuracy 0.72, AUC 0.77, F1-score 0.69), Linear Discriminant Analysis (accuracy 0.7, AUC 0.77, F1-score 0.67), and Ada Boost (accuracy 0.7, AUC 0.77, F1-score 0.68). CONCLUSION: Mapping higher-dimensional data to a low-dimensional space can result in information loss, but reducing sparsity can be beneficial for Machine Learning modeling due to improved predictive ability. In this study, we addressed the issue of data sparsity in electronic health records and created a model that incorporates both prevalent and rare medical conditions, leading to more accurate and effective predictive modeling. The identification of complex associations between multimorbidity and the severity of COVID-19 highlights potential areas of focus for future research, including long COVID and intervention efforts.

Bias correction in random effects models with sparse binary responses.

Sparse correlated binary data are frequently encountered in many applications involving either rare event cases or small sample sizes. In this study, we consider correlated binary data and a logit random effects model framework. We discuss h-likelihood estimates and how the computational procedure is affected by sparseness. We propose an adjustment to the fitting process that involves the adaption of the regression calibration method to the estimation of random effects. Using this adjustment, we correct for the bias in the random effects estimates resulting in better properties for the fixed effects estimates of the model. This is supported by the results of the simulation study that was conducted under different sparseness levels. The proposed adjusted h-likelihood estimation approach is also used for the analysis of two real meta-analyses data sets.

Efficient feature extraction from highly sparse binary genotype data for cancer prognosis prediction using an auto-encoder.

Genomics involving tens of thousands of genes is a complex system determining phenotype. An interesting and vital issue is how to integrate highly sparse genetic genomics data with a mass of minor effects into a prediction model for improving prediction power. We find that the deep learning method can work well to extract features by transforming highly sparse dichotomous data to lower-dimensional continuous data in a non-linear way. This may provide benefits in risk prediction-associated genotype data. We developed a multi-stage strategy to extract information from highly sparse binary genotype data and applied it for cancer prognosis. Specifically, we first reduced the size of binary biomarkers via a univariable regression model to a moderate size. Then, a trainable auto-encoder was used to learn compact features from the reduced data. Next, we performed a LASSO problem process to select the optimal combination of extracted features. Lastly, we applied such feature combination to real cancer prognostic models and evaluated the raw predictive effect of the models. The results indicated that these compressed transformation features could better improve the model's original predictive performance and might avoid an overfitting problem. This idea may be enlightening for everyone involved in cancer research, risk reduction, treatment, and patient care via integrating genomics data.

The "DOLPHINS" Project: A Low-Cost Real-Time Multivariate Process Control From Large Sensor Arrays Providing Sparse Binary Data.

The "DOLPHINS" project started in 2018 under a collaboration between three partners: CNH Industrial Iveco (CHNi), RADA (an informatics company), and the Chemistry Department of the University of Turin. The project's main aim was to establish a predictive maintenance method in real-time at a pilot plant (CNHi Iveco, Brescia, Italy). This project currently allows maintenance technicians to intervene on machinery preventively, avoiding breakdowns or stops in the production process. For this purpose, several predictive maintenance models were tested starting from databases on programmable logic controllers (PLCs) already available, thus taking advantage of Machine Learning techniques without investing additional resources in purchasing or installing new sensors. The instrumentation and PLCs related to the truck sides' paneling phase were considered at the beginning of the project. The instrumentation under evaluation was equipped with sensors already connected to PLCs (only on/off switches, i.e., neither analog sensors nor continuous measurements are available, and the data are in sparse binary format) so that the data provided by PLCs were acquired in a binary way before being processed by multivariate data analysis (MDA) models. Several MDA approaches were tested (e.g., PCA, PLS-DA, SVM, XGBoost, and SIMCA) and validated in the plant (in terms of repeated double cross-validation strategies). The optimal approach currently used involves combining PCA and SIMCA models, whose performances are continuously monitored, and the various models are updated and tested weekly. Tuning the time range predictions enabled the shop floor and the maintenance operators to achieve sensitivity and specificity values higher than 90%, but the performance results are constantly improved since new data are collected daily. Furthermore, the information on where to carry out intervention is provided to the maintenance technicians between 30 min and 3 h before the breakdown.

An integrative shrinkage estimator for random-effects meta-analysis of rare binary events.

Meta-analysis has been a powerful tool for inferring the treatment effect between two experimental conditions from multiple studies of rare binary events. Recently, under a random-effects (RE) model, Bhaumik et al. developed a simple average (SA) estimator and showed that with the continuity correction factor 0.5, the SA estimator was the least biased among a set of commonly used estimators. In this paper, under various RE models that allow for treatment groups with equal and unequal variability (in either direction), we develop an integrative shrinkage (iSHRI) estimator based on the SA estimator, which aims to improve estimation efficiency in terms of mean squared error (MSE) that accounts for the bias-variance tradeoff. Through simulation, we find that iSHRI has better performance in general when compared with existing methods, in terms of bias, MSE, type I error and confidence interval coverage. Data examples of rosiglitazone meta-analysis are provided as well, where iSHRI yields competitive results.