Integration of incomplete multi-omics data using Knowledge Distillation and Supervised Variational Autoencoders for disease progression prediction.

OBJECTIVE: The rapid advancement of high-throughput technologies in the biomedical field has resulted in the accumulation of diverse omics data types, such as mRNA expression, DNA methylation, and microRNA expression, for studying various diseases. Integrating these multi-omics datasets enables a comprehensive understanding of the molecular basis of cancer and facilitates accurate prediction of disease progression. METHODS: However, conventional approaches face challenges due to the dimensionality curse problem. This paper introduces a novel framework called Knowledge Distillation and Supervised Variational AutoEncoders utilizing View Correlation Discovery Network (KD-SVAE-VCDN) to address the integration of high-dimensional multi-omics data with limited common samples. Through our experimental evaluation, we demonstrate that the proposed KD-SVAE-VCDN architecture accurately predicts the progression of breast and kidney carcinoma by effectively classifying patients as long- or short-term survivors. Furthermore, our approach outperforms other state-of-the-art multi-omics integration models. RESULTS: Our findings highlight the efficacy of the KD-SVAE-VCDN architecture in predicting the disease progression of breast and kidney carcinoma. By enabling the classification of patients based on survival outcomes, our model contributes to personalized and targeted treatments. The favorable performance of our approach in comparison to several existing models suggests its potential to contribute to the advancement of cancer understanding and management. CONCLUSION: The development of a robust predictive model capable of accurately forecasting disease progression at the time of diagnosis holds immense promise for advancing personalized medicine. By leveraging multi-omics data integration, our proposed KD-SVAE-VCDN framework offers an effective solution to this challenge, paving the way for more precise and tailored treatment strategies for patients with different types of cancer.

Neonatal encephalopathy prediction of poor outcome with diffusion-weighted imaging connectome and fixel-based analysis.

BACKGROUND: Better biomarkers of eventual outcome are needed for neonatal encephalopathy. To identify the most potent neonatal imaging marker associated with 2-year outcomes, we retrospectively performed diffusion-weighted imaging connectome (DWIC) and fixel-based analysis (FBA) on magnetic resonance imaging (MRI) obtained in the first 4 weeks of life in term neonatal encephalopathy newborns. METHODS: Diffusion tractography was available in 15 out of 24 babies with MRI, five each with normal, abnormal motor outcome, or death. All 15 except one underwent hypothermia as initial treatment. In abnormal motor and death groups, DWIC found 19 white matter pathways with severely disrupted fiber orientation distributions. RESULTS: Using random forest classification, these disruptions predicted the follow-up outcomes with 89-99% accuracy. These pathways showed reduced integrity in abnormal motor and death vs. normal tone groups (p < 10-6). Using ranked supervised multi-view canonical correlation and depicting just three of the five dimensions of the analysis, the abnormal motor and death were clearly differentiated from each other and the normal tone group. CONCLUSIONS: This study suggests that a machine-learning model for prediction using early DWIC and FBA could be a possible way of developing biomarkers in large MRI datasets having clinical outcomes. IMPACT: Early connectome and FBA of clinically acquired DWI provide a new noninvasive imaging tool to predict the long-term motor outcomes after birth, based on the severity of white matter injury. Disrupted white matter connectivity as a novel neonatal marker achieves high accuracy of 89-99% to predict 2-year motor outcomes using conventional machine-learning classification. The proposed neonatal marker may allow better prognostication that is important to elucidate neural repair mechanisms and evaluate treatment modalities in neonatal encephalopathy.

A tree based approach for multi-class classification of surgical procedures using structured and unstructured data.

BACKGROUND: In surgical department, CPT code assignment has been a complicated manual human effort, that entails significant related knowledge and experience. While there are several studies using CPTs to make predictions in surgical services, literature on predicting CPTs in surgical and other services using text features is very sparse. This study improves the prediction of CPTs by the means of informative features and a novel re-prioritization algorithm. METHODS: The input data used in this study is composed of both structured and unstructured data. The ground truth labels (CPTs) are obtained from medical coding databases using relative value units which indicates the major operational procedures in each surgery case. In the modeling process, we first utilize Random Forest multi-class classification model to predict the CPT codes. Second, we extract the key information such as label probabilities, feature importance measures, and medical term frequency. Then, the indicated factors are used in a novel algorithm to rearrange the alternative CPT codes in the list of potential candidates based on the calculated weights. RESULTS: To evaluate the performance of both phases, prediction and complementary improvement, we report the accuracy scores of multi-class CPT prediction tasks for datasets of 5 key surgery case specialities. The Random Forest model performs the classification task with 74-76% when predicting the primary CPT (accuracy@1) versus the CPT set (accuracy@2) with respect to two filtering conditions on CPT codes. The complementary algorithm improves the results from initial step by 8% on average. Furthermore, the incorporated text features enhanced the quality of the output by 20-35%. The model outperforms the state-of-the-art neural network model with respect to accuracy, precision and recall. CONCLUSIONS: We have established a robust framework based on a decision tree predictive model. We predict the surgical codes more accurately and robust compared to the state-of-the-art deep neural structures which can help immensely in both surgery billing and scheduling purposes in such units.

Nanoscale imaging using differential expansion microscopy.

Expensive and time-consuming approaches of immunoelectron microscopy of biopsy tissues continues to serve as the gold-standard for diagnostic pathology. The recent development of the new approach of expansion microscopy (ExM) capable of fourfold lateral expansion of biological specimens for their morphological examination at approximately 70 nm lateral resolution using ordinary diffraction limited optical microscopy, is a major advancement in cellular imaging. Here we report (1) an optimized fixation protocol for retention of cellular morphology while obtaining optimal expansion, (2) an ExM procedure for up to eightfold lateral and over 500-fold volumetric expansion, (3) demonstrate that ExM is anisotropic or differential between tissues, cellular organelles and domains within organelles themselves, and (4) apply image analysis and machine learning (ML) approaches to precisely assess differentially expanded cellular structures. We refer to this enhanced ExM approach combined with ML as differential expansion microscopy (DiExM), applicable to profiling biological specimens at the nanometer scale. DiExM holds great promise for the precise, rapid and inexpensive diagnosis of disease from pathological specimen slides.

Unsupervised abnormality detection in neonatal MRI brain scans using deep learning.

Analysis of 3D medical imaging data has been a large topic of focus in the area of Machine Learning/Artificial Intelligence, though little work has been done in algorithmic (particularly unsupervised) analysis of neonatal brain MRI's. A myriad of conditions can manifest at an early age, including neonatal encephalopathy (NE), which can result in lifelong physical consequences. As such, there is a dire need for better biomarkers of NE and other conditions. The objective of the study is to improve identification of anomalies and prognostication of neonatal MRI brain scans. We introduce a framework designed to support the analysis and assessment of neonatal MRI brain scans, the results of which can be used as an aid to neuroradiologists. We explored the efficacy of the framework through iterations of several deep convolutional Autoencoder (AE) unsupervised modeling architectures designed to learn normalcy of the neonatal brain structure. We tested this framework on the developing human connectome project (dHCP) dataset with 97 patients that were previously categorized by severity. Our framework demonstrated the model's ability to identify and distinguish subtle morphological signatures present in brain structures. Normal and abnormal neonatal brain scans can be distinguished with reasonable accuracy, correctly categorizing them in up to 83% of cases. Most critically, new brain anomalies originally missed during the radiological reading were identified and corroborated by a neuroradiologist. This framework and our modeling approach demonstrate an ability to improve prognostication of neonatal brain conditions and are able to localize new anomalies.

Biomarker-driven drug repurposing on biologically similar cancers with DNA-repair deficiencies.

Similar molecular and genetic aberrations among diseases can lead to the discovery of jointly important treatment options across biologically similar diseases. Oncologists closely looked at several hormone-dependent cancers and identified remarkable pathological and molecular similarities in their DNA repair pathway abnormalities. Although deficiencies in Homologous Recombination (HR) pathway plays a significant role towards cancer progression, there could be other DNA-repair pathway deficiencies that requires careful investigation. In this paper, through a biomarker-driven drug repurposing model, we identified several potential drug candidates for breast and prostate cancer patients with DNA-repair deficiencies based on common specific biomarkers and irrespective of the organ the tumors originated from. Normalized discounted cumulative gain (NDCG) and sensitivity analysis were used to assess the performance of the drug repurposing model. Our results showed that Mitoxantrone and Genistein were among drugs with high therapeutic effects that significantly reverted the gene expression changes caused by the disease (FDR adjusted p-values for prostate cancer =1.225e-4 and 8.195e-8, respectively) for patients with deficiencies in their homologous recombination (HR) pathways. The proposed multi-cancer treatment framework, suitable for patients whose cancers had common specific biomarkers, has the potential to identify promising drug candidates by enriching the study population through the integration of multiple cancers and targeting patients who respond poorly to organ-specific treatments.

Identifying indicators influencing emergency department performance during a medical surge: A consensus-based modified fuzzy Delphi approach.

During a medical surge, resource scarcity and other factors influence the performance of the healthcare systems. To enhance their performance, hospitals need to identify the critical indicators that affect their operations for better decision-making. This study aims to model a pertinent set of indicators for improving emergency departments' (ED) performance during a medical surge. The framework comprises a three-stage process to survey, evaluate, and rank such indicators in a systematic approach. The first stage consists of a survey based on the literature and interviews to extract quality indicators that impact the EDs' performance. The second stage consists of forming a panel of medical professionals to complete the survey questionnaire and applying our proposed consensus-based modified fuzzy Delphi method, which integrates text mining to address the fuzziness and obtain the sentiment scores in expert responses. The final stage ranks the indicators based on their stability and convergence. Here, twenty-nine potential indicators are extracted in the first stage, categorized into five healthcare performance factors, are reduced to twenty consentaneous indicators monitoring ED's efficacy. The Mann-Whitney test confirmed the stability of the group opinions (p < 0.05). The agreement percentage indicates that ED beds (77.8%), nurse staffing per patient seen (77.3%), and length of stay (75.0%) are among the most significant indicators affecting the ED's performance when responding to a surge. This research proposes a framework that helps hospital administrators determine essential indicators to monitor, manage, and improve the performance of EDs systematically during a surge event.

Integration of Multimodal Data from Disparate Sources for Identifying Disease Subtypes.

Studies over the past decade have generated a wealth of molecular data that can be leveraged to better understand cancer risk, progression, and outcomes. However, understanding the progression risk and differentiating long- and short-term survivors cannot be achieved by analyzing data from a single modality due to the heterogeneity of disease. Using a scientifically developed and tested deep-learning approach that leverages aggregate information collected from multiple repositories with multiple modalities (e.g., mRNA, DNA Methylation, miRNA) could lead to a more accurate and robust prediction of disease progression. Here, we propose an autoencoder based multimodal data fusion system, in which a fusion encoder flexibly integrates collective information available through multiple studies with partially coupled data. Our results on a fully controlled simulation-based study have shown that inferring the missing data through the proposed data fusion pipeline allows a predictor that is superior to other baseline predictors with missing modalities. Results have further shown that short- and long-term survivors of glioblastoma multiforme, acute myeloid leukemia, and pancreatic adenocarcinoma can be successfully differentiated with an AUC of 0.94, 0.75, and 0.96, respectively.

Acute coronary syndrome prediction in emergency care: A machine learning approach.

BACKGROUND AND OBJECTIVE: Clinical concern for acute coronary syndrome (ACS) is one of emergency medicine's most common patient encounters. This study aims to develop an ensemble learning-driven framework as a diagnostic support tool to prevent misdiagnosis. METHODS: We obtained extensive clinical electronic health data on patient encounters with clinical concerns for ACS from a large urban emergency department (ED) between January 2017 and August 2020. We applied an analytical framework equipped with many well-developed algorithms to improve the data quality by addressing missing values, dimensionality reduction, and data imbalance. We trained ensemble learning algorithms to classify patients with ACS or non-ACS etiologies of their symptoms. We used performance evaluation metrics such as accuracy, sensitivity, precision, F1-score, and the area under the receiver operating characteristic (AUROC) to measure the model's performance. RESULTS: The analysis included 31,228 patients, of whom 563 (1.8%) had ACS and 30,665 (98.2%) had alternative diagnoses. Eleven features, including systolic blood pressure, brain natriuretic peptide, chronic heart disease, coronary artery disease, creatinine, glucose, heart attack, heart rate, nephrotic syndrome, red cell distribution width, and troponin level, are reported as significantly contributing risk factors. The proposed framework successfully classifies these cohorts with sensitivity and AUROC as high as 86.3% and 93.3%. Our proposed model's accuracy, precision, specificity, Matthew's correlation coefficient, and F1-score were 85.7%, 86.3%, 93%, 80%, and 86.3%, respectively. CONCLUSION: Our proposed framework can identify early patients with ACS through further refinement and validation.

A Comparison of Univariate and Multivariate Forecasting Models Predicting Emergency Department Patient Arrivals during the COVID-19 Pandemic.

The COVID-19 pandemic has heightened the existing concern about the uncertainty surrounding patient arrival and the overutilization of resources in emergency departments (EDs). The prediction of variations in patient arrivals is vital for managing limited healthcare resources and facilitating data-driven resource planning. The objective of this study was to forecast ED patient arrivals during a pandemic over different time horizons. A secondary objective was to compare the performance of different forecasting models in predicting ED patient arrivals. We included all ED patient encounters at an urban teaching hospital between January 2019 and December 2020. We divided the data into training and testing datasets and applied univariate and multivariable forecasting models to predict daily ED visits. The influence of COVID-19 lockdown and climatic factors were included in the multivariable models. The model evaluation consisted of the root mean square error (RMSE) and mean absolute error (MAE) over different forecasting horizons. Our exploratory analysis illustrated that monthly and weekly patterns impact daily demand for care. The Holt-Winters approach outperformed all other univariate and multivariable forecasting models for short-term predictions, while the Long Short-Term Memory approach performed best in extended predictions. The developed forecasting models are able to accurately predict ED patient arrivals and peaks during a surge when tested on two years of data from a high-volume urban ED. These short- and long-term prediction models can potentially enhance ED and hospital resource planning.

Discovery of primary prostate cancer biomarkers using cross cancer learning.

Prostate cancer (PCa), the second leading cause of cancer death in American men, is a relatively slow-growing malignancy with multiple early treatment options. Yet, a significant number of low-risk PCa patients are over-diagnosed and over-treated with significant and long-term quality of life effects. Further, there is ever increasing evidence of metastasis and higher mortality when hormone-sensitive or castration-resistant PCa tumors are treated indistinctively. Hence, the critical need is to discover clinically-relevant and actionable PCa biomarkers by better understanding the biology of PCa. In this paper, we have discovered novel biomarkers of PCa tumors through cross-cancer learning by leveraging the pathological and molecular similarities in the DNA repair pathways of ovarian, prostate, and breast cancer tumors. Cross-cancer disease learning enriches the study population and identifies genetic/phenotypic commonalities that are important across diseases with pathological and molecular similarities. Our results show that ADIRF, SLC2A5, C3orf86, HSPA1B are among the most significant PCa biomarkers, while MTRNR2L1, EEPD1, TEPP and VN1R2 are jointly important biomarkers across prostate, breast and ovarian cancers. Our validation results have further shown that the discovered biomarkers can predict the disease state better than any randomly selected subset of differentially expressed prostate cancer genes.

Automated Surgical Term Clustering: A Text Mining Approach for Unstructured Textual Surgery Descriptions.

High costs in health care and everlasting need for quality improvement in care delivery is increasingly becoming the motivating factor for novel predictive studies in health care informatics. Surgical services impact both the operating theatre costs and revenues and play critical role in care quality. Efficiency of such units relies extremely on effective operational planning and inventory management. A key ingredient to such planning activities is the structured and unstructured data available prior to the surgery day from the electronic health records and other information systems. Unstructured data, such as textual features of procedure description and notes, provide additional information while structured data alone is not sufficient. To effectively utilize textual information using text mining, textual features should be easily identifiable, i.e., without typographical errors and ad hoc abbreviations. While there exists numerous spelling correction and abbreviation identification tools, they are not suitable for the surgical medical text as they require a dictionary and cannot accommodate ad hoc words such as abbreviations. This study proposes a novel preprocessing framework for surgical text data to detect misspellings and abbreviations prior to the application of any text mining and predictive modeling. The proposed approach helps extract the most salient text features from the unstructured principal procedure and additional notes by effectively reducing the raw feature set dimension. The transformed (text) feature set thus improves subsequent prediction tasks in surgery units. We test and validate the proposed approach using datasets from multiple hospitals' surgical departments and benchmark feature sets.

Integrated Cancer Subtyping using Heterogeneous Genome-Scale Molecular Datasets.

Vast repositories of heterogeneous data from existing sources present unique opportunities. Taken individually, each of the datasets offers solutions to important domain and source-specific questions. Collectively, they represent complementary views of related data entities with an aggregate information value often well exceeding the sum of its parts. Integration of heterogeneous data is therefore paramount to i) obtain a more unified picture and comprehensive view of the relations, ii) achieve more robust results, iii) improve the accuracy and integrity, and iv) illuminate the complex interactions among data features. In this paper, we have proposed a data integration methodology to identify subtypes of cancer using multiple data types (mRNA, methylation, microRNA and somatic variants) and different data scales that come from different platforms (microarray, sequencing, etc.). The Cancer Genome Atlas (TCGA) dataset is used to build the data integration and cancer subtyping framework. The proposed data integration and disease subtyping approach accurately identifies novel subgroups of patients with significantly different survival profiles. With current availability of vast genomics, and variant data for cancer, the proposed data integration system will better differentiate cancer and patient subtypes for risk and outcome prediction and targeted treatment planning without additional cost and precious lost time.

Human skeletal muscle cell atlas: Unraveling cellular secrets utilizing 'muscle-on-a-chip', differential expansion microscopy, mass spectrometry, nanothermometry and machine learning.

The 'Human Cell Atlas' project has been launched to obtain a comprehensive understanding of all cell types, the fundamental living units that constitute the human body. This is a global partnership and effort involving experts from many disciplines, from computer science, engineering to medicine, and is supported by several private and public organizations, among them, the Chan Zuckerberg Foundation, the National Institutes of Health, and Google, that will greatly benefit humanity. Nearly 37 trillion cells of various shapes, sizes, and composition, are precisely organized to constitute the human body. Humans, like all other living organisms, are dynamic, and therefore a comprehensive understanding of different cells in their various dynamic states is required to provide a reference map for the early diagnosis and various preventive approach to disease, and in the development of precision therapeutics. Skeletal muscles being the most abundant tissue and the largest locomotor and metabolic organ in the human body, requires a global understanding of its structure, composition, and function. The objective of creating a 'Human Skeletal Muscle Cell Atlas', necessitates therefore a comprehensive understanding of the emergent properties of skeletal muscle cell growth, development, structure, function and chemistry, under conditions of activity and inactivity. To achieve this objective would require a very precise yet rapid and cost-effective approach of combined multimodal imaging, including our new and novel 'Differential Expansion Microscopy', our 'Nanoscale Thermometry', combined with 'Mass Spectrometry', 'Motor Protein Motility Assay' and 'Machine Learning' tools.

Continence Index: a new screening questionnaire to predict the probability of future incontinence in older women in the community.

PURPOSE: Urinary incontinence (UI) is a chronic, costly condition that impairs quality of life. To identify older women most at risk, the Medical Epidemiologic and Social Aspects of Aging (MESA) datasets were mined to create a set of questions that can reliably predict future UI. METHODS: MESA data were collected during four household interviews at approximately 1 year intervals. Factors associated with becoming incontinent at the second interview (HH2) were identified using logistic regression (construction datasets). Based on p values and odds ratios, eight potential predictive factors with their 256 combinations and corresponding prediction probabilities formed the Continence Index. Its predictive and discriminatory capability was tested against the same cohort's outcome in the fourth survey (HH4 validation datasets). Sensitivity analysis, area under receiver operating characteristic (ROC) curve, predicted probabilities and confidence intervals were used to statistically validate the Continence Index. RESULTS: Body mass index, sneezing, post-partum UI, urinary frequency, mild UI, belief of developing UI in the future, difficulty stopping urinary stream and remembering names emerged as the strongest predictors of UI. The confidence intervals for prediction probabilities strongly agreed between construction and validation datasets. Calculated sensitivity, specificity, false-positive and false-negative values revealed that the areas under the ROCs (0.802 and 0.799) for the construction and validation datasets, respectively, indicated good discriminatory capabilities of the index as a predictor. CONCLUSION: The Continence Index will help identify older women most at risk of UI in order to apply targeted prevention strategies in women that are most likely to benefit.

Novel application of statistical methods to identify new urinary incontinence risk factors.

Longitudinal data for studying urinary incontinence (UI) risk factors are rare. Data from one study, the hallmark Medical, Epidemiological, and Social Aspects of Aging (MESA), have been analyzed in the past; however, repeated measures analyses that are crucial for analyzing longitudinal data have not been applied. We tested a novel application of statistical methods to identify UI risk factors in older women. MESA data were collected at baseline and yearly from a sample of 1955 men and women in the community. Only women responding to the 762 baseline and 559 follow-up questions at one year in each respective survey were examined. To test their utility in mining large data sets, and as a preliminary step to creating a predictive index for developing UI, logistic regression, generalized estimating equations (GEEs), and proportional hazard regression (PHREG) methods were used on the existing MESA data. The GEE and PHREG combination identified 15 significant risk factors associated with developing UI out of which six of them, namely, urinary frequency, urgency, any urine loss, urine loss after emptying, subject's anticipation, and doctor's proactivity, are found most highly significant by both methods. These six factors are potential candidates for constructing a future UI predictive index.