Applying Naive Bayesian Networks to Disease Prediction: a Systematic Review

Machine learning-based prediction of in-hospital mortality for critically ill patients with sepsis-associated acute kidney injury

OBJECTIVES

This study aims to develop and validate a prediction model in-hospital mortality in critically ill patients with sepsis-associated acute kidney injury (SA-AKI) based on machine learning algorithms.

METHODS

Patients who met the criteria for inclusion were identified in the Medical Information Mart for Intensive Care-IV (MIMIC-IV) database and divided according to the validation (n = 2440) and development (n = 9756, 80%) queues. Ensemble stepwise feature selection method was used to screen for effective features. The prediction models of short-term mortality were developed by seven machine learning algorithms. Ten-fold cross-validation was used to verify the performance of the algorithm in the development queue. The area under the receiver operating characteristic curve (ROC-AUC) was used to evaluate the differentiation accuracy and performance of the prediction model in the validation queue. The best-performing model was interpreted by Shapley additive explanations (SHAP).

RESULTS

A total of 12,196 patients were enrolled in this study. Eleven variables were finally chosen to develop the prediction model. The AUC of the random forest (RF) model was the highest value both in the Ten-fold cross-validation and evaluation (AUC: 0.798, 95% CI: 0.774-0.821). According to the SHAP plots, old age, low Glasgow Coma Scale (GCS) score, high AKI stage, reduced urine output, high Simplified Acute Physiology Score (SAPS II), high respiratory rate, low temperature, low absolute lymphocyte count, high creatinine level, dysnatremia, and low body mass index (BMI) increased the risk of poor prognosis.

CONCLUSIONS

The RF model developed in this study is a good predictor of in-hospital mortality for patients with SA-AKI in the intensive care unit (ICU), which may have potential applications in mortality prediction.

Artificial intelligence in the diagnosis and treatment of acute appendicitis: a narrative review

Artificial intelligence is transforming healthcare. Artificial intelligence can improve patient care by analyzing large amounts of data to help make more informed decisions regarding treatments and enhance medical research through analyzing and interpreting data from clinical trials and research projects to identify subtle but meaningful trends beyond ordinary perception. Artificial intelligence refers to the simulation of human intelligence in computers, where systems of artificial intelligence can perform tasks that require human-like intelligence like speech recognition, visual perception, pattern-recognition, decision-making, and language processing. Artificial intelligence has several subdivisions, including machine learning, natural language processing, computer vision, and robotics. By automating specific routine tasks, artificial intelligence can improve healthcare efficiency. By leveraging machine learning algorithms, the systems of artificial intelligence can offer new opportunities for enhancing both the efficiency and effectiveness of surgical procedures, particularly regarding training of minimally invasive surgery. As artificial intelligence continues to advance, it is likely to play an increasingly significant role in the field of surgical learning. Physicians have assisted to a spreading role of artificial intelligence in the last decade. This involved different medical specialties such as ophthalmology, cardiology, urology, but also abdominal surgery. In addition to improvements in diagnosis, ascertainment of efficacy of treatment and autonomous actions, artificial intelligence has the potential to improve surgeons' ability to better decide if acute surgery is indicated or not. The role of artificial intelligence in the emergency departments has also been investigated. We considered one of the most common condition the emergency surgeons have to face, acute appendicitis, to assess the state of the art of artificial intelligence in this frequent acute disease. The role of artificial intelligence in diagnosis and treatment of acute appendicitis will be discussed in this narrative review.

Composition and micromechanical properties of the femoral neck compact bone in relation to patient age, sex and hip fracture occurrence

Current clinical methods of bone health assessment depend to a great extent on bone mineral density (BMD) measurements. However, these methods only act as a proxy for bone strength and are often only carried out after the fracture occurs. Besides BMD, composition and tissue-level mechanical properties are expected to affect the whole bone's strength and toughness. While the elastic properties of the bone extracellular matrix (ECM) have been extensively investigated over the past two decades, there is still limited knowledge of the yield properties and their relationship to composition and architecture. In the present study, morphological, compositional and micropillar compression bone data was collected from patients who underwent hip arthroplasty. Femoral neck samples from 42 patients were collected together with anonymous clinical information about age, sex and primary diagnosis (coxarthrosis or hip fracture). The femoral neck cortex from the inferomedial region was analyzed in a site-matched manner using a combination of micromechanical testing (nanoindentation, micropillar compression) together with micro-CT and quantitative polarized Raman spectroscopy for both morphological and compositional characterization. Mechanical properties, as well as the sample-level mineral density, were constant over age. Only compositional properties demonstrate weak dependence on patient age: decreasing mineral to matrix ratio (p = 0.02, R2 = 0.13, 2.6 % per decade) and increasing amide I sub-peak ratio I∼1660/I∼1683 (p = 0.04, R2 = 0.11, 1.5 % per decade). The patient's sex and diagnosis did not seem to influence investigated bone properties. A clear zonal dependence between interstitial and osteonal cortical zones was observed for compositional and elastic bone properties (p < 0.0001). Site-matched microscale analysis confirmed that all investigated mechanical properties except yield strain demonstrate a positive correlation with the mineral fraction of bone. The output database is the first to integrate the experimentally assessed microscale yield properties, local tissue composition and morphology with the available patient clinical information. The final dataset was used for bone fracture risk prediction in-silico through the principal component analysis and the Naïve Bayes classification algorithm. The analysis showed that the mineral to matrix ratio, indentation hardness and micropillar yield stress are the most relevant parameters for bone fracture risk prediction at 70 % model accuracy (0.71 AUC). Due to the low number of samples, further studies to build a universal fracture prediction algorithm are anticipated with the higher number of patients (N > 200). The proposed classification algorithm together with the output dataset of bone tissue properties can be used for the future comparison of existing methods to evaluate bone quality as well as to form a better understanding of the mechanisms through which bone tissue is affected by aging or disease.

Disulfidptosis-associated lncRNAs predict breast cancer subtypes

Disulfidptosis is a newly discovered mode of cell death. However, its relationship with breast cancer subtypes remains unclear. In this study, we aimed to construct a disulfidptosis-associated breast cancer subtype prediction model. We obtained 19 disulfidptosis-related genes from published articles and performed correlation analysis with lncRNAs differentially expressed in breast cancer. We then used the random forest algorithm to select important lncRNAs and establish a breast cancer subtype prediction model. We identified 132 lncRNAs significantly associated with disulfidptosis (FDR < 0.01, |R|> 0.15) and selected the first four important lncRNAs to build a prediction model (training set AUC = 0.992). The model accurately predicted breast cancer subtypes (test set AUC = 0.842). Among the key lncRNAs, LINC02188 had the highest expression in the Basal subtype, while LINC01488 and GATA3-AS1 had the lowest expression in Basal. In the Her2 subtype, LINC00511 had the highest expression level compared to other key lncRNAs. GATA3-AS1 had the highest expression in LumA and LumB subtypes, while LINC00511 had the lowest expression in these subtypes. In the Normal subtype, GATA3-AS1 had the highest expression level compared to other key lncRNAs. Our study also found that key lncRNAs were closely related to RNA methylation modification and angiogenesis (FDR < 0.05, |R|> 0.1), as well as immune infiltrating cells (P.adj < 0.01, |R|> 0.1). Our random forest model based on disulfidptosis-related lncRNAs can accurately predict breast cancer subtypes and provide a new direction for research on clinical therapeutic targets for breast cancer.

A Bayesian latent class extension of naive Bayesian classifier and its application to the classification of gastric cancer patients

BACKGROUND

The Naive Bayes (NB) classifier is a powerful supervised algorithm widely used in Machine Learning (ML). However, its effectiveness relies on a strict assumption of conditional independence, which is often violated in real-world scenarios. To address this limitation, various studies have explored extensions of NB that tackle the issue of non-conditional independence in the data. These approaches can be broadly categorized into two main categories: feature selection and structure expansion. In this particular study, we propose a novel approach to enhancing NB by introducing a latent variable as the parent of the attributes. We define this latent variable using a flexible technique called Bayesian Latent Class Analysis (BLCA). As a result, our final model combines the strengths of NB and BLCA, giving rise to what we refer to as NB-BLCA. By incorporating the latent variable, we aim to capture complex dependencies among the attributes and improve the overall performance of the classifier.

METHODS

Both Expectation-Maximization (EM) algorithm and the Gibbs sampling approach were offered for parameter learning. A simulation study was conducted to evaluate the classification of the model in comparison with the ordinary NB model. In addition, real-world data related to 976 Gastric Cancer (GC) and 1189 Non-ulcer dyspepsia (NUD) patients was used to show the model's performance in an actual application. The validity of models was evaluated using the 10-fold cross-validation.

RESULTS

The presented model was superior to ordinary NB in all the simulation scenarios according to higher classification sensitivity and specificity in test data. The NB-BLCA model using Gibbs sampling accuracy was 87.77 (95% CI: 84.87-90.29). This index was estimated at 77.22 (95% CI: 73.64-80.53) and 74.71 (95% CI: 71.02-78.15) for the NB-BLCA model using the EM algorithm and ordinary NB classifier, respectively.

CONCLUSIONS

When considering the modification of the NB classifier, incorporating a latent component into the model offers numerous advantages, particularly within medical and health-related contexts. By doing so, the researchers can bypass the extensive search algorithm and structure learning required in the local learning and structure extension approach. The inclusion of latent class variables allows for the integration of all attributes during model construction. Consequently, the NB-BLCA model serves as a suitable alternative to conventional NB classifiers when the assumption of independence is violated, especially in domains pertaining to health and medicine.

Remote fit wrist braces through artificial intelligence

INTRODUCTION

Physical boundaries to access skilled orthotist or hand therapy care may be hindered by multiple factors, such as geography, or availability. This study evaluated the accuracy of fitting a prefabricated wrist splint using an app on a smart device. We hypothesize that remote brace fitting by artificial intelligence (AI) can accurately determine the brace size the patient needs without in-person fitting.

METHODS

Healthy volunteers were recruited to fit wrist braces. Using 2 standardized calibrated images captured by the smart device, each subject's image was loaded into the machine learning software (AI). Later, hand features were extracted, calibrated, and measured the application, calculated the correct splint size, and compared with the splint chosen by our subjects to improve its own accuracy. As a control (control 1), the subjects independently selected the best brace fit from an array of available splints. Subject selection was recorded and compared with the AI fit splint. As the second method of fitting (control 2), we compared the manufacturer recommended brace size (based on measured wrist circumference and provided sizing chart/insert brochure) with the AI fit splint.

RESULTS

A total of 54 volunteers were included. Thirty-two splints predicted by the algorithm matched the exact size chosen by each subject yielding 70% accuracy with a standard deviation of 10% ( p < 0.001). The accuracy increased to 90% with 5% standard deviation if the splints were predicted within the next size category. Fit by manufacturer sizing chart was only 33% in agreement with participant selection.

CONCLUSION

Remote brace fitting using AI prediction model may be an acceptable alternative to current standards because it can accurately predict wrist splint size. As more subjects were analyzed, the AI algorithm became more accurate predicting proper brace fit. In addition, AI fit braces are more than twice as accurate as relying on the manufacturer sizing chart.

Towards artificial intelligence-enabled extracellular vesicle precision drug delivery

Extracellular Vesicles (EVs), particularly exosomes, recently exploded into nanomedicine as an emerging drug delivery approach due to their superior biocompatibility, circulating stability, and bioavailability in vivo. However, EV heterogeneity makes molecular targeting precision a critical challenge. Deciphering key molecular drivers for controlling EV tissue targeting specificity is in great need. Artificial intelligence (AI) brings powerful prediction ability for guiding the rational design of engineered EVs in precision control for drug delivery. This review focuses on cutting-edge nano-delivery via integrating large-scale EV data with AI to develop AI-directed EV therapies and illuminate the clinical translation potential. We briefly review the current status of EVs in drug delivery, including the current frontier, limitations, and considerations to advance the field. Subsequently, we detail the future of AI in drug delivery and its impact on precision EV delivery. Our review discusses the current universal challenge of standardization and critical considerations when using AI combined with EVs for precision drug delivery. Finally, we will conclude this review with a perspective on future clinical translation led by a combined effort of AI and EV research.

Aerobic Fitness as an Important Moderator Risk Factor for Loneliness in Physically Trained Older People: An Explanatory Case Study Using Machine Learning

BACKGROUND

Loneliness in older people seems to have emerged as an increasingly prevalent social problem.

OBJECTIVE

To apply a machine learning (ML) algorithm to the task of understanding the influence of sociodemographic variables, physical fitness, physical activity levels (PAL), and sedentary behavior (SB) on the loneliness feelings of physically trained older people.

MATERIALS AND METHODS

The UCLA loneliness scale was used to evaluate loneliness, the Functional Fitness Test Battery was used to evaluate the correlation of sociodemographic variables, physical fitness, PAL, and SB in the loneliness feelings scores of 23 trained older people (19 women and 4 men). For this purpose, a naive Bayes ML algorithm was applied.

RESULTS

After analysis, we inferred that aerobic fitness (AF), hand grip strength (HG), and upper limb strength (ULS) comprised the most relevant variables panel to cause high participant loneliness with 100% accuracy and F-1 score.

CONCLUSIONS

The naive Bayes algorithm with leave-one-out cross-validation (LOOCV) predicted loneliness in trained older with a high precision. In addition, AF was the most potent variable in reducing loneliness risk.

A twin logistic regression method based on attribute-oriented fuzzy rough set

Forecasting is making predictions about what will happen or how things will change. This can help people avoid blindness and losses and play a significant role in their lives. In multi-attribute prediction problems, the correlation between attributes is often ignored, which affects prediction accuracy. Based on fuzzy rough sets and logistic regression, this paper proposes a new logistic regression method that fully considers attribute correlation, namely a twin logistic regression method based on attribute-oriented fuzzy rough sets. Firstly, attribute-oriented fuzzy rough sets are studied and analyzed. Then, the optimistic and pessimistic predictions are achieved by fuzzy rough sets and logistic regression, and the final result is obtained by fusing the optimistic and pessimistic predictions. Finally, the effectiveness of the twin logistic regression method is verified.

Identification of species-specific RNA N6-methyladinosine modification sites from RNA sequences

N6-methyladinosine (m6A) modification is the most abundant co-transcriptional modification in eukaryotic RNA and plays important roles in cellular regulation. Traditional high-throughput sequencing experiments used to explore functional mechanisms are time-consuming and labor-intensive, and most of the proposed methods focused on limited species types. To further understand the relevant biological mechanisms among different species with the same RNA modification, it is necessary to develop a computational scheme that can be applied to different species. To achieve this, we proposed an attention-based deep learning method, adaptive-m6A, which consists of convolutional neural network, bi-directional long short-term memory and an attention mechanism, to identify m6A sites in multiple species. In addition, three conventional machine learning (ML) methods, including support vector machine, random forest and logistic regression classifiers, were considered in this work. In addition to the performance of ML methods for multi-species prediction, the optimal performance of adaptive-m6A yielded an accuracy of 0.9832 and the area under the receiver operating characteristic curve of 0.98. Moreover, the motif analysis and cross-validation among different species were conducted to test the robustness of one model towards multiple species, which helped improve our understanding about the sequence characteristics and biological functions of RNA modifications in different species.

Evaluation of a decided sample size in machine learning applications

BACKGROUND

An appropriate sample size is essential for obtaining a precise and reliable outcome of a study. In machine learning (ML), studies with inadequate samples suffer from overfitting of data and have a lower probability of producing true effects, while the increment in sample size increases the accuracy of prediction but may not cause a significant change after a certain sample size. Existing statistical approaches using standardized mean difference, effect size, and statistical power for determining sample size are potentially biased due to miscalculations or lack of experimental details. This study aims to design criteria for evaluating sample size in ML studies. We examined the average and grand effect sizes and the performance of five ML methods using simulated datasets and three real datasets to derive the criteria for sample size. We systematically increase the sample size, starting from 16, by randomly sampling and examine the impact of sample size on classifiers' performance and both effect sizes. Tenfold cross-validation was used to quantify the accuracy.

RESULTS

The results demonstrate that the effect sizes and the classification accuracies increase while the variances in effect sizes shrink with the increment of samples when the datasets have a good discriminative power between two classes. By contrast, indeterminate datasets had poor effect sizes and classification accuracies, which did not improve by increasing sample size in both simulated and real datasets. A good dataset exhibited a significant difference in average and grand effect sizes. We derived two criteria based on the above findings to assess a decided sample size by combining the effect size and the ML accuracy. The sample size is considered suitable when it has appropriate effect sizes (≥ 0.5) and ML accuracy (≥ 80%). After an appropriate sample size, the increment in samples will not benefit as it will not significantly change the effect size and accuracy, thereby resulting in a good cost-benefit ratio.

CONCLUSION

We believe that these practical criteria can be used as a reference for both the authors and editors to evaluate whether the selected sample size is adequate for a study.

Machine Learning Approaches for the Prediction of Hepatitis B and C Seropositivity

(1) Background: The identification of patients at risk for hepatitis B and C viral infection is a challenge for the clinicians and public health specialists. The aim of this study was to evaluate and compare the predictive performances of four machine learning-based models for the prediction of HBV and HCV status. (2) Methods: This prospective cohort screening study evaluated adults from the North-Eastern and South-Eastern regions of Romania between January 2022 and November 2022 who underwent viral hepatitis screening in their family physician's offices. The patients' clinical characteristics were extracted from a structured survey and were included in four machine learning-based models: support vector machine (SVM), random forest (RF), naïve Bayes (NB), and K nearest neighbors (KNN), and their predictive performance was assessed. (3) Results: All evaluated models performed better when used to predict HCV status. The highest predictive performance was achieved by KNN algorithm (accuracy: 98.1%), followed by SVM and RF with equal accuracies (97.6%) and NB (95.7%). The predictive performance of these models was modest for HBV status, with accuracies ranging from 78.2% to 97.6%. (4) Conclusions: The machine learning-based models could be useful tools for HCV infection prediction and for the risk stratification process of adult patients who undergo a viral hepatitis screening program.

Prediction of HELLP Syndrome Severity Using Machine Learning Algorithms-Results from a Retrospective Study

(1) Background: HELLP (hemolysis, elevated liver enzymes, and low platelets) syndrome is a rare and life-threatening complication of preeclampsia. The aim of this study was to evaluate and compare the predictive performances of four machine learning-based models for the prediction of HELLP syndrome, and its subtypes according to the Mississippi classification; (2) Methods: This retrospective case-control study evaluated pregnancies that occurred in women who attended a tertiary maternity hospital in Romania between January 2007 and December 2021. The patients' clinical and paraclinical characteristics were included in four machine learning-based models: decision tree (DT), naïve Bayes (NB), k-nearest neighbors (KNN), and random forest (RF), and their predictive performance were assessed; (3) Results: Our results showed that HELLP syndrome was best predicted by RF (accuracy: 89.4%) and NB (accuracy: 86.9%) models, while DT (accuracy: 91%) and KNN (accuracy: 87.1%) models had the highest performance when used to predict class 1 HELLP syndrome. The predictive performance of these models was modest for class 2 and 3 of HELLP syndrome, with accuracies ranging from 65.2% and 83.8%; (4) Conclusions: The machine learning-based models could be useful tools for predicting HELLP syndrome, and its most severe form-class 1.

Machine Learning for Endometrial Cancer Prediction and Prognostication

Endometrial cancer (EC) is a prevalent uterine cancer that remains a major contributor to cancer-associated morbidity and mortality. EC diagnosed at advanced stages shows a poor therapeutic response. The clinically utilized EC diagnostic approaches are costly, time-consuming, and are not readily available to all patients. The rapid growth in computational biology has enticed substantial research attention from both data scientists and oncologists, leading to the development of rapid and cost-effective computer-aided cancer surveillance systems. Machine learning (ML), a subcategory of artificial intelligence, provides opportunities for drug discovery, early cancer diagnosis, effective treatment, and choice of treatment modalities. The application of ML approaches in EC diagnosis, therapies, and prognosis may be particularly relevant. Considering the significance of customized treatment and the growing trend of using ML approaches in cancer prediction and monitoring, a critical survey of ML utility in EC may provide impetus research in EC and assist oncologists, molecular biologists, biomedical engineers, and bioinformaticians to further collaborative research in EC. In this review, an overview of EC along with risk factors and diagnostic methods is discussed, followed by a comprehensive analysis of the potential ML modalities for prevention, screening, detection, and prognosis of EC patients.

Artificial Intelligence (AI) and Internet of Medical Things (IoMT) Assisted Biomedical Systems for Intelligent Healthcare

Artificial intelligence (AI) is a modern approach based on computer science that develops programs and algorithms to make devices intelligent and efficient for performing tasks that usually require skilled human intelligence. AI involves various subsets, including machine learning (ML), deep learning (DL), conventional neural networks, fuzzy logic, and speech recognition, with unique capabilities and functionalities that can improve the performances of modern medical sciences. Such intelligent systems simplify human intervention in clinical diagnosis, medical imaging, and decision-making ability. In the same era, the Internet of Medical Things (IoMT) emerges as a next-generation bio-analytical tool that combines network-linked biomedical devices with a software application for advancing human health. In this review, we discuss the importance of AI in improving the capabilities of IoMT and point-of-care (POC) devices used in advanced healthcare sectors such as cardiac measurement, cancer diagnosis, and diabetes management. The role of AI in supporting advanced robotic surgeries developed for advanced biomedical applications is also discussed in this article. The position and importance of AI in improving the functionality, detection accuracy, decision-making ability of IoMT devices, and evaluation of associated risks assessment is discussed carefully and critically in this review. This review also encompasses the technological and engineering challenges and prospects for AI-based cloud-integrated personalized IoMT devices for designing efficient POC biomedical systems suitable for next-generation intelligent healthcare.

Artificial intelligence and machine learning in precision and genomic medicine

The advancement of precision medicine in medical care has led behind the conventional symptom-driven treatment process by allowing early risk prediction of disease through improved diagnostics and customization of more effective treatments. It is necessary to scrutinize overall patient data alongside broad factors to observe and differentiate between ill and relatively healthy people to take the most appropriate path toward precision medicine, resulting in an improved vision of biological indicators that can signal health changes. Precision and genomic medicine combined with artificial intelligence have the potential to improve patient healthcare. Patients with less common therapeutic responses or unique healthcare demands are using genomic medicine technologies. AI provides insights through advanced computation and inference, enabling the system to reason and learn while enhancing physician decision making. Many cell characteristics, including gene up-regulation, proteins binding to nucleic acids, and splicing, can be measured at high throughput and used as training objectives for predictive models. Researchers can create a new era of effective genomic medicine with the improved availability of a broad range of datasets and modern computer techniques such as machine learning. This review article has elucidated the contributions of ML algorithms in precision and genome medicine.

Artificial intelligence and machine learning in precision and genomic medicine

The advancement of precision medicine in medical care has led behind the conventional symptom-driven treatment process by allowing early risk prediction of disease through improved diagnostics and customization of more effective treatments. It is necessary to scrutinize overall patient data alongside broad factors to observe and differentiate between ill and relatively healthy people to take the most appropriate path toward precision medicine, resulting in an improved vision of biological indicators that can signal health changes. Precision and genomic medicine combined with artificial intelligence have the potential to improve patient healthcare. Patients with less common therapeutic responses or unique healthcare demands are using genomic medicine technologies. AI provides insights through advanced computation and inference, enabling the system to reason and learn while enhancing physician decision making. Many cell characteristics, including gene up-regulation, proteins binding to nucleic acids, and splicing, can be measured at high throughput and used as training objectives for predictive models. Researchers can create a new era of effective genomic medicine with the improved availability of a broad range of datasets and modern computer techniques such as machine learning. This review article has elucidated the contributions of ML algorithms in precision and genome medicine.

Artificial intelligence and machine learning approaches using gene expression and variant data for personalized medicine

Precision medicine uses genetic, environmental and lifestyle factors to more accurately diagnose and treat disease in specific groups of patients, and it is considered one of the most promising medical efforts of our time. The use of genetics is arguably the most data-rich and complex components of precision medicine. The grand challenge today is the successful assimilation of genetics into precision medicine that translates across different ancestries, diverse diseases and other distinct populations, which will require clever use of artificial intelligence (AI) and machine learning (ML) methods. Our goal here was to review and compare scientific objectives, methodologies, datasets, data sources, ethics and gaps of AI/ML approaches used in genomics and precision medicine. We selected high-quality literature published within the last 5 years that were indexed and available through PubMed Central. Our scope was narrowed to articles that reported application of AI/ML algorithms for statistical and predictive analyses using whole genome and/or whole exome sequencing for gene variants, and RNA-seq and microarrays for gene expression. We did not limit our search to specific diseases or data sources. Based on the scope of our review and comparative analysis criteria, we identified 32 different AI/ML approaches applied in variable genomics studies and report widely adapted AI/ML algorithms for predictive diagnostics across several diseases.

Applying machine learning to explore the association between biological stress and near misses in emergency medicine residents

Physician stress is associated with near misses and adverse medical events. However, little is known about physiological mechanisms linking stress to such events. We explored the utility of machine learning to determine whether the catabolic stress hormone cortisol and the anabolic, anti-stress hormone dehydroepiandrosterone sulfate (DHEA-S), as well as the cortisol to DHEA-S ratio relate to near misses in emergency medicine residents during active duty in a trauma 1 emergency department. Compared to statistical models better suited for inference, machine learning models allow for prediction in situations that have not yet occurred, and thus better suited for clinical applications. This exploratory study used multiple machine learning models to determine possible relationships between biomarkers and near misses. Of the various models tested, support vector machine with radial bias function kernels and support vector machine with linear kernels performed the best, with training accuracies of 85% and 79% respectively. When evaluated on a test dataset, both models had prediction accuracies of around 80%. The pre-shift cortisol to DHEA-S ratio was shown to be the most important predictor in interpretable models tested. Results suggest that interventions that help emergency room physicians relax before they begin their shift could reduce risk of errors and improve patient and physician outcomes. This pilot demonstrates promising results regarding using machine learning to better understand the stress biology of near misses. Future studies should use larger groups and relate these variables to information in electronic medical records, such as objective and patient-reported quality measures.

Feature optimization in CNN using MROA for disease classification

Electronic Medical Records (EMR) carry important information about a patient’s journey. The past decade shows substantial use of Natural Language Processing (NLP)-based Information Retrieval (IR) techniques to extract insights such as symptoms, diseases, and tests from these unstructured records. The state-of-the-art shows that convolutional neural networks (CNN) make a significant contribution to the disease classification task.A significant improvement in precise knowledge mining is possible with precise feature extraction. Feature selection addresses undesirable, unneeded, or irrelevant features. This article proposes a Modified Rider Optimization Algorithm (MROA) to choose important features by selecting optimal weights from a pool of randomly generated weights based on high accuracy and less training time in the CNN algorithm. A modified approach is trained on 114 N2C2 patients’ records to extract symptoms, disease, and tests are performed on them to perform disease classification tasks. The proposed approach is found to be accurate, with 97.77% accuracy in the disease classification and treatment prediction task from EMR.

Machine learning classifiers do not improve prediction of hospitalization > 2 days after fast-track hip and knee arthroplasty compared with a classical statistical risk model

Background and purpose: Prediction of postoperative outcomes and length of hospital stay (LOS) of patients is vital for allocation of healthcare resources. We investigated the performance of prediction models based on machinelearning algorithms compared with a previous risk stratification model using traditional multiple logistic regression, for predicting the risk of a LOS of > 2 days after fast-track total hip and knee replacement. Patients and methods: 3 different machine learning classifiers were trained on data from the Lundbeck Centre for Fast-track Hip and Knee Replacement Database (LCDB) collected from 9,512 patients between 2016 and 2017. The chosen classifiers were a random forest classifier (RF), a support vector machine classifier with a polynomial kernel (SVM), and a multinomial Naïve-Bayes classifier (NB). Results: Comparing performance measures of the classifiers with the traditional model revealed that all the models had a similar performance in terms of F1 score, accuracy, sensitivity, specificity, area under the receiver operating curve (AUC), and area under the precision-recall curve (AUPRC). A feature importance analysis of the RF classifier found hospital, age, use of walking aid, living alone, and joint operated on to be the most relevant input features. None of the classifiers reached a clinically relevant performance with the input data from the LCDB. Interpretation: Despite the promising prospects of machine-learning practices for disease and risk prediction, none of the machine learning models tested outperformed the traditional multiple regression model in predicting which patients in this cohort had a LOS > 2 days.

Learning Disease Causality Knowledge from Web of Health Data

Health information becomes importantly valuable for protecting public health in the current coronavirus situation. Knowledge-based information systems can play a crucial role in helping individuals to practice risk assessment and remote diagnosis. We introduce a novel approach that will develop causality-focused knowledge learning in a robust and transparent manner. Then, the machine gains the causality and probability knowledge for inference (thinking) and accurate prediction later. Besides, the hidden knowledge can be discovered beyond the existing understanding of the diseases. The whole approach is built on a Causal Probability Description Logic Framework that combines Natural Language Processing (NLP), Causality Analysis and extended Knowledge Graph (KG) technologies together. The experimental work has processed 801 diseases in total (from the UK NHS website linking with DBpedia datasets). As a result, the machine learnt comprehensive health causal knowledge and relations among the diseases, symptoms, and other facts efficiently.

Prediction of Autonomy Loss in Alzheimer’s Disease

The evolution of functional autonomy loss leads to institutionalization of people affected by Alzheimer’s disease (AD), to an alteration of their quality of life and that of their caregivers. To predict loss of functional autonomy could optimize prevention strategies, aids and cost of care. The aim of this study was to develop and to cross-validate a model to predict loss of functional autonomy as assessed by Instrumental Activities of Daily Living (IADL) score. Outpatients with probable AD and with 2 or more visits to the Clinical and Research Memory Centre of the University Hospital were included. Four Tree-Augmented Naïve bayesian networks (6, 12, 18 and 24 months of follow-up) were built. Variables included in the model were demographic data, IADL score, MMSE score, comorbidities, drug prescription (psychotropics and AD-specific drugs). A 10-fold cross-validation was conducted to evaluate robustness of models. The study initially included 485 patients in the prospective cohort. The best performance after 10-fold cross-validation was obtained with the model able to predict loss of functional autonomy at 18 months (area under the curve of the receiving operator characteristic curve = 0.741, 27% of patients misclassified, positive predictive value = 77% and negative predictive value = 73%). The 13 variables used explain 41.6% of the evolution of functional autonomy at 18 months. A high-performing predictive model of AD evolution of functional autonomy was obtained. An external validation is needed to use the model in clinical routine so as to optimize the patient care.

Mining real-world high dimensional structured data in medicine and its use in decision support. Some different perspectives on unknowns, interdependency, and distinguishability

There are many difficulties in extracting and using knowledge for medical analytic and predictive purposes from Real-World Data, even when the data is already well structured in the manner of a large spreadsheet. Preparative curation and standardization or "normalization" of such data involves a variety of chores but underlying them is an interrelated set of fundamental problems that can in part be dealt with automatically during the datamining and inference processes. These fundamental problems are reviewed here and illustrated and investigated with examples. They concern the treatment of unknowns, the need to avoid independency assumptions, and the appearance of entries that may not be fully distinguished from each other. Unknowns include errors detected as implausible (e.g., out of range) values that are subsequently converted to unknowns. These problems are further impacted by high dimensionality and problems of sparse data that inevitably arise from high-dimensional datamining even if the data is extensive. All these considerations are different aspects of incomplete information, though they also relate to problems that arise if care is not taken to avoid or ameliorate consequences of including the same information twice or more, or if misleading or inconsistent information is combined. This paper addresses these aspects from a slightly different perspective using the Q-UEL language and inference methods based on it by borrowing some ideas from the mathematics of quantum mechanics and information theory. It takes the view that detection and correction of probabilistic elements of knowledge subsequently used in inference need only involve testing and correction so that they satisfy certain extended notions of coherence between probabilities. This is by no means the only possible view, and it is explored here and later compared with a related notion of consistency.

Validating machine learning models for the prediction of labour induction intervention using routine data: a registry-based retrospective cohort study at a tertiary hospital in northern Tanzania

OBJECTIVES

We aimed at identifying the important variables for labour induction intervention and assessing the predictive performance of machine learning algorithms.

SETTING

We analysed the birth registry data from a referral hospital in northern Tanzania. Since July 2000, every birth at this facility has been recorded in a specific database.

PARTICIPANTS

21 578 deliveries between 2000 and 2015 were included. Deliveries that lacked information regarding the labour induction status were excluded.

PRIMARY OUTCOME

Deliveries involving labour induction intervention.

RESULTS

Parity, maternal age, body mass index, gestational age and birth weight were all found to be important predictors of labour induction. Boosting method demonstrated the best discriminative performance (area under curve, AUC=0.75: 95% CI (0.73 to 0.76)) while logistic regression presented the least (AUC=0.71: 95% CI (0.70 to 0.73)). Random forest and boosting algorithms showed the highest net-benefits as per the decision curve analysis.

CONCLUSION

All of the machine learning algorithms performed well in predicting the likelihood of labour induction intervention. Further optimisation of these classifiers through hyperparameter tuning may result in an improved performance. Extensive research into the performance of other classifier algorithms is warranted.

Naïve Bayesian network-based contribution analysis of tumor biology and healthcare factors to racial disparity in breast cancer stage-at-diagnosis

Background

Variation in breast cancer stage at initial diagnosis (including racial disparities) is driven both by tumor biology and healthcare factors.

Methods

We studied women age 67-74 with initial diagnosis of breast cancer from 2006 through 2014 in the SEER-Medicare database. We extracted variables related to tumor biology (histologic grade and hormone receptor status) and healthcare factors (screening mammography [SM] utilization and time delay from mammography to diagnostic biopsy). We used naïve Bayesian networks (NBNs) to illustrate the relationships among patient-specific factors and stage-at-diagnosis for African American (AA) and white patients separately. After identifying and controlling confounders, we conducted counterfactual inference through the NBN, resulting in an unbiased evaluation of the causal effects of individual factors on the expected utility of stage-at-diagnosis. An NBN-based decomposition mechanism was developed to evaluate the contributions of each patient-specific factor to an actual racial disparity in stage-at-diagnosis. 2000 bootstrap samples from our training patients were used to compute the 95% confidence intervals (CIs) of these contributions.

Results

Using a causal-effect contribution analysis, the relative contributions of each patient-specific factor to the actual racial disparity in stage-at-diagnosis were as follows: tumor grade, 45.1% (95% CI: 44.5%, 45.8%); hormone receptor status, 5.0% (4.5%, 5.4%); mammography utilization, 23.1% (22.4%, 24.0%); and biopsy delay 26.8% (26.1%, 27.3%).

Conclusion

The modifiable mechanisms of mammography utilization and biopsy delay drive about 49.9% of racial difference in stage-at-diagnosis, potentially guiding more targeted interventions to eliminate cancer outcome disparities.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13755-021-00165-5.

Forecasting Erroneous Neural Machine Translation of Disease Symptoms: Development of Bayesian Probabilistic Classifiers for Cross-Lingual Health Translation

BACKGROUND

Machine translation (MT) technologies have increasing applications in healthcare. Despite their convenience, cost-effectiveness, and constantly improved accuracy, research shows that the use of MT tools in medical or healthcare settings poses risks to vulnerable populations.

OBJECTIVES

We aimed to develop machine learning classifiers (MNB and RVM) to forecast nuanced yet significant MT errors of clinical symptoms in Chinese neural MT outputs.

METHODS

We screened human translations of MSD Manuals for information on self-diagnosis of infectious diseases and produced their matching neural MT outputs for subsequent pairwise quality assessment by trained bilingual health researchers. Different feature optimisation and normalisation techniques were used to identify the best feature set.

RESULTS

The RVM classifier using optimised, normalised (L₂ normalisation) semantic features achieved the highest sensitivity, specificity, AUC, and accuracy. MNB achieved similar high performance using the same optimised semantic feature set. The best probability threshold of the best performing RVM classifier was found at 0.6, with a very high positive likelihood ratio (LR+) of 27.82 (95% CI: 3.99, 193.76), and a low negative likelihood ratio (LR-) of 0.19 (95% CI: 0.08, 046), suggesting the high diagnostic utility of our model to predict the probabilities of erroneous MT of disease symptoms to help reverse potential inaccurate self-diagnosis of diseases among vulnerable people without adequate medical knowledge or an ability to ascertain the reliability of MT outputs.

CONCLUSION

Our study demonstrated the viability, flexibility, and efficiency of introducing machine learning models to help promote risk-aware use of MT technologies to achieve optimal, safer digital health outcomes for vulnerable people.

Machine learning: principles and applications for thoracic surgery

OBJECTIVES

Machine learning (ML) has experienced a revolutionary decade with advances across many disciplines. We seek to understand how recent advances in ML are going to specifically influence the practice of surgery in the future with a particular focus on thoracic surgery.

METHODS

Review of relevant literature in both technical and clinical domains.

RESULTS

ML is a revolutionary technology that promises to change the way that surgery is practiced in the near future. Spurred by an advance in computing power and the volume of data produced in healthcare, ML has shown remarkable ability to master tasks that had once been reserved for physicians. Supervised learning, unsupervised learning and reinforcement learning are all important techniques that can be leveraged to improve care. Five key applications of ML to cardiac surgery include diagnostics, surgical skill assessment, postoperative prognostication, augmenting intraoperative performance and accelerating translational research. Some key limitations of ML include lack of interpretability, low quality and volumes of relevant clinical data, ethical limitations and difficulties with clinical implementation.

CONCLUSIONS

In the future, the practice of cardiac surgery will be greatly augmented by ML technologies, ultimately leading to improved surgical performance and better patient outcomes.

A Clinical Decision Support Tool to Find the Best Initial Intravenous Cyclosporine Regimen in Pediatric Hematopoietic Stem Cell Transplantation

To optimize cyclosporine A (CsA) dosing regimen in pediatric patients undergoing hematopoietic stem cell transplantation (HSCT), we aimed to provide clinicians with a validated decision support tool for determining the most suitable first dose of intravenous CsA. We used a 10-year monocentric data set of pediatric patients undergoing HSCT. Discretization of all variables was performed according to literature or thanks to algorithms using Shannon entropy (from information theory) or equal width intervals. The first 8 years were used to build the Bayesian network model. This model underwent a 10-fold cross-validation, and then a prospective validation with data of the last 2 years. There were 3.3% and 4.1% of missing values in the training and the validation data set, respectively. After prospective validation, the Tree-Augmented Naïve Bayesian network shows interesting prediction performances with an average area under the receiver operating characteristic curve of 0.804, 32.8% of misclassified patients, a true-positive rate of 0.672, and a false-positive rate of 0.285. This validated model allows good predictions to propose an optimized and personalized initial CsA dose for pediatric patients undergoing HSCT. The clinical impact of its use should be further evaluated.

Bayesian network models with decision tree analysis for management of childhood malaria in Malawi

BACKGROUND

Malaria is a major cause of death in children under five years old in low- and middle-income countries such as Malawi. Accurate diagnosis and management of malaria can help reduce the global burden of childhood morbidity and mortality. Trained healthcare workers in rural health centers manage malaria with limited supplies of malarial diagnostic tests and drugs for treatment. A clinical decision support system that integrates predictive models to provide an accurate prediction of malaria based on clinical features could aid healthcare workers in the judicious use of testing and treatment. We developed Bayesian network (BN) models to predict the probability of malaria from clinical features and an illustrative decision tree to model the decision to use or not use a malaria rapid diagnostic test (mRDT).

METHODS

We developed two BN models to predict malaria from a dataset of outpatient encounters of children in Malawi. The first BN model was created manually with expert knowledge, and the second model was derived using an automated method. The performance of the BN models was compared to other statistical models on a range of performance metrics at multiple thresholds. We developed a decision tree that integrates predictions with the costs of mRDT and a course of recommended treatment.

RESULTS

The manually created BN model achieved an area under the ROC curve (AUC) equal to 0.60 which was statistically significantly higher than the other models. At the optimal threshold for classification, the manual BN model had sensitivity and specificity of 0.74 and 0.42 respectively, and the automated BN model had sensitivity and specificity of 0.45 and 0.68 respectively. The balanced accuracy values were similar across all the models. Sensitivity analysis of the decision tree showed that for values of probability of malaria below 0.04 and above 0.40, the preferred decision that minimizes expected costs is not to perform mRDT.

CONCLUSION

In resource-constrained settings, judicious use of mRDT is important. Predictive models in combination with decision analysis can provide personalized guidance on when to use mRDT in the management of childhood malaria. BN models can be efficiently derived from data to support clinical decision making.

R-HEFS: Rough set based heterogeneous ensemble feature selection method for medical data classification

Feature selection is one of the trustworthy processes of dimensionality reduction technique to select a subset of relevant and non-redundant features from large datasets. Ensemble feature selection (EFS) approach is a recent technique aiming at accumulating diversity in the subset of selected features. It improves the performance of learning algorithms and obtains more stable and robust results. In this paper, a novel rough set theory (RST) based heterogeneous EFS method (R-HEFS) is proposed for selecting the less redundant and highly relevant features during the aggregation of diverse feature subsets by applying the feature-class, feature-feature rough dependency and feature-significance measures. In R-HEFS five state-of-the-art RST based filter methods are used as a base feature selectors. Experiments are carried out on 10 benchmark medical datasets collected from the UCI repository. For the imputation of the missing values and discretization of the continuous features, k nearest neighbor (kNN) imputation method and RST based discretization techniques are applied. The effectiveness of the proposed R-HEFS method is evaluated and analyzed by using four benchmark classifiers viz., Naïve Bayes (NB), random forest (RF), support vector machine (SVM), and AdaBoost. The proposed R-HEFS method turns out to be effective by removing the non-relevant and redundant features during the process of aggregation of base feature selectors and it assists to increase the classification accuracy. Out of 10 different medical datasets, on 7 datasets, R-HEFS has achieved better average classification accuracy. So, the overall results strongly suggest that the proposed R-HEFS method can reduce the dimension of large medical datasets and may help the physicians or medical experts to diagnose (classify) different diseases with lesser computational complexities.

Identification of Most Relevant Features for Classification of Francisella tularensis using Machine Learning

Background:

Francisella tularensis is a stealth pathogen fatal for animals and humans. Ease of its propagation, coupled with high capacity for ailment and death makes it a potential candidate for biological weapon.

Objective:

Work related to the pathogen’s classification and factors affecting its prolonged existence in soil is limited to statistical measures. Machine learning other than conventional analysis methods may be applied to better predict epidemiological modeling for this soil-borne pathogen.

Methods:

Feature-ranking algorithms namely; relief, correlation and oneR are used for soil attribute ranking. Moreover, classification algorithms; SVM, random forest, naive bayes, logistic regression and MLP are used for classification of the soil attribute dataset for Francisella tularensis positive and negative soils.

Results:

Feature-ranking methods concluded that clay, nitrogen, organic matter, soluble salts, zinc, silt and nickel are the most significant attributes while potassium, phosphorous, iron, calcium, copper, chromium and sand are the least contributing risk factors for the persistence of the pathogen. However, clay is the most significant and potassium is the least contributing attribute. Data analysis suggests that feature-ranking using relief produced classification accuracy of 84.35% for multilayer perceptron; 82.99% for linear regression; 80.27% for SVM and random forest; and 78.23% for naive bayes, which is better than other ranking methods. MLP outperforms other classifiers by generating an accuracy of 84.35%, 82.99% and 81.63% for feature-ranking using relief, correlation and oneR algorithms, respectively.

Conclusion:

These models can significantly improve accuracy and can minimize the risk of incorrect classification. They further help in controlling epidemics and thereby minimizing the socio-economic impact on the society.

Automated detection of patients with dementia whose symptoms have been identified in primary care but have no formal diagnosis: a retrospective case-control study using electronic primary care records

OBJECTIVES

UK statistics suggest only two-thirds of patients with dementia get a diagnosis recorded in primary care. General practitioners (GPs) report barriers to formally diagnosing dementia, so some patients may be known by GPs to have dementia but may be missing a diagnosis in their patient record. We aimed to produce a method to identify these 'known but unlabelled' patients with dementia using data from primary care patient records.

DESIGN

Retrospective case-control study using routinely collected primary care patient records from Clinical Practice Research Datalink.

SETTING

UK general practice.

PARTICIPANTS

English patients aged >65 years, with a coded diagnosis of dementia recorded in 2000-2012 (cases), matched 1:1 with patients with no diagnosis code for dementia (controls).

INTERVENTIONS

Eight coded and nine keyword concepts indicating symptoms, screening tests, referrals and care for dementia recorded in the 5 years before diagnosis. We trialled machine learning classifiers to discriminate between cases and controls (logistic regression, naïve Bayes, random forest).

PRIMARY AND SECONDARY OUTCOMES

The outcome variable was dementia diagnosis code; the accuracy of classifiers was assessed using area under the receiver operating characteristic curve (AUC); the order of features contributing to discrimination was examined.

RESULTS

93 426 patients were included; the median age was 83 years (64.8% women). Three classifiers achieved high discrimination and performed very similarly. AUCs were 0.87-0.90 with coded variables, rising to 0.90-0.94 with keywords added. Feature prioritisation was different for each classifier; commonly prioritised features were Alzheimer's prescription, dementia annual review, memory loss and dementia keywords.

CONCLUSIONS

It is possible to detect patients with dementia who are known to GPs but unlabelled with a diagnostic code, with a high degree of accuracy in electronic primary care record data. Using keywords from clinic notes and letters improves accuracy compared with coded data alone. This approach could improve identification of dementia cases for record-keeping, service planning and delivery of good quality care.

A Comprehensive Analysis Identified Hub Genes and Associated Drugs in Alzheimer's Disease

Alzheimer's disease (AD) is the most common neurodegenerative disease among the elderly and has become a growing global health problem causing great concern. However, the pathogenesis of AD is unclear and no specific therapeutics are available to provide the sustained remission of the disease. In this study, we used comprehensive bioinformatics to determine 158 potential genes, whose expression levels changed between the entorhinal and temporal lobe cortex samples from cognitively normal individuals and patients with AD. Then, we clustered these genes in the protein-protein interaction analysis and identified six significant genes that had more biological functions. Besides, we conducted a drug-gene interaction analysis of module genes in the drug-gene interaction database and obtained 26 existing drugs that might be applied for the prevention and treatment of AD. In addition, a predictive model was built based on the selected genes using different machine learning algorithms to identify individuals with AD. These findings may provide new insights into AD therapy.

Effectiveness of the Early Response to COVID-19: Data Analysis and Modelling

Governments around the world have introduced a number of stringent policies to try to contain COVID-19 outbreaks, but the relative importance of such measures, in comparison to the community response to these restrictions, the amount of testing conducted, and the interconnections between them, is not well understood yet. In this study, data were collected from numerous online sources, pre-processed and analysed, and a number of Bayesian Network models were developed, in an attempt to unpack such complexity. Results show that early, high-volume testing was the most crucial factor in successfully monitoring and controlling the outbreaks; when testing was low, early government and community responses were found to be both critical in predicting how rapidly cases and deaths grew in the first weeks of the outbreak. Results also highlight that in countries with low early test numbers, the undiagnosed cases could have been up to five times higher than the officially diagnosed cases. The conducted analysis and developed models can be refined in the future with more data and variables, to understand/model potential second waves of contagions.

Human Tongue Thermography Could Be a Prognostic Tool for Prescreening the Type II Diabetes Mellitus

Diabetes mellitus is one of the life threatening diseases over the globe, and an early prediction of diabetes is of utmost importance in this current scenario. International Diabetes Federation (IDF) reported nearly half of the world's population was undiagnosed and unaware of being developed into diabetes. In 2017, around 84 million individuals were living with diabetes, and it might increase to 156 million by the end of 2045 stated by IDF. Generally, the diagnosis of diabetes relies on the biochemical method that may cause uneasiness and probability of infections to the subjects. To overcome such difficulties, a noninvasive method is much needed around the globe for primary screening. A change in body temperature is an indication of various diseases. Infrared thermal imaging is relatively a novel technique for skin temperature measurement and turned out to be well known in the medical field due to being noninvasive, risk-free, and repeatable. According to traditional Chinese medicine, the human tongue is a sensitive mirror that reflects the body's pathophysiological condition. So, we have (i) analysed and classified diabetes based on thermal variations at human tongue, (ii) segmented the hot spot regions from tongue thermogram by RGB (red, green, blue) based color histogram image segmentation method and extracted the features using gray level co-occurrence matrix algorithm, (iii) classified normal and diabetes using various machine learning algorithms, and (iv) developed computer aided diagnostic system to classify diabetes mellitus. The baseline measurements and tongue thermograms were obtained from 140 subjects. The measured tongue surface temperature of the diabetic group was found to be greater than normal. The statistical correlation between the HbA_1c and the thermal distribution in the tongue region was found to be r² = 0.5688. The Convolutional Neural Network has outperformed the other classifiers with 94.28% accuracy rate. Thus, tongue thermograms could be used as a preliminary screening approach for diabetes prognosis.

Artificial intelligence with multi-functional machine learning platform development for better healthcare and precision medicine

Precision medicine is one of the recent and powerful developments in medical care, which has the potential to improve the traditional symptom-driven practice of medicine, allowing earlier interventions using advanced diagnostics and tailoring better and economically personalized treatments. Identifying the best pathway to personalized and population medicine involves the ability to analyze comprehensive patient information together with broader aspects to monitor and distinguish between sick and relatively healthy people, which will lead to a better understanding of biological indicators that can signal shifts in health. While the complexities of disease at the individual level have made it difficult to utilize healthcare information in clinical decision-making, some of the existing constraints have been greatly minimized by technological advancements. To implement effective precision medicine with enhanced ability to positively impact patient outcomes and provide real-time decision support, it is important to harness the power of electronic health records by integrating disparate data sources and discovering patient-specific patterns of disease progression. Useful analytic tools, technologies, databases, and approaches are required to augment networking and interoperability of clinical, laboratory and public health systems, as well as addressing ethical and social issues related to the privacy and protection of healthcare data with effective balance. Developing multifunctional machine learning platforms for clinical data extraction, aggregation, management and analysis can support clinicians by efficiently stratifying subjects to understand specific scenarios and optimize decision-making. Implementation of artificial intelligence in healthcare is a compelling vision that has the potential in leading to the significant improvements for achieving the goals of providing real-time, better personalized and population medicine at lower costs. In this study, we focused on analyzing and discussing various published artificial intelligence and machine learning solutions, approaches and perspectives, aiming to advance academic solutions in paving the way for a new data-centric era of discovery in healthcare.

Finding Diagnostically Useful Patterns in Quantitative Phenotypic Data

Trio-based whole-exome sequence (WES) data have established confident genetic diagnoses in ∼40% of previously undiagnosed individuals recruited to the Deciphering Developmental Disorders (DDD) study. Here we aim to use the breadth of phenotypic information recorded in DDD to augment diagnosis and disease variant discovery in probands. Median Euclidean distances (mEuD) were employed as a simple measure of similarity of quantitative phenotypic data within sets of ≥10 individuals with plausibly causative de novo mutations (DNM) in 28 different developmental disorder genes. 13/28 (46.4%) showed significant similarity for growth or developmental milestone metrics, 10/28 (35.7%) showed similarity in HPO term usage, and 12/28 (43%) showed no phenotypic similarity. Pairwise comparisons of individuals with high-impact inherited variants to the 32 individuals with causative DNM in ANKRD11 using only growth z-scores highlighted 5 likely causative inherited variants and two unrecognized DNM resulting in an 18% diagnostic uplift for this gene. Using an independent approach, naive Bayes classification of growth and developmental data produced reasonably discriminative models for the 24 DNM genes with sufficiently complete data. An unsupervised naive Bayes classification of 6,993 probands with WES data and sufficient phenotypic information defined 23 in silico syndromes (ISSs) and was used to test a “phenotype first” approach to the discovery of causative genotypes using WES variants strictly filtered on allele frequency, mutation consequence, and evidence of constraint in humans. This highlighted heterozygous de novo nonsynonymous variants in SPTBN2 as causative in three DDD probands.

Bidirectional General Graphs for inference. Principles and implications for medicine

Probabilistic inference methods require a more general and realistic description of the world as a Bidirectional General Graph (BGG). While in its original form the Bayes Net (BN) has been promoted as a predictive tool, it is more immediately a way of testing a hypothesis or model about interactions in a system usually considered on a causal basis. Once established, the model can be used in a predictive way, but the problem here is that for a traditional BN the hypotheses or models that can be formed are limited to the Directed Acyclic Graph (DAG) by definition. Three interrelated features are highlighted that represent deficiencies of the DAG which are corrected by conversion to a method based on a BGG: (i) lack of intrinsic representation of coherence by Bayes' rule, (ii) relatedly the need to consider interdependence in parent nodes, and (iii) the need for management of a property called recurrence. These deficiencies can represent large errors in absolute estimates of probabilities, and while relative and renormalized probabilities ameliorate that, they can often make much of a net superfluous through cancelations by division. The Hyperbolic Dirac Net (HDN) based on Dirac's quantum mechanics is a solution that led naturally to avoiding these deficiencies. It encodes bidirectional probabilities in an h-complex value rediscovered by Dirac, i.e. with the imaginary number h such that hh = +1. Properties of the HDN described previously are reviewed (though emphasis is on descriptions in familiar probability terms), the issue of recurrence is introduced, methods of construction are simplified, and the severity of the quantitative differences between BNs and analogous HDNs are exemplified. There is also discussion of how results compare with other approaches in practice.

Early Identification of High-Risk TIA or Minor Stroke Using Artificial Neural Network

Background and Purpose: The risk of recurrent stroke following a transient ischemic attack (TIA) or minor stroke is high, despite of a significant reduction in the past decade. In this study, we investigated the feasibility of using artificial neural network (ANN) for risk stratification of TIA or minor stroke patients.

Methods: Consecutive patients with acute TIA or minor ischemic stroke presenting at a tertiary hospital during a 2-year period were recruited. We collected demographics, clinical and imaging data at baseline. The primary outcome was recurrent ischemic stroke within 1 year. We developed ANN models to predict the primary outcome. We randomly down-sampled patients without a primary outcome to 1:1 match with those with a primary outcome to mitigate data imbalance. We used a 5-fold cross-validation approach to train and test the ANN models to avoid overfitting. We employed 19 independent variables at baseline as the input neurons in the ANN models, using a learning algorithm based on backpropagation to minimize the loss function. We obtained the sensitivity, specificity, accuracy and the c statistic of each ANN model from the 5 rounds of cross-validation and compared that of support vector machine (SVM) and Naïve Bayes classifier in risk stratification of the patients.

Results: A total of 451 acute TIA or minor stroke patients were enrolled. Forty (8.9%) patients had a recurrent ischemic stroke within 1 year. Another 40 patients were randomly selected from those with no recurrent stroke, so that data from 80 patients in total were used for 5 rounds of training and testing of ANN models. The median sensitivity, specificity, accuracy and c statistic of the ANN models to predict recurrent stroke at 1 year was 75%, 75%, 75%, and 0.77, respectively. ANN model outperformed SVM and Naïve Bayes classifier in our dataset for predicting relapse after TIA or minor stroke.

Conclusion: This pilot study indicated that ANN may yield a novel and effective method in risk stratification of TIA and minor stroke. Further studies are warranted for verification and improvement of the current ANN model.

Mortality prediction in patients with isolated moderate and severe traumatic brain injury using machine learning models

BACKGROUND

The purpose of this study was to build a model of machine learning (ML) for the prediction of mortality in patients with isolated moderate and severe traumatic brain injury (TBI).

METHODS

Hospitalized adult patients registered in the Trauma Registry System between January 2009 and December 2015 were enrolled in this study. Only patients with an Abbreviated Injury Scale (AIS) score ≥ 3 points related to head injuries were included in this study. A total of 1734 (1564 survival and 170 non-survival) and 325 (293 survival and 32 non-survival) patients were included in the training and test sets, respectively.

RESULTS

Using demographics and injury characteristics, as well as patient laboratory data, predictive tools (e.g., logistic regression [LR], support vector machine [SVM], decision tree [DT], naive Bayes [NB], and artificial neural networks [ANN]) were used to determine the mortality of individual patients. The predictive performance was evaluated by accuracy, sensitivity, and specificity, as well as by area under the curve (AUC) measures of receiver operator characteristic curves. In the training set, all five ML models had a specificity of more than 90% and all ML models (except the NB) achieved an accuracy of more than 90%. Among them, the ANN had the highest sensitivity (80.59%) in mortality prediction. Regarding performance, the ANN had the highest AUC (0.968), followed by the LR (0.942), SVM (0.935), NB (0.908), and DT (0.872). In the test set, the ANN had the highest sensitivity (84.38%) in mortality prediction, followed by the SVM (65.63%), LR (59.38%), NB (59.38%), and DT (43.75%).

CONCLUSIONS

The ANN model provided the best prediction of mortality for patients with isolated moderate and severe TBI.

Development of a store-and-forward telescreening system of diabetic retinopathy: lessons learned from Iran

Background

The present study describes the development and identity phases of a teleophthalmology system used for screening of diabetic retinopathy.

Methods

A questionnaire was used to identify the main factors necessary for diagnosis of diabetic retinopathy and the features required for a teleophthalmology system. In the second phase, a web-based prototype of the system was designed using the data collected in the previous phase. In the final phase, the system was optimized based on the users’ ideas and comments; then, it was evaluated through a standard usability testing questionnaire.

Results

The results showed that the lowest average percentages were related to ethnicity (61%), optometrist’s office address (61%), and consultant physician’s office address (65%). A web-based prototype was designed using the Visual Studio and Dreamweaver programming tools. This system comprised patient identity data, medical history, clinical data, and retinal images of the patient. The mean score of usability testing and user satisfaction including specialists, residents, and optometrist was 7.3, 7.1 and 7.3 (out of a total 9), respectively. The evaluation results revealed that the system was classified as good.

Conclusion

The telescreening system suggested in the current study could be helpful in timely diagnosis. Moreover, it would reduce the treatment costs and complexities. Regardless of the positive points of telemedicine systems, one of the most challenges in this study was the Internet infrastructure to design and apply the system. The future studies, therefore, could focus on the application of cell phone technology for rendering teleophthalmology.

Application of Ethics for Providing Telemedicine Services and Information Technology

INTRODUCTION

Advanced technology has increased the use of telemedicine and Information Technology (IT) in treating or rehabilitating diseases. An increased use of technology increases the importance of the ethical issues involved. The need for keeping patients' information confidential and secure, controlling a number of therapists' inefficiency as well as raising the quality of healthcare services necessitates adequate heed to ethical issues in telemedicine provision.

AIM

The goal of this review is gathering all articles that are published through 5 years until now (2012-2017) for detecting ethical issues for providing telemedicine services and Information technology. The reason of this time is improvement of telemedicine and technology through these years. This article is important for clinical practice and also to world, because of knowing ethical issues in telemedicine and technology are always important factors for physician and health providers.

MATERIAL AND METHODS

the required data in this research were derived from published electronic sources and credible academic articles published in such databases as PubMed, Scopus and Science Direct. The following key words were searched for in separation and combination: tele-health, telemedicine, ethical issues in telemedicine. A total of 503 articles were found. After excluding the duplicates (n= 93), the titles and abstracts of 410 articles were skimmed according to the inclusion criteria. Finally, 64 articles remained. They were reviewed in full text and 36 articles were excluded. At the end, 28 articles were chosen which met our eligibility criteria and were included in this study.

RESULTS

Ethics has been of a great significance in IT and telemedicine especially the Internet since there are more chances provided for accessing information. It is, however, accompanied by a threat to patients' personal information. Therefore, suggestions are made to investigate ethics in technology, to offer standards and guidelines to therapists. Due to the advancement in technology, access to information has become simpler than the past. This has prompted hackers to seize the opportunity.

DISCUSSION

This research shows that the ethical issues in telemedicine can be investigated from several aspects like technology, doctor-patient relationship, data confidentiality and security, informed consent, patient's and family's satisfaction with telemedicine services. Following ethical issues in telemedicine is a primary aspect of high quality services. In other words, if therapists abide by ethical rules, they can provide better services for patients. Attention to ethical issues in telemedicine guarantees a safer use of the services.