Development of a graphical model of causal gene regulatory networks using medical big data and Bayesian machine learning

Data collection from medicine and biomedical science is becoming a large task and increasingly complicated with each passing day. Machine learning methods have been applied to elucidate interactions between genes and genes and their environment.Current Concepts: Many machine learning methods have been used to determine the statistical meaning or relationship in the prediction or progression of diseases through the creation of causal networks based on medical big data. Through these analyses, the occurrence and progression of diseases have been shown to be related to several genes and environmental factors. However, these methods cannot identify the key upstream regulators inferred from genomic, clinical, and environmental medical data.Discussion and Conclusion: The causal Bayesian network (CBN) is a machine learning method that can be used to understand a causal network inferred from the gene expression data. The CBN can help identify the key upstream regulators through examining the causal network inferred from medical big data having genomic information. We can easily improve the clinical outcome through regulation of these identified key upstream factors. Therefore, the CBN may be a powerful and flexible tool in the era of precision medicine.

Erratum to: Causal Inference Network of Genes Related with Bone Metastasis of Breast Cancer and Osteoblasts Using Causal Bayesian Networks

The Acknowledgement was published incorrectly.

Causal Inference Network of Genes Related with Bone Metastasis of Breast Cancer and Osteoblasts Using Causal Bayesian Networks

BACKGROUND: The causal networks among genes that are commonly expressed in osteoblasts and during bone metastasis (BM) of breast cancer (BC) are not well understood. Here, we developed a machine learning method to obtain a plausible causal network of genes that are commonly expressed during BM and in osteoblasts in BC. METHODS: We selected BC genes that are commonly expressed during BM and in osteoblasts from the Gene Expression Omnibus database. Bayesian Network Inference with Java Objects (Banjo) was used to obtain the Bayesian network. Genes registered as BC related genes were included as candidate genes in the implementation of Banjo. Next, we obtained the Bayesian structure and assessed the prediction rate for BM, conditional independence among nodes, and causality among nodes. Furthermore, we reported the maximum relative risks (RRs) of combined gene expression of the genes in the model. RESULTS: We mechanistically identified 33 significantly related and plausibly involved genes in the development of BC BM. Further model evaluations showed that 16 genes were enough for a model to be statistically significant in terms of maximum likelihood of the causal Bayesian networks (CBNs) and for correct prediction of BM of BC. Maximum RRs of combined gene expression patterns showed that the expression levels of UBIAD1, HEBP1, BTNL8, TSPO, PSAT1, and ZFP36L2 significantly affected development of BM from BC. CONCLUSIONS: The CBN structure can be used as a reasonable inference network for accurately predicting BM in BC.

Development of Decision Support Formulas for the Prediction of Bladder Outlet Obstruction and Prostatic Surgery in Patients With Lower Urinary Tract Symptom/Benign Prostatic Hyperplasia: Part I, Development of the Formula and its Internal Validation / 대한배뇨장애요실금학회지

PURPOSE: As the elderly population increases, a growing number of patients have lower urinary tract symptom (LUTS)/benign prostatic hyperplasia (BPH). The aim of this study was to develop decision support formulas and nomograms for the prediction of bladder outlet obstruction (BOO) and for BOO-related surgical decision-making, and to validate them in patients with LUTS/BPH. METHODS: Patient with LUTS/BPH between October 2004 and May 2014 were enrolled as a development cohort. The available variables included age, International Prostate Symptom Score, free uroflowmetry, postvoid residual volume, total prostate volume, and the results of a pressure-flow study. A causal Bayesian network analysis was used to identify relevant parameters. Using multivariate logistic regression analysis, formulas were developed to calculate the probabilities of having BOO and requiring prostatic surgery. Patients between June 2014 and December 2015 were prospectively enrolled for internal validation. Receiver operating characteristic curve analysis, calibration plots, and decision curve analysis were performed. RESULTS: A total of 1,179 male patients with LUTS/BPH, with a mean age of 66.1 years, were included as a development cohort. Another 253 patients were enrolled as an internal validation cohort. Using multivariate logistic regression analysis, 2 and 4 formulas were established to estimate the probabilities of having BOO and requiring prostatic surgery, respectively. Our analysis of the predictive accuracy of the model revealed area under the curve values of 0.82 for BOO and 0.87 for prostatic surgery. The sensitivity and specificity were 53.6% and 87.0% for BOO, and 91.6% and 50.0% for prostatic surgery, respectively. The calibration plot indicated that these prediction models showed a good correspondence. In addition, the decision curve analysis showed a high net benefit across the entire spectrum of probability thresholds. CONCLUSIONS: We established nomograms for the prediction of BOO and BOO-related prostatic surgery in patients with LUTS/BPH. Internal validation of the nomograms demonstrated that they predicted both having BOO and requiring prostatic surgery very well.

Development of Decision Support Formulas for the Prediction of Bladder Outlet Obstruction and Prostatic Surgery in Patients With Lower Urinary Tract Symptom/Benign Prostatic Hyperplasia: Part II, External Validation and Usability Testing of a Smartphone A / 대한배뇨장애요실금학회지

PURPOSE: We aimed to externally validate the prediction model we developed for having bladder outlet obstruction (BOO) and requiring prostatic surgery using 2 independent data sets from tertiary referral centers, and also aimed to validate a mobile app for using this model through usability testing. METHODS: Formulas and nomograms predicting whether a subject has BOO and needs prostatic surgery were validated with an external validation cohort from Seoul National University Bundang Hospital and Seoul Metropolitan Government-Seoul National University Boramae Medical Center between January 2004 and April 2015. A smartphone-based app was developed, and 8 young urologists were enrolled for usability testing to identify any human factor issues of the app. RESULTS: A total of 642 patients were included in the external validation cohort. No significant differences were found in the baseline characteristics of major parameters between the original (n=1,179) and the external validation cohort, except for the maximal flow rate. Predictions of requiring prostatic surgery in the validation cohort showed a sensitivity of 80.6%, a specificity of 73.2%, a positive predictive value of 49.7%, and a negative predictive value of 92.0%, and area under receiver operating curve of 0.84. The calibration plot indicated that the predictions have good correspondence. The decision curve showed also a high net benefit. Similar evaluation results using the external validation cohort were seen in the predictions of having BOO. Overall results of the usability test demonstrated that the app was user-friendly with no major human factor issues. CONCLUSIONS: External validation of these newly developed a prediction model demonstrated a moderate level of discrimination, adequate calibration, and high net benefit gains for predicting both having BOO and requiring prostatic surgery. Also a smartphone app implementing the prediction model was user-friendly with no major human factor issue.

Factors Influencing Nonabsolute Indications for Surgery in Patients With Lower Urinary Tract Symptoms Suggestive of Benign Prostatic Hyperplasia: Analysis Using Causal Bayesian Networks / 대한배뇨장애요실금학회지

PURPOSE: To identify the factors affecting the surgical decisions of experienced physicians when treating patients with lower urinary tract symptoms that are suggestive of benign prostatic hyperplasia (LUTS/BPH). METHODS: Patients with LUTS/BPH treated by two physicians between October 2004 and August 2013 were included in this study. The causal Bayesian network (CBN) model was used to analyze factors influencing the surgical decisions of physicians and the actual performance of surgery. The accuracies of the established CBN models were verified using linear regression (LR) analysis. RESULTS: A total of 1,108 patients with LUTS/BPH were analyzed. The mean age and total prostate volume (TPV) were 66.2 (+/-7.3, standard deviation) years and 47.3 (+/-25.4) mL, respectively. Of the total 1,108 patients, 603 (54.4%) were treated by physician A and 505 (45.6%) were treated by physician B. Although surgery was recommended to 699 patients (63.1%), 589 (53.2%) actually underwent surgery. Our CBN model showed that the TPV (R=0.432), treating physician (R=0.370), bladder outlet obstruction (BOO) on urodynamic study (UDS) (R=0.324), and International Prostate Symptom Score (IPSS) question 3 (intermittency; R=0.141) were the factors directly influencing the surgical decision. The transition zone volume (R=0.396), treating physician (R=0.340), and BOO (R=0.300) directly affected the performance of surgery. Compared to the LR model, the area under the receiver operating characteristic curve of the CBN surgical decision model was slightly compromised (0.803 vs. 0.847, P<0.001), whereas that of the actual performance of surgery model was similar (0.801 vs. 0.820, P=0.063) to the LR model. CONCLUSIONS: The TPV, treating physician, BOO on UDS, and the IPSS item of intermittency were factors that directly influenced decision-making in physicians treating patients with LUTS/BPH.

Big Data Analysis Using Modern Statistical and Machine Learning Methods in Medicine / 대한배뇨장애요실금학회지

In this article we introduce modern statistical machine learning and bioinformatics approaches that have been used in learning statistical relationships from big data in medicine and behavioral science that typically include clinical, genomic (and proteomic) and environmental variables. Every year, data collected from biomedical and behavioral science is getting larger and more complicated. Thus, in medicine, we also need to be aware of this trend and understand the statistical tools that are available to analyze these datasets. Many statistical analyses that are aimed to analyze such big datasets have been introduced recently. However, given many different types of clinical, genomic, and environmental data, it is rather uncommon to see statistical methods that combine knowledge resulting from those different data types. To this extent, we will introduce big data in terms of clinical data, single nucleotide polymorphism and gene expression studies and their interactions with environment. In this article, we will introduce the concept of well-known regression analyses such as linear and logistic regressions that has been widely used in clinical data analyses and modern statistical models such as Bayesian networks that has been introduced to analyze more complicated data. Also we will discuss how to represent the interaction among clinical, genomic, and environmental data in using modern statistical models. We conclude this article with a promising modern statistical method called Bayesian networks that is suitable in analyzing big data sets that consists with different type of large data from clinical, genomic, and environmental data. Such statistical model form big data will provide us with more comprehensive understanding of human physiology and disease.