Machine learning refers to a series of processes in which a computer finds rules from a vast amount of data. With recent advances in computer technology and the availability of a wide variety of health data, machine learning has rapidly developed and been applied in medical research. Currently, there are three types of machine learning: supervised, unsupervised, and reinforcement learning. In medical research, supervised learning is commonly used for diagnoses and prognoses, while unsupervised learning is used for phenotyping a disease, and reinforcement learning for maximizing favorable results, such as optimization of total patients’ waiting time in the emergency department. The present article focuses on the concept and application of supervised learning in medicine, the most commonly used machine learning approach in medicine, and provides a brief explanation of four algorithms widely used for prediction (random forests, gradient-boosted decision tree, support vector machine, and neural network). Among these algorithms, the neural network has further developed into deep learning algorithms to solve more complex tasks. Along with simple classification problems, deep learning is commonly used to process medical imaging, such as retinal fundus photographs for diabetic retinopathy diagnosis. Although machine learning can bring new insights into medicine by processing a vast amount of data that are often beyond human capacity, algorithms can also fail when domain knowledge is neglected. The combination of algorithms and human cognitive ability is a key to the successful application of machine learning in medicine.
Clinical prediction models include a diagnostic prediction model to estimate the probability of an individual currently having a disease (e.g., pulmonary embolism) and a prognostic prediction model to estimate the probability of an individual developing a specific health outcome over a specific time period (e.g., myocardial infarction and stroke in 10 years). Clinical prediction models can be developed by applying traditional regression models (e.g., logistic and Cox regression models) or emerging machine learning models to real-world data, such as electronic health records and administrative claims data. For derivation, researchers select candidate variables based on a literature review and clinical knowledge, and predictor variables in the final model based on pre-defined criteria (e.g., thresholds for the size of relative risk and p-values) or strategies such as the stepwise regression and the least absolute shrinkage and selection operator (LASSO) regression. For validation, the clinical prediction model’s performance is evaluated in terms of goodness of fit (e.g., R2), discrimination (e.g., area under the receiver operating characteristic curve or c-statistics), and calibration (e.g., calibration plot and Hosmer-Lemeshow test). Performance of a new variable added to an existing clinical prediction model is evaluated in terms of reclassification (e.g., net reclassification improvement and integrated discrimination improvement). The model should be validated using the original data to examine internal validity through methods such as resampling (e.g., cross-validation and bootstrapping) and using other participants’ data to examine external validity. For successful implementation of a clinical prediction model in actual clinical practice, presentation methods such as paper-based (nomogram) or web-based calculator and an easy-to-use risk score should be considered.
The controversy concerning the benefits of pulmonary artery catheter (PAC)-based hemodynamic monitoring in cardiac surgeries has not been adequately addressed. This study aims to compare the all-cause mortality between the PAC with venous oxygen saturation monitoring and the Vigileo/FloTrac (FloTrac) system with central venous oxygen saturation monitoring in cardiac surgeries.
This nationwide retrospective study includes adult patients who underwent elective cardiac surgeries between April 2010 and October 2014, based on the Japanese health insurance claims database. The main outcome was 30-day all-cause mortality. Propensity scores (PS) were used to adjust for the confounding factors. Treatment effects were estimated using multivariable logistic regression analysis, including PS.
A total of 5,838 patients were included in this study. The crude 30-day mortality rates were 2.4% (8/334) and 1.7% (96/5,504) in the FloTrac and PAC groups, respectively. After PS matching, the ORs for 30-day all-cause mortality, in-hospital mortality after PAC placement (vs. FloTrac) were 0.36 (95% CI: 0.05–2.37; p = 0.28) and 0.59 (95% CI: 0.16–2.20; p = 0.43), respectively. The amount of dobutamine was larger in the PAC group (281 ± 31 mg vs 155 ± 19 mg; p < 0.001). There were no significant differences in the amounts of other inotropes, the volume of fluids, or blood transfusions.
The association between PAC (with venous oxygen saturation monitoring) and mortality in patients who underwent elective cardiac surgeries was unclear compared to FloTrac (with central venous oxygen saturation monitoring). Additional investigation is needed to evaluate the benefits of PAC-specific hemodynamic parameters in this population.
Data on the evaluation of the clinical course of coronavirus disease 2019 (COVID-19) and the efficacy of treatments after hospitalization in Japan are limited.
This study aimed to construct a database of confirmed COVID-19 cases in Japan and promptly address unresolved research issues.
This multicenter observational study included patients who had a laboratory-confirmed severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection and were discharged from each participating institution between January 1 and September 31, 2020. We called for participating facilities and research proposals until the end of September 2020. The research steering committee members provided advice to co-investigators on refining their research proposals and analyses. After developing the research proposal, we collected clinical information, facility information, and laboratory data from each participating institution. Clinical information was also obtained from the Diagnosis Procedure Combination (DPC) data using a dedicated software called DPC hash application.
We planned to conduct an analysis based on the research proposal. Overall, 66 institutions from Japan announced their participation, and 102 research proposals were selected for the analyses. Research areas from the proposals included epidemiology, pathophysiology, therapeutic agents, ventilator settings, cost-benefit analyses, and prognosis prediction for COVID-19.
CONTRIBUTION AND SIGNIFICANCE TO THE FIELD
We have established an efficient data collection system and clinical research team for COVID-19 infection studies. The results of this study may be utilized in future response strategies for COVID-19.