Comparison of machine-learning and logistic regression models for prediction of 30-day unplanned readmission in electronic health records: A development and validation study.

It is expected but unknown whether machine-learning models can outperform regression models, such as a logistic regression (LR) model, especially when the number and types of predictor variables increase in electronic health records (EHRs). We aimed to compare the predictive performance of gradient-boosted decision tree (GBDT), random forest (RF), deep neural network (DNN), and LR with the least absolute shrinkage and selection operator (LR-LASSO) for unplanned readmission. We used EHRs of patients discharged alive from 38 hospitals in 2015-2017 for derivation and in 2018 for validation, including basic characteristics, diagnosis, surgery, procedure, and drug codes, and blood-test results. The outcome was 30-day unplanned readmission. We created six patterns of data tables having different numbers of binary variables (that ≥5% or ≥1% of patients or ≥10 patients had) with and without blood-test results. For each pattern of data tables, we used the derivation data to establish the machine-learning and LR models, and used the validation data to evaluate the performance of each model. The incidence of outcome was 6.8% (23,108/339,513 discharges) and 6.4% (7,507/118,074 discharges) in the derivation and validation datasets, respectively. For the first data table with the smallest number of variables (102 variables that ≥5% of patients had, without blood-test results), the c-statistic was highest for GBDT (0.740), followed by RF (0.734), LR-LASSO (0.720), and DNN (0.664). For the last data table with the largest number of variables (1543 variables that ≥10 patients had, including blood-test results), the c-statistic was highest for GBDT (0.764), followed by LR-LASSO (0.755), RF (0.751), and DNN (0.720), suggesting that the difference between GBDT and LR-LASSO was small and their 95% confidence intervals overlapped. In conclusion, GBDT generally outperformed LR-LASSO to predict unplanned readmission, but the difference of c-statistic became smaller as the number of variables was increased and blood-test results were used.

Height, body mass index, physical activity, and risk of colorectal cancer in relation to expression of insulin receptor: The Japan Public Health Center-based Prospective Study.

To ascertain the involvement of insulin receptors (IRs) in colorectal carcinogenesis, we investigated the association of height, body mass index (BMI), and physical activity with colorectal cancer (CRC) and two subtypes of CRC according to the expression level of IR. We utilized data from a large-scale, population-based prospective cohort study of 18,158 middle-aged and elderly subjects in Akita and Okinawa, Japan. In the statistical analysis, we used the Cox proportional hazards model and estimated hazard ratios (HRs) and 95% confidence intervals (CIs) for the risk of CRC and its subtypes as defined by immunohistochemistry of IRß, a transmembrane subunit of IR. In the IRß-defined subtypes, height showed no apparent association with the risk of IRß-positive CRC. In contrast, a multivariable HR of IRß-positive CRC was 1.77 (95% CI = 1.04-3.03) with a BMI of ≥30.0 kg/m2 (i.e., obesity), compared to a BMI of <25.0 kg/m2. Further, an increase in physical activity was significantly associated with decreased risk of IRß-positive CRC (multivariable HR per 5 METs-hour/day = 0.93, 95% CI = 0.88-0.99). Meanwhile, we found no significant association between any exposure and IRß-negative CRC. Likewise, heterogeneity between the two subtypes of CRC was not statistically significant. These findings imply that obesity and physical activity exert promoting and suppressing effects on the development of CRC expressing IRs, respectively.

Importance of circulating leptin and adiponectin in the causal pathways between obesity and the development of colorectal cancer in Japanese men.

BACKGROUND: The mechanistic associations between obesity and risk of colorectal cancer (CRC) remain unclear. Here, using body mass index (BMI) as an obesity indicator, we decomposed the total effects of obesity on the risk of CRC into: (1) direct effects, which are possibly mediated by unmeasured or currently unknown factors; (2) indirect effects mediated by circulating leptin and adiponectin; and (3) indirect effects that are not mediated by circulating leptin and adiponectin but by hyperinsulinemia and chronic inflammation (assessed via circulating connecting peptide and C-reactive protein, respectively). METHODS: We adopted a causal mediation framework, using data from a large prospective cohort study of 44,271 Japanese men. RESULTS: BMI was not associated with the risk of CRC due to direct and indirect effects that were not mediated by circulating leptin and adiponectin. By contrast, individuals with BMIs of 25.0-27.4 kg/m2 (risk ratio, 1.29; 95% confidence interval, 0.98-1.69) and ≥27.5 kg/m2 (risk ratio, 1.28; 95% confidence interval, 0.98-1.68) had a higher risk of CRC due to indirect effects of circulating leptin and adiponectin. CONCLUSIONS: Our mediation analyses suggest that the association between BMI and CRC risk may be largely mediated by a pathway involving circulating leptin and adiponectin.

The epidemiology of postoperative dobutamine and phosphodiesterase inhibitors after adult elective cardiac surgery and its impact on the length of hospital stay: a post hoc analysis from the multicenter retrospective observational study.

The optimal administration of inotrope after cardiac surgery is unknown. This study aimed to investigate the impact of postoperative inotrope on clinical outcomes in adult elective cardiac surgery patients. Data from the Blood Pressure and Relative Optimal Target after Heart Surgery in Epidemiologic Registry study were analyzed, employing propensity score considering the hospital of admission. The primary outcome was the length of hospital stay evaluated using quantile regression. Secondary outcomes were kidney injury progression, renal replacement therapy, atrial fibrillation, mortality, mechanical ventilation duration, and length of intensive care unit (ICU) stay. Among 870 patients from 14 ICUs in Japan, 535 received inotropes within 24 h of ICU admission, with usage rates ranging from 40 to 100% among facilities. After propensity score matching, 218 patients were included in each group. The inotrope group had a significantly longer hospital stay compared to the control group (16 days vs. 14 days; median difference 1.78 [95% confidence interval [CI] 0.31-3.24]; p = 0.018). However, no significant differences were observed in the secondary outcomes, except for mechanical ventilation duration. The results of the sensitivity analysis using a mixed-effects quantile regression analysis considering the hospital of admission for length of hospital stay in the original cohort were consistent with the results of the propensity analyses (median difference in days, 2.35 [95% CI, 0.35-4.36]; p = 0.022). The use of inotropes within 24 h of ICU admission in adult elective cardiac surgery patients was associated with an extended hospitalization period of approximately 2 days, without offering any prognostic benefit. Clinical trial registration: UMIN-CTR, https://www.umin.ac.jp/ctr/index-j.htm , UMIN000037074.

Genetic architecture of alcohol consumption identified by a genotype-stratified GWAS and impact on esophageal cancer risk in Japanese people.

An East Asian-specific variant on aldehyde dehydrogenase 2 (ALDH2 rs671, G>A) is the major genetic determinant of alcohol consumption. We performed an rs671 genotype-stratified genome-wide association study meta-analysis of alcohol consumption in 175,672 Japanese individuals to explore gene-gene interactions with rs671 behind drinking behavior. The analysis identified three genome-wide significant loci (GCKR, KLB, and ADH1B) in wild-type homozygotes and six (GCKR, ADH1B, ALDH1B1, ALDH1A1, ALDH2, and GOT2) in heterozygotes, with five showing genome-wide significant interaction with rs671. Genetic correlation analyses revealed ancestry-specific genetic architecture in heterozygotes. Of the discovered loci, four (GCKR, ADH1B, ALDH1A1, and ALDH2) were suggested to interact with rs671 in the risk of esophageal cancer, a representative alcohol-related disease. Our results identify the genotype-specific genetic architecture of alcohol consumption and reveal its potential impact on alcohol-related disease risk.

Third Report of the Japan Diabetes Society (JDS)/Japanese Cancer Association (JCA) Joint Committee on diabetes and cancer: summary of the results of a questionnaire survey of oncologists and diabetologists-secondary publication.

The Japan Diabetes Society (JDS) and the Japan Cancer Association (JCA) launched a joint committee and published their "First Joint Committee Report on Diabetes and Cancer" in 2013, compiling recommendations for physicians and healthcare providers as well as for the general population. In 2016, the "Second Joint Committee Report on Diabetes and Cancer" summarized the current evidence on glycemic control and cancer risk in patients with diabetes. The current "Third Joint Committee Report on Diabetes and Cancer", for which the joint committee also enlisted the assistance of the Japanese Society of Clinical Oncology (JSCO) and the Japanese Society of Medical Oncology (JSMO), reports on the results from the questionnaire survey, "Diabetes Management in Patients Receiving Cancer Therapy," which targeted oncologists responsible for cancer management and diabetologists in charge of glycemic control in cancer patients. The results of the current survey demonstrated that there is a general consensus among oncologists and diabetologists with regard to the need for guidelines on glycemic control goals, the relevance of glycemic control, and glycemic control during cancer therapy in cancer patients.

Third Report of the Japan Diabetes Society/Japanese Cancer Association Joint Committee on Diabetes and Cancer: Summary of the results of a questionnaire survey of oncologists and diabetologists-Secondary publication.

The Japan Diabetes Society and the Japan Cancer Association launched a joint committee and published their "First Joint Committee Report on Diabetes and Cancer" in 2013, compiling recommendations for physicians and health-care providers as well as for the general population. In 2016, the "Second Joint Committee Report on Diabetes and Cancer" summarized the current evidence on glycemic control and cancer risk in patients with diabetes. The current "Third Joint Committee Report on Diabetes and Cancer", for which the joint committee also enlisted the assistance of the Japanese Society of Clinical Oncology and the Japanese Society of Medical Oncology, reports on the results from the questionnaire survey, "Diabetes Management in Patients Receiving Cancer Therapy," which targeted oncologists responsible for cancer management and diabetologists in charge of glycemic control in cancer patients. The results of the current survey indicated that there is a general consensus among oncologists and diabetologists with regard to the need for guidelines on glycemic control goals, the relevance of glycemic control, and glycemic control during cancer therapy in cancer patients.