The 30-days hospital readmission risk in diabetic patients: predictive modeling with machine learning classifiers.

BACKGROUND AND OBJECTIVES: Diabetes mellitus is a major chronic disease that results in readmissions due to poor disease control. Here we established and compared machine learning (ML)-based readmission prediction methods to predict readmission risks of diabetic patients. METHODS: The dataset analyzed in this study was acquired from the Health Facts Database, which includes over 100,000 records of diabetic patients from 1999 to 2008. The basic data distribution characteristics of this dataset were summarized and then analyzed. In this study, 30-days readmission was defined as a readmission period of less than 30 days. After data preprocessing and normalization, multiple risk factors in the dataset were examined for classifier training to predict the probability of readmission using ML models. Different ML classifiers such as random forest, Naive Bayes, and decision tree ensemble were adopted to improve the clinical efficiency of the classification. In this study, the Konstanz Information Miner platform was used to preprocess and model the data, and the performances of the different classifiers were compared. RESULTS: A total of 100,244 records were included in the model construction after the data preprocessing and normalization. A total of 23 attributes, including race, sex, age, admission type, admission location, length of stay, and drug use, were finally identified as modeling risk factors. Comparison of the performance indexes of the three algorithms revealed that the RF model had the best performance with a higher area under receiver operating characteristic curve (AUC) than the other two algorithms, suggesting that its use is more suitable for making readmission predictions. CONCLUSION: The factors influencing 30-days readmission predictions in diabetic patients, including number of inpatient admissions, age, diagnosis, number of emergencies, and sex, would help healthcare providers to identify patients who are at high risk of short-term readmission and reduce the probability of 30-days readmission. The RF algorithm with the highest AUC is more suitable for making 30-days readmission predictions and  deserves further validation in clinical trials.

iT2DMS: a Standard-Based Diabetic Disease Data Repository and its Pilot Experiment on Diabetic Retinopathy Phenotyping and Examination Results Integration.

Type 2 diabetes mellitus (T2DM) is a common chronic disease, and the fragment data collected through separated vendors makes continuous management of DM patients difficult. The lack of standard of fragment data from those diabetic patients also makes the further potential phenotyping based on the diabetic data difficult. Traditional T2DM data repository only supports data collection from T2DM patients, lack of phenotyping ability and relied on standalone database design, limiting the secondary usage of these valuable data. To solve these issues, we proposed a novel T2DM data repository framework, which was based on standards. This repository can integrate data from various sources. It would be used as a standardized record for further data transfer as well as integration. Phenotyping was conducted based on clinical guidelines with KNIME workflow. To evaluate the phenotyping performance of the proposed system, data was collected from local community by healthcare providers and was then tested using algorithms. The results indicated that the proposed system could detect DR cases with an average accuracy of about 82.8%. Furthermore, these results had the promising potential of addressing fragmented data. The proposed system has integrating and phenotyping abilities, which could be used for diabetes research in future studies.

The clinical value of combination of immune checkpoint inhibitors in cancer patients: A meta-analysis of efficacy and safety.

The use of immune checkpoint inhibitors (ICIs) in combination therapy is an emerging trend in tumor immunology. However, the value of combination immunotherapy remains controversial, because of the toxic effects induced by combination. The added benefit of each additional drug has not been assessed against the added toxicity. We searched for clinical trials that evaluated ICI monotherapies and combination therapies in lung cancer and melanoma patients. The overall response rate (ORR), grade 3/4 treatment-related adverse event rate, overall survival (OS), and progression-free survival (PFS) were extracted from the most recently published studies to determine the relative risk (RR), hazard ratios (HRs), and 95% confidence intervals (CIs). Seven randomized controlled trials and one open-label study were identified (n = 3,097). Treatments included combinations of several ICIs, a combination of an ICI and dacarbazine, two combinations of an ICI, paclitaxel and carboplatin, and a combination of an ICI and gp100 vaccine. Higher ORR (RR: 1.51, 95% CI: 1.03-2.20, p = 0.034), OS (HR: 0.86, 95% CI: 0.78-0.95, p = 0.000), and PFS (HR: 0.93, 95% CI: 0.72-1.14, p = 0.000) values were observed in combination therapy than in monotherapy. In addition, the toxicity of combination ICI immunotherapy was higher (RR: 1.50, 95% CI: 1.03-2.19, p = 0.036) than that of monotherapy. This meta-analysis showed that the addition of nivolumab to ipilimumab better benefits PFS and ORR. Adding sargramostim was associated with better OS and safety. The efficacy and safety of a nivolumab-ipilimumab-sargramostim combination should be investigated further.

iT2DM Project: A Framework for Secondary Use of EHR Data for High-Throughput Phenotyping in Diabetes.

Diabetes is one of the major burdens in health care, but could be controlled if the relevant data are well-managed. Referring to current successful cases, we designed a framework for the interoperability and integration of medical data in compliance with both archetype and reference information model specification. The clinical data model (CDM) was designed on the basis of OpenEHR archetypes and self-made patient generated health data (PGHD). Integrating healthcare enterprise (IHE) protocol was taken into integrating different modality data. After terminology mapping, the personal health record could be transferred and shared in different clinical information vendors complying with HL7 standards. Many fragment data such as blood glucose and gene data were also integrated to system. Those patients suspected of higher risk of diabetic retinopathy (DR) were grouped as case and other patients could be filtered as control cohort. Furthermore, the framework could be further developed for precision medicine.