Interpretation Method for Continuous Glucose Monitoring with Subsequence Time-Series Clustering.

We propose mini-batch top-n k-medoids to sequential pattern mining to improve CGM interpretation. Mecical workers can treat specific patient groups better by understanding the time series variation of blood glucose results. For 10 years, continuous glucose monitoring (CGM) has provided time-series data of blood glucose thanks to the invention of devices with low measurement errors. We conducted two experiments. In the first experiment, we evaluated the proposed method with a manually created dataset and confirmed that the method provides more accurate patterns than other clustering methods. In the second experiment, we applied the proposed method to a CGM dataset consisting of real data from 163 patients. We created two labels based on blood glucose (BG) statistics and found patterns that correlated with a specific label in each case.

Feature Set for a Prediction Model of Diabetic Kidney Disease Progression.

In this paper, we propose feature extraction method for prediction model for at the early stage of diabetic kidney disease (DKD) progression. DKD needs continuous treatment; however, a hospital visit interval of a patient at the early stage of DKD is normally from one month to three months, and this is not a short time period. Therefore it makes difficult to apply sophisticated approaches such as using convolutional neural networks because of the data limitation. The propose method uses with hierarchical clustering that can estimate a suitable interval for grouping inputted sequences. We evaluate the proposed method with a real-EMR dataset that consists of 30,810 patient records and conclude that the proposed method outperforms the baseline methods derived from related work.

Artificial intelligence predicts the progression of diabetic kidney disease using big data machine learning.

Artificial intelligence (AI) is expected to support clinical judgement in medicine. We constructed a new predictive model for diabetic kidney diseases (DKD) using AI, processing natural language and longitudinal data with big data machine learning, based on the electronic medical records (EMR) of 64,059 diabetes patients. AI extracted raw features from the previous 6 months as the reference period and selected 24 factors to find time series patterns relating to 6-month DKD aggravation, using a convolutional autoencoder. AI constructed the predictive model with 3,073 features, including time series data using logistic regression analysis. AI could predict DKD aggravation with 71% accuracy. Furthermore, the group with DKD aggravation had a significantly higher incidence of hemodialysis than the non-aggravation group, over 10 years (N = 2,900). The new predictive model by AI could detect progression of DKD and may contribute to more effective and accurate intervention to reduce hemodialysis.

Risk Prediction of Diabetic Nephropathy via Interpretable Feature Extraction from EHR Using Convolutional Autoencoder.

This paper describes a technology for predicting the aggravation of diabetic nephropathy from electronic health record (EHR). For the prediction, we used features extracted from event sequence of lab tests in EHR with a stacked convolutional autoencoder which can extract both local and global temporal information. The extracted features can be interpreted as similarities to a small number of typical sequences of lab tests, that may help us to understand the disease courses and to provide detailed health guidance. In our experiments on real-world EHRs, we confirmed that our approach performed better than baseline methods and that the extracted features were promising for understanding the disease.

Posterior-mean super-resolution with a causal Gaussian Markov random field prior.

We propose a Bayesian image super-resolution (SR) method with a causal Gaussian Markov random field (MRF) prior. SR is a technique to estimate a spatially high-resolution image from given multiple low-resolution images. An MRF model with the line process supplies a preferable prior for natural images with edges. We improve the existing image transformation model, the compound MRF model, and its hyperparameter prior model. We also derive the optimal estimator--not the joint maximum a posteriori (MAP) or the marginalized maximum likelihood (ML) but the posterior mean (PM)--from the objective function of the L2-norm-based (mean square error) peak signal-to-noise ratio. Point estimates such as MAP and ML are generally not stable in ill-posed high-dimensional problems because of overfitting, whereas PM is a stable estimator because all the parameters in the model are evaluated as distributions. The estimator is numerically determined by using the variational Bayesian method. The variational Bayesian method is a widely used method that approximately determines a complicated posterior distribution, but it is generally hard to use because it needs the conjugate prior. We solve this problem with simple Taylor approximations. Experimental results have shown that the proposed method is more accurate or comparable to existing methods.