Incorporating Glucose Variability into Glucose Forecasting Accuracy Assessment Using the New Glucose Variability Impact Index and the Prediction Consistency Index: An LSTM Case Example

BACKGROUND

In this work, we developed glucose forecasting algorithms trained and evaluated on a large dataset of free-living people with type 1 diabetes (T1D) using closed-loop (CL) and sensor-augmented pump (SAP) therapies; and we demonstrate how glucose variability impacts accuracy. We introduce the glucose variability impact index (GVII) and the glucose prediction consistency index (GPCI) to assess the accuracy of prediction algorithms.

METHODS

A long-short-term-memory (LSTM) neural network was designed to predict glucose up to 60 minutes in the future using continuous glucose measurements and insulin data collected from 175 people with T1D (41,318 days) and evaluated on 75 people (11,333 days) from the Tidepool Big Data Donation Dataset. LSTM was compared with two naïve forecasting algorithms as well as Ridge linear regression and a random forest using root-mean-square error (RMSE). Parkes error grid quantified clinical accuracy. Regression analysis was used to derive the GVII and GPCI.

RESULTS

The LSTM had highest accuracy and best GVII and GPCI. RMSE for CL was 19.8 ± 3.2 and 33.2 ± 5.4 mg/dL for 30- and 60-minute prediction horizons, respectively. RMSE for SAP was 19.6 ± 3.8 and 33.1 ± 7.3 mg/dL for 30- and 60-minute prediction horizons, respectively; 99.6% and 97.6% of predictions were within zones A+B of the Parkes error grid at 30- and 60-minute prediction horizons, respectively. Glucose variability was strongly correlated with RMSE (R≥0.64, P < 0.001); GVII and GPCI demonstrated a means to compare algorithms across datasets with different glucose variability.

CONCLUSIONS

The LSTM model was accurate on a large real-world free-living dataset. Glucose variability should be considered when assessing prediction accuracy using indices such as GVII and GPCI.

geneBasis: an iterative approach for unsupervised selection of targeted gene panels from scRNA-seq

scRNA-seq datasets are increasingly used to identify gene panels that can be probed using alternative technologies, such as spatial transcriptomics, where choosing the best subset of genes is vital. Existing methods are limited by a reliance on pre-existing cell type labels or by difficulties in identifying markers of rare cells. We introduce an iterative approach, geneBasis, for selecting an optimal gene panel, where each newly added gene captures the maximum distance between the true manifold and the manifold constructed using the currently selected gene panel. Our approach outperforms existing strategies and can resolve cell types and subtle cell state differences.

Feasibility Studies of an Insulin-Only Bionic Pancreas in a Home-Use Setting

BACKGROUND

We tested the safety and performance of the "insulin-only" configuration of the bionic pancreas (BP) closed-loop blood-glucose control system in a home-use setting to assess glycemic outcomes using different static and dynamic glucose set-points.

METHOD

This is an open-label non-randomized study with three consecutive intervention periods. Participants had consecutive weeks of usual care followed by the insulin-only BP with (1) an individualized static set-point of 115 or 130 mg/dL and (2) a dynamic set-point that automatically varied within 110 to 130 mg/dL, depending on hypoglycemic risk. Human factors (HF) testing was conducted using validated surveys. The last five days of each study arm were used for data analysis.

RESULTS

Thirteen participants were enrolled with a mean age of 28 years, mean A1c of 7.2%, and mean daily insulin dose of 0.6 U/kg (0.4-1.0 U/kg). The usual care arm had an average glucose of 145 ± 20 mg/dL, which increased in the static set-point arm (159 ± 8 mg/dL, P = .004) but not in the dynamic set-point arm (154 ± 10 mg/dL, P = ns). There was no significant difference in time spent in range (70-180 mg/dL) among the three study arms. There was less time <70 mg/dL with both the static (1.8% ± 1.4%, P = .009) and dynamic set-point (2.7±1.5, P = .051) arms compared to the usual-care arm (5.5% ± 4.2%). HF testing demonstrated preliminary user satisfaction and no increased risk of diabetes burden or distress.

CONCLUSIONS

The insulin-only configuration of the BP using either static or dynamic set-points and initialized only with body weight performed similarly to other published insulin-only systems.