2016-11-03T14:11:40

Cation disorder plays a major role in the performance deterioration of Cu2ZnSn(S,Se)4 (CZTSSe) photovoltaic (PV) absorbers. Unfortunately the quantitative impact of cation disorder in this material is not well understood. Here, we show that changes in microstate configurational entropy (S MCE), predicted by a combination of statistical mechanics, density functional theory, and phonon calculations, can quantitatively describe cation disorder and the associated properties in CZTSSe. For example, the predicted critical temperature of the second-order phase transition and the thermal expansion anomaly of CZTS based on S MCE are in good agreement with experiments. We further reveal that the more separated is the same kind of cations (Cu, Zn, or Sn), the lower the energy of the corresponding microstate will be, implying the maximum entropy probability distribution of cations at high temperatures. It is suggested that the introduction of S MCE can serve as a framework to quantitatively understand and tailor cation disorder and associated properties for synthesis of the high-quality CZTS-based solar cells.

Predicting success on a new step in therapy currently relies upon clinical expertise and heuristics. Even after the transition to bolus insulin, many people with diabetes fail to achieve glycemic control. Our objective was to develop a predictive model to classify patients into expected levels of success upon bolus insulin initiation. Machine learning methods were applied to a large nationally representative insurance claims database. Commercially insured adults with T2DM were identified at first observed use of bolus insulin injections. Those with ≥2 HbA1c values in the 15-month follow-up period (N=15,331) were included. HbA1c goal achievement was defined as <8% or at least a 1% reduction from baseline. We trained boosted decision tree ensembles (XGBoost) to assign people into Class 0 (never meeting goal), Class 1 (meeting but not maintaining goal), or Class 2 (meeting and maintaining goal) based on the demographic and clinical data available prior to initiating bolus insulin. Overall, the model’s ROC was 0.79 with greater performance on predicting those in Class 2 (ROC=0.92) than those in Classes 0 and 1 (ROC’s = 0.71 and 0.62, respectively). Among the patients with highest (top 10%) probability of being in each of the three categories, the model accurately separates Class 2 from the others but had difficulty distinguishing between Class 0 and 1 patients. Over 98% of patients predicted to be either Class 0 or Class 1 with high probability were true Class 0 or Class 1 patients. However, only 42% of those highly predicted as Class 0 were Class 0, with >50% being Class 1. Average HbA1c values in the year prior to starting bolus insulin was the most influential predictive factor. Results suggest that predictive modeling using routine healthcare data was reasonably accurate in classifying patients initiating bolus insulin who will achieve and maintain HbA1c goals. The model was less accurate differentiating between patients who never met and those who did not maintain goals. Incorporation of behavioral, psychosocial, or other factors could improve predictions. Disclosure E.L. Eby: Employee; Self; Eli Lilly and Company. N.R. Kelly: Consultant; Self; Eli Lilly and Company. Employee; Spouse/Partner; Optum, Inc. Employee; Self; Optum, Inc. Stock/Shareholder; Spouse/Partner; UnitedHealth Group. Stock/Shareholder; Self; UnitedHealth Group. M. Blodgett: None. C. Stubbins: None. E. Meadows: Employee; Self; Eli Lilly and Company. Employee; Spouse/Partner; Eli Lilly and Company. Stock/Shareholder; Spouse/Partner; Eli Lilly and Company. Stock/Shareholder; Self; Eli Lilly and Company. B.D. Benneyworth: Employee; Self; Eli Lilly and Company. D.E. Faries: Employee; Self; Eli Lilly and Company. Funding Eli Lilly and Company