Estimating ecosystem evapotranspiration (ET) is important to understanding the global water cycle and to study land-atmosphere interactions. We developed a physics constrained machine learning (ML) model (hybrid model) to estimate latent heat flux (LE), which conserves the surface energy budget. By comparing model predictions with observations at 82 eddy covariance tower sites, our hybrid model shows similar performance to the pure ML model in terms of mean metrics (e.g., mean absolute percent errors) but, importantly, the hybrid model conserves the surface energy balance, while the pure ML model does not. A second key result is that the hybrid model extrapolates much better than the pure ML model, emphasizing the benefits of combining physics with ML for increased generalizations. The hybrid model allows inferring the structural dependence of ET and surface resistance (r s ), and we find that vegetation height and soil moisture are the main regulators of ET and r s . Plain Language SummaryA physics constrained machine learning model is developed using the FLUXNET2015 Tier 1 data set. This new approach is able to effectively retrieve latent heat flux while constraining energy conservation in the surface energy budget. This hybrid model has better performance in extrapolation than a pure machine learning model. Key Points: • A physics-constrained machine learning model of evapotranspiration (hybrid model) is developed and trained using the FLUXNET 2015 data set • The evapotranspiration retrieved by the hybrid model is as accurate as pure machine learning model and also conserves surface energy balance • The hybrid model better reproduces extremes and thus better extrapolates compared to the pure machine learning approach Supporting Information:• Supporting Information S1• Figure S1 • Table S1

No abstract

This study aims to construct a logistically derived additive score for predicting in-hospital mortality risk in Chinese patients undergoing coronary artery bypass surgery (CABG). Data from 9839 consecutive CABG patients in 43 Chinese centers were collected between 2007 and 2008 from the Chinese Coronary Artery Bypass Grafting Registry. This database was randomly divided into developmental and validation subsets (9:1). The data in the developmental dataset were used to develop the model using logistic regression. Calibration and discrimination characteristics were assessed using the validation dataset. Thresholds were defined for each model to distinguish different risk groups. After excluding 275 patients with incomplete information, the overall mortality rate of the remaining 9564 patients was 2.5%. The SinoSCORE model was constructed based on 11 variables: age, preoperative NYHA stage III or IV, chronic renal failure, extracardiac arteriopathy, chronic obstructive pulmonary disease, preoperative atrial fibrillation or flutter (within 2 weeks), left ventricular ejection fraction, other elective surgery, combined valve procedures, preoperative critical state, and BMI. In the developmental dataset, calibration using a Hosmer-Lemeshow (HL) test was at P = 0.44 and discrimination based on the area under the receiver operating characteristic curve (ROC) was 0.80. In the validation dataset, the HL test was at P = 0.34 and the area under the ROC (AUC) was 0.78. A logistically derived additive model for predicting in-hospital mortality among Chinese patients undergoing CABG was developed based on the most up-to-date multi-center data from China.

No abstract