•Uncertainty predictions in digital soil mapping are not optimally validated.•The prediction interval coverage probability cannot account for one-sided bias.•One-sided bias can occur with some ...probabilistic prediction models.•Other validation metrics from the broader probabilistic literature are preferable.
It is quite common in digital soil mapping (DSM) to quantify the uncertainty of issued predictions, that is to make probabilistic predictions. Yet, little attention has been paid to its validation. Probabilistic predictions are only of value for end users if they are reliable and ideally also sharp. Reliability refers to the consistency between predicted conditional probabilities and observed frequencies of independent test data. Sharpness refers to the concentration of a conditional probability distribution function, i.e. its narrowness. The prediction interval coverage probability (PICP) is currently used in DSM to validate the reliability of prediction intervals but it is ignorant of a potential one-sided bias of its boundaries. Therefore, we propose to extend the current validation procedure with metrics used in the broader probabilistic literature. These metrics not only evaluate probabilistic predictions in prediction interval format but also quantiles or full conditional probability distributions. We suggest the quantile coverage probability (QCP) and probability integral transform (PIT) histogram as alternatives to PICP and proper scoring rules for relative comparisons of competing probabilistic models. As scoring rules, we present the interval score (IS) and the continuous ranked probability score (CRPS), which can be decomposed into a reliability part (RELI). We illustrated the use of these metrics in a case study using soil pH and soil organic carbon from the LUCAS-soil database. Thereby, probabilistic predictions of five different models were compared: a reference null model (NM), quantile regression forest (QRF), quantile regression post-processing of a random forest (QRPP RF), kriging with external drift (KED) and quantile regression neural network (QRNN). For KED and QRNN, one-sided bias was found. This was not apparent from PICP but was shown by use of the PIT histogram and QCP. RELI summarized the trends found in QCP, PICP and PIT histograms to one numerical value. CRPS and IS were especially harsh to outliers and low sharpness. According to CRPS and IS, the best probabilistic predictions were obtained by QRF and QRPP RF and the worst by NM.
Site-specific estimation of lime requirement requires high-resolution maps of soil organic carbon (SOC), clay and pH. These maps can be generated with digital soil mapping models fitted on covariates ...observed by proximal soil sensors. However, the quality of the derived maps depends on the applied methodology. We assessed the effects of (i) training sample size (5–100); (ii) sampling design (simple random sampling (SRS), conditioned Latin hypercube sampling (cLHS) and k-means sampling (KM)); and (iii) prediction model (multiple linear regression (MLR) and random forest (RF)) on the prediction performance for the above mentioned three soil properties. The case study is based on conditional geostatistical simulations using 250 soil samples from a 51 ha field in Eastern Germany. Lin’s concordance correlation coefficient (CCC) and root-mean-square error (RMSE) were used to evaluate model performances. Results show that with increasing training sample sizes, relative improvements of RMSE and CCC decreased exponentially. We found the lowest median RMSE values with 100 training observations i.e., 1.73%, 0.21% and 0.3 for clay, SOC and pH, respectively. However, already with a sample size of 10, models of moderate quality (CCC > 0.65) were obtained for all three soil properties. cLHS and KM performed significantly better than SRS. MLR showed lower median RMSE values than RF for SOC and pH for smaller sample sizes, but RF outperformed MLR if at least 25–30 or 75–100 soil samples were used for SOC or pH, respectively. For clay, the median RMSE was lower with RF, regardless of sample size.
Patients with cancer are at risk of venous thromboembolism (VTE). Tumor-related factors could help estimate patients' individual risk for VTE. Currently, only scarce information on the association ...between tumor grade and VTE is available. We thus evaluated the role of tumor grade and its association with VTE.
The Vienna Cancer and Thrombosis Study is a prospective, observational cohort study including patients with newly diagnosed cancer or progression of disease after remission. Study end point is the occurrence of symptomatic VTE.
Seven hundred forty-seven patients with solid tumors received follow-up for a median of 526 days. VTE occurred in 52 patients (7.0%). At study inclusion, 468 patients had low-grade tumors (G1 and G2) and 279 had high-grade tumors (G3 and G4). In multivariable Cox regression analysis including tumor grade, tumor histology, tumor sites, stage, sex, and age, patients with high-grade tumors had a significantly higher risk of VTE compared with those with low-grade tumors (hazard ratio, 2.0; 95% CI, 1.1 to 3.5; P = .015). The cumulative probability of developing VTE after 6 months was higher in patients with high-grade tumors than in those with low-grade tumors (8.2% v 4.0%; log-rank test P = .037). Patients with high-grade tumors had higher D-dimer levels (P = .008) and leukocyte counts (P < .001), and lower hemoglobin levels (P = .008).
The tumor grade may help identify patients with cancer who are at high risk of VTE. The association of tumor grade with recently identified biomarkers indicates a link between tumor differentiation and pathogenesis of cancer-associated VTE.
Summary Background Nivolumab plus ipilimumab is approved as first-line regimen for intermediate-risk or poor-risk metastatic renal cell carcinoma, and nivolumab monotherapy as second-line therapy for ...all risk groups. We aimed to examine the efficacy and safety of nivolumab monotherapy and nivolumab plus ipilimumab combination as an immunotherapeutic boost after no response to nivolumab monotherapy in patients with intermediate-risk and poor-risk clear-cell metastatic renal cell carcinoma. Methods TITAN-RCC is a multicentre, single-arm, phase 2 trial, done at 28 hospitals and cancer centres across Europe (Austria, Belgium, Czech Republic, France, Germany, Italy, Spain, and the UK). Adults (aged ≥18 years) with histologically confirmed intermediate-risk or poor-risk clear-cell metastatic renal cell carcinoma who were formerly untreated (first-line population) or pretreated with one previous systemic therapy (anti-angiogenic or temsirolimus; second-line population) were eligible. Patients had to have a Karnofsky Performance Status score of at least 70 and measurable disease per Response Evaluation Criteria in Solid Tumours (version 1.1). Patients started with intravenous nivolumab 240 mg once every 2 weeks. On early progressive disease (week 8) or non-response at week 16, patients received two or four doses of intravenous nivolumab (3 mg/kg) and ipilimumab (1 mg/kg) boosts (once every 3 weeks), whereas responders continued with intravenous nivolumab (240 mg, once every 2 weeks), but could receive two to four boost doses of nivolumab plus ipilimumab for subsequent progressive disease. The primary endpoint was confirmed investigator-assessed objective response rate in the full analysis set, which included all patients who received at least one dose of study medication; safety was also assessed in this population. An objective response rate of more than 25% was required to reject the null hypothesis and show improvement, on the basis of results from the pivotal phase 3 CheckMate-025 trial. This study is registered with ClinicalTrials.gov, NCT02917772, and is complete. Findings Between Oct 28, 2016, and Nov 30, 2018, 207 patients were enrolled and all received nivolumab induction (109 patients in the first-line group; 98 patients in the second-line group). 60 (29%) of 207 patients were female and 147 (71%) were male. 147 (71%) of 207 patients had intermediate-risk metastatic renal cell carcinoma and 51 (25%) had poor-risk disease. After median follow-up of 27·6 months (IQR 10·5–34·8), 39 (36%, 90% CI 28–44; p=0·0080) of 109 patients in the first-line group and 31 (32%, 24–40; p=0·083) of 98 patients in the second-line group had a confirmed objective response for nivolumab with and without nivolumab plus ipilimumab. Confirmed response to nivolumab at week 8 or 16 was observed in 31 (28%) of 109 patients in the first-line group and 18 (18%) of 98 patients in the second-line group. The most frequent grade 3–4 treatment-related adverse events (reported in ≥5% of patients) were increased lipase (15 7% of 207 patients), colitis (13 6%), and diarrhoea (13 6%). Three deaths were reported that were deemed to be treatment-related: one due to possible ischaemic stroke, one due to respiratory failure, and one due to pneumonia. Interpretation In treatment-naive patients, nivolumab induction with or without nivolumab plus ipilimumab boosts significantly improved the objective response rate compared with that reported for nivolumab monotherapy in the CheckMate-025 trial. However, overall efficacy seemed inferior when compared with approved upfront nivolumab plus ipilimumab. For second-line treatment, nivolumab plus ipilimumab could be a rescue strategy on progression with approved nivolumab monotherapy. Funding Bristol Myers Squibb.