Molecular Assays Simulator to Unravel Predictors Hacking in Goal-Directed Molecular Generations.

Generative models are being increasingly used in drug discovery, very often coupled with absorption, distribution, metabolism, and excretion (ADME) bioassays or quantitative structure-activity relationship (QSAR) models to optimize a given set of properties. The molecules proposed by these algorithms are often revealed to be false positives; that is, they are predicted to be active and turn out to be inactive after synthesis and testing, mostly due to overoptimization of the predicted scores, which leads to an actual decrease or stagnation of the real scores. This behavior is also known as the "hacking" of the predictive models by the generative model during the optimization step. This issue is reminiscent of adversarial examples in machine learning and it can be seen as enunciated by Goodhart's law: "when a measure becomes a target, it ceases to be a good measure." This issue is even more apparent in a multiparameter optimization (MPO) case, where the models need to extrapolate outside the training set distribution because there are no known molecules satisfying all the objectives simultaneously in the initial training set. Experimental evaluation of this problem is a hard and expensive task since it requires synthesis and testing of the generated molecules. Thus, efforts have been made to develop in silico "oracles"─real-valued functions used as proxies for molecular properties─to help with the evaluation of these generative-model-based pipelines. However, these oracles have had a limited value so far because they are often too easy to model in comparison with biological assays and are usually limited to mono-objective cases. In this work, we introduce a simulator of multitarget assays using a smartly initialized neural network (NN) that returns continuous values for any input molecule. We use this oracle to replicate a real-world prospective lead optimization (LO) scenario. First, we trained predictive models on an initial small sample of molecules aimed at predicting their oracle values. Afterward, we generated new optimized molecules using the open-source GuacaMol package coupled with the previously built predictive models. Finally, we selected compounds matching the candidate drug target profile (CDTP) according to the predicted values and evaluated them by computing the true oracle values. We observed that even when the predictive models had excellent estimated performance metrics, the final selection still contained multiple false positives according to the NN-based oracle. Then, we evaluated the optimization behavior in mono- and bi-objective scenarios using either a logistic regression or a random forest predictive model. We also propose and evaluate several methods to help mitigate the hacking issue.

Assessing the impact of transitional care units on dialysis patient outcomes: A multicenter, propensity score-matched analysis.

INTRODUCTION: Inadequate predialysis care and education impacts the selection of a dialysis modality and is associated with adverse clinical outcomes. Transitional care units (TCUs) aim to meet the unmet educational needs of incident dialysis patients, but their impact beyond increasing home dialysis utilization has been incompletely characterized. METHODS: This retrospective study included adults initiating in-center hemodialysis at a TCU, matched to controls (1:4) with no TCU history initiating in-center hemodialysis. Patients were followed for up to 14 months. TCUs are dedicated spaces where staff provide personalized education and as-needed adjustments to dialysis prescriptions. For many patients, therapy was initiated with four to five weekly dialysis sessions, with at least some sessions delivered by home dialysis machines. Outcomes included survival, first hospitalization, transplant waiting-list status, post-TCU dialysis modality, and vascular access type. FINDINGS: The study included 724 patients initiating dialysis across 48 TCUs, with 2892 well-matched controls. At the end of 14 months, patients initiating dialysis in a TCU were significantly more likely to be referred and/or wait-listed for a kidney transplant than controls (57% vs. 42%; p < 0.0001). Initiation of dialysis at a TCU was also associated with significantly lower rates of receiving in-center hemodialysis at 14 months (74% vs. 90%; p < 0.0001) and higher rates of arteriovenous access (70% vs. 63%; p = 0.003). Although not statistically significant, TCU patients were more likely to survive and less likely to be hospitalized during follow-up than controls. DISCUSSION: Although TCUs are sometimes viewed as only a means for enhancing utilization of home dialysis, patients attending TCUs exhibited more favorable outcomes across all endpoints. In addition to being 2.5-fold more likely to receive home dialysis, TCU patients were 42% more likely to be referred for transplantation. Our results support expanding utilization of TCUs for patients with inadequate predialysis support.

On the Frustration to Predict Binding Affinities from Protein-Ligand Structures with Deep Neural Networks.

Accurate prediction of binding affinities from protein-ligand atomic coordinates remains a major challenge in early stages of drug discovery. Using modular message passing graph neural networks describing both the ligand and the protein in their free and bound states, we unambiguously evidence that an explicit description of protein-ligand noncovalent interactions does not provide any advantage with respect to ligand or protein descriptors. Simple models, inferring binding affinities of test samples from that of the closest ligands or proteins in the training set, already exhibit good performances, suggesting that memorization largely dominates true learning in the deep neural networks. The current study suggests considering only noncovalent interactions while omitting their protein and ligand atomic environments. Removing all hidden biases probably requires much denser protein-ligand training matrices and a coordinated effort of the drug design community to solve the necessary protein-ligand structures.

The instantaneous retrieval of precipitation over land by temporal variation at 19 GHz.

The primary signal used in all current passive microwave precipitation retrieval algorithms over land is the depression of the instantaneous brightness temperature (TB) caused by ice scattering. This study presents a new methodology to retrieve instantaneous precipitation rate over land by using TB temporal variation (ΔTB) at 19 GHz, which primarily reflects the surface emissivity variation due to the precipitation impact. As a proof-of-concept, we exploit observations from five polar-orbiting satellites over the Southern Great Plains (SGP) of the United States. Results show that ΔTB at 19 GHz correlate well with the instantaneous precipitation rate. Further analysis shows that ΔTB at 19 GHz is better correlated with the precipitation rate when multiple satellite observations are used due to the much shorter re-visit time for a certain location. The retrieved instantaneous precipitation rate over SGP from ΔTB at 19 GHz reasonably agrees with the surface radar observations, with the correlation, the root mean square error and the bias being 0.49, 2.39 mm/hr and 6.54%, respectively. Future work seeks to combine the ice scattering signal at high frequencies and this surface emissivity variation signal at low frequencies to achieve an optimal retrieval performance.

The NxStage System One.

Given the results of recent randomized controlled trials as well as staffing and budget challenges that today face many institutions across North America, a novel therapeutic approach is likely necessary to enable improvements in clinical outcomes for renal failure patients. The NxStage System One was developed to address these challenges. The system is an innovative, flexible device that delivers hemodialysis, hemofiltration, and/or ultrafiltration therapies to patients with renal failure or fluid overload. The unique characteristics of this system include a highly automated system design with a drop-in cartridge to facilitate training and simple operation; portable size and independence from dedicated infrastructure to minimize practical barriers to where therapy may be administered; use of high-quality premixed treatment fluids to enable capture of the potential clinical benefits of fluid purity without the hassles of local water treatment; and wide operating ranges to allow clinician flexibility in patient therapy prescriptions. In both the chronic and acute care environments, the System One presents clinicians with a new platform for delivering patient therapy improvements within real-world constraints.

Dose determinants in continuous renal replacement therapy.

Increasing attention is being paid to quantifying the dose of dialysis prescribed and delivered to critically ill patients with acute renal failure (ARF). Recent trials in both the intermittent hemodialysis (IHD) and continuous renal replacement therapy (CRRT) realms have suggested that a direct relationship between dose and survival exists for both of these therapies. The purpose of this review, first, is to analyze critically the above-mentioned dose/outcome studies in acute dialysis. Subsequently, the factors influencing dose prescription and delivery are discussed, with the focus on continuous venovenous hemofiltration (CVVH). Specifically, differences between postdilution and predilution CVVH will be highlighted, and the importance of blood flow rate in dose delivery for these therapies will be discussed.