Impact of heart failure on reoperation in adult congenital heart disease: An innovative machine learning model.

OBJECTIVES: The study objectives were to evaluate the association between preoperative heart failure and reoperative cardiac surgical outcomes in adult congenital heart disease and to develop a risk model for postoperative morbidity/mortality. METHODS: Single-institution retrospective cohort study of adult patients with congenital heart disease undergoing reoperative cardiac surgery between January 1, 2010, and March 30, 2022. Heart failure defined clinically as preoperative diuretic use and either New York Heart Association Class II to IV or systemic ventricular ejection fraction less than 40%. Composite outcome included operative mortality, mechanical circulatory support, dialysis, unplanned noncardiac reoperation, persistent neurologic deficit, and cardiac arrest. Multivariable logistic regression and machine learning analysis using gradient boosting technology were performed. Shapley statistics determined feature influence, or impact, on model output. RESULTS: Preoperative heart failure was present in 376 of 1011 patients (37%); those patients had longer postoperative length of stay (6 [5-8] vs 5 [4-7] days, P < .001), increased postoperative mechanical circulatory support (21/376 [6%] vs 16/635 [3%], P = .015), and decreased long-term survival (84% [80%-89%] vs 90% [86%-93%]) at 10 years (P = .002). A 7-feature machine learning risk model for the composite outcome achieved higher area under the curve (0.76) than logistic regression, and ejection fraction was most influential (highest mean |Shapley value|). Additional risk factors for the composite outcome included age, number of prior cardiopulmonary bypass operations, urgent/emergency procedure, and functionally univentricular physiology. CONCLUSIONS: Heart failure is common among adult patients with congenital heart disease undergoing cardiac reoperation and associated with longer length of stay, increased postoperative mechanical circulatory support, and decreased long-term survival. Machine learning yields a novel 7-feature risk model for postoperative morbidity/mortality, in which ejection fraction was the most influential.

A machine learning model of response to hypomethylating agents in myelodysplastic syndromes.

Hypomethylating agents (HMA) prolong survival and improve cytopenias in individuals with higher-risk myelodysplastic syndrome (MDS). Only 30-40% of patients, however, respond to HMAs, and responses may not occur for more than 6 months after HMA initiation. We developed a model to more rapidly assess HMA response by analyzing early changes in patients' blood counts. Three institutions' data were used to develop a model that assessed patients' response to therapy 90 days after the initiation using serial blood counts. The model was developed with a training cohort of 424 patients from 2 institutions and validated on an independent cohort of 90 patients. The final model achieved an area under the receiver operating characteristic curve (AUROC) of 0.79 in the train/test group and 0.84 in the validation group. The model provides cohort-wide and individual-level explanations for model predictions, and model certainty can be interrogated to gauge the reliability of a given prediction.

Artificial Intelligence in Oncology: Current Capabilities, Future Opportunities, and Ethical Considerations.

The promise of highly personalized oncology care using artificial intelligence (AI) technologies has been forecasted since the emergence of the field. Cumulative advances across the science are bringing this promise to realization, including refinement of machine learning- and deep learning algorithms; expansion in the depth and variety of databases, including multiomics; and the decreased cost of massively parallelized computational power. Examples of successful clinical applications of AI can be found throughout the cancer continuum and in multidisciplinary practice, with computer vision-assisted image analysis in particular having several U.S. Food and Drug Administration-approved uses. Techniques with emerging clinical utility include whole blood multicancer detection from deep sequencing, virtual biopsies, natural language processing to infer health trajectories from medical notes, and advanced clinical decision support systems that combine genomics and clinomics. Substantial issues have delayed broad adoption, with data transparency and interpretability suffering from AI's "black box" mechanism, and intrinsic bias against underrepresented persons limiting the reproducibility of AI models and perpetuating health care disparities. Midfuture projections of AI maturation involve increasing a model's complexity by using multimodal data elements to better approximate an organic system. Far-future positing includes living databases that accumulate all aspects of a person's health into discrete data elements; this will fuel highly convoluted modeling that can tailor treatment selection, dose determination, surveillance modality and schedule, and more. The field of AI has had a historical dichotomy between its proponents and detractors. The successful development of recent applications, and continued investment in prospective validation that defines their impact on multilevel outcomes, has established a momentum of accelerated progress.

A geno-clinical decision model for the diagnosis of myelodysplastic syndromes.

The differential diagnosis of myeloid malignancies is challenging and subject to interobserver variability. We used clinical and next-generation sequencing (NGS) data to develop a machine learning model for the diagnosis of myeloid malignancies independent of bone marrow biopsy data based on a 3-institution, international cohort of patients. The model achieves high performance, with model interpretations indicating that it relies on factors similar to those used by clinicians. In addition, we describe associations between NGS findings and clinically important phenotypes and introduce the use of machine learning algorithms to elucidate clinicogenomic relationships.

Personalized Prediction Model to Risk Stratify Patients With Myelodysplastic Syndromes.

PURPOSE: Patients with myelodysplastic syndromes (MDS) have a survival that can range from months to decades. Prognostic systems that incorporate advanced analytics of clinical, pathologic, and molecular data have the potential to more accurately and dynamically predict survival in patients receiving various therapies. METHODS: A total of 1,471 MDS patients with comprehensively annotated clinical and molecular data were included in a training cohort and analyzed using machine learning techniques. A random survival algorithm was used to build a prognostic model, which was then validated in external cohorts. The accuracy of the proposed model, compared with other established models, was assessed using a concordance (c)index. RESULTS: The median age for the training cohort was 71 years. Commonly mutated genes included SF3B1, TET2, and ASXL1. The algorithm identified chromosomal karyotype, platelet, hemoglobin levels, bone marrow blast percentage, age, other clinical variables, seven discrete gene mutations, and mutation number as having prognostic impact on overall and leukemia-free survivals. The model was validated in an independent external cohort of 465 patients, a cohort of patients with MDS treated in a prospective clinical trial, a cohort of patients with paired samples at different time points during the disease course, and a cohort of patients who underwent hematopoietic stem-cell transplantation. CONCLUSION: A personalized prediction model on the basis of clinical and genomic data outperformed established prognostic models in MDS. The new model was dynamic, predicting survival and leukemia transformation probabilities at different time points that are unique for a given patient, and can upstage and downstage patients into more appropriate risk categories.

Machine learning integrates genomic signatures for subclassification beyond primary and secondary acute myeloid leukemia.

Although genomic alterations drive the pathogenesis of acute myeloid leukemia (AML), traditional classifications are largely based on morphology, and prototypic genetic founder lesions define only a small proportion of AML patients. The historical subdivision of primary/de novo AML and secondary AML has shown to variably correlate with genetic patterns. The combinatorial complexity and heterogeneity of AML genomic architecture may have thus far precluded genomic-based subclassification to identify distinct molecularly defined subtypes more reflective of shared pathogenesis. We integrated cytogenetic and gene sequencing data from a multicenter cohort of 6788 AML patients that were analyzed using standard and machine learning methods to generate a novel AML molecular subclassification with biologic correlates corresponding to underlying pathogenesis. Standard supervised analyses resulted in modest cross-validation accuracy when attempting to use molecular patterns to predict traditional pathomorphologic AML classifications. We performed unsupervised analysis by applying the Bayesian latent class method that identified 4 unique genomic clusters of distinct prognoses. Invariant genomic features driving each cluster were extracted and resulted in 97% cross-validation accuracy when used for genomic subclassification. Subclasses of AML defined by molecular signatures overlapped current pathomorphologic and clinically defined AML subtypes. We internally and externally validated our results and share an open-access molecular classification scheme for AML patients. Although the heterogeneity inherent in the genomic changes across nearly 7000 AML patients was too vast for traditional prediction methods, machine learning methods allowed for the definition of novel genomic AML subclasses, indicating that traditional pathomorphologic definitions may be less reflective of overlapping pathogenesis.

Personalized Prediction of Hospital Mortality in COVID-19-Positive Patients.

OBJECTIVE: To develop predictive models for in-hospital mortality and length of stay (LOS) for coronavirus disease 2019 (COVID-19)-positive patients. PATIENTS AND METHODS: We performed a multicenter retrospective cohort study of hospitalized COVID-19-positive patients. A total of 764 patients admitted to 14 different hospitals within the Cleveland Clinic from March 9, 2020, to May 20, 2020, who had reverse transcriptase-polymerase chain reaction-proven coronavirus infection were included. We used LightGBM, a machine learning algorithm, to predict in-hospital mortality at different time points (after 7, 14, and 30 days of hospitalization) and in-hospital LOS. Our final cohort was composed of 764 patients admitted to 14 different hospitals within our system. RESULTS: The median LOS was 5 (range, 1-44) days for patients admitted to the regular nursing floor and 10 (range, 1-38) days for patients admitted to the intensive care unit. Patients who died during hospitalization were older, initially admitted to the intensive care unit, and more likely to be white and have worse organ dysfunction compared with patients who survived their hospitalization. Using the 10 most important variables only, the final model's area under the receiver operating characteristics curve was 0.86 for 7-day, 0.88 for 14-day, and 0.85 for 30-day mortality in the validation cohort. CONCLUSION: We developed a decision tool that can provide explainable and patient-specific prediction of in-hospital mortality and LOS for COVID-19-positive patients. The model can aid health care systems in bed allocation and distribution of vital resources.

Genome Sequence of the Virulent Model Herpes Simplex Virus 1 Strain McKrae Demonstrates the Presence of at Least Two Widely Used Variant Strains.

Herpes simplex virus 1 (HSV-1) strain McKrae was isolated in 1965 and has been utilized by many laboratories. Three HSV-1 strain McKrae stocks have been sequenced previously, revealing discrepancies in key genes. We sequenced the genome of HSV-1 strain McKrae from the laboratory of James M. Hill to better understand the genetic differences between isolates.

The Evolving Landscape of Myelodysplastic Syndrome Prognostication.

Myelodysplastic syndromes (MDSs) are potentially devastating monoclonal deviations of hematopoiesis that lead to bone marrow dysplasia and variable cytopenias. Predicting severity of disease progression and likelihood to undergo acute myeloid leukemia transformation is the basis of treatment strategy. Some patients belong to a low-risk cohort best managed with conservative supportive care, whereas others are included in a high-risk cohort that requires decisive therapy with hematopoietic cell transplantation or hypomethylating agent administration. Risk scoring systems for MDS prognostication were traditionally based on karyotype characteristics and clinical factors readily available from chart review, and validation was typically conducted on de novo MDS patients. However, retrospective analysis found a large subset of patients incorrectly risk-stratified. In this review, the most commonly used scoring systems are evaluated, and pitfalls therein are identified. Emerging technologies such as personal genomics and machine learning are then explored for efficacy in MDS risk modeling. Barriers to clinical adoption of artificial intelligence-derived models are discussed, with focus on approaches meant to increase model interpretability and clinical relevance. Finally, a guiding set of recommendations is proposed for best designing an accurate and universally applicable prognostic model for MDS, which is supported by more than 20 years of observation of traditional scoring system performance, as well as modern efforts in creating hybrid genomic-clinical scoring systems.

Novel Prognostic Models for Myelodysplastic Syndromes.

Myelodysplastic syndromes are disorders of clonal myelopoiesis having a range of clinical manifestations, from benign and indolent to aggressive with very poor prognosis. Classifying the likely trajectory of disease within a patient largely guides therapeutic decision making and therefore survival. Traditional methods of risk-stratification systems rely on clinical features: simple blood tests, peripheral smears, bone marrow biopsies, and cytogenetics, but do not adequately predict disease severity for a substantial proportion of patients. This article reviews the state of stratification at use in the clinic, describes emerging systems that leverage large-scale genomic data, and summarizes efforts toward truly personalized prediction models.

Expression profiles of cell-wall related genes vary broadly between two common maize inbreds during stem development.

BACKGROUND: The cellular machinery for cell wall synthesis and metabolism is encoded by members of large multi-gene families. Maize is both a genetic model for grass species and a potential source of lignocellulosic biomass from crop residues. Genetic improvement of maize for its utility as a bioenergy feedstock depends on identification of the specific gene family members expressed during secondary wall development in stems. RESULTS: High-throughput sequencing of transcripts expressed in developing rind tissues of stem internodes provided a comprehensive inventory of cell wall-related genes in maize (Zea mays, cultivar B73). Of 1239 of these genes, 854 were expressed among the internodes at ≥95 reads per 20 M, and 693 of them at ≥500 reads per 20 M. Grasses have cell wall compositions distinct from non-commelinid species; only one-quarter of maize cell wall-related genes expressed in stems were putatively orthologous with those of the eudicot Arabidopsis. Using a slope-metric algorithm, five distinct patterns for sub-sets of co-expressed genes were defined across a time course of stem development. For the subset of genes associated with secondary wall formation, fifteen sequence motifs were found in promoter regions. The same members of gene families were often expressed in two maize inbreds, B73 and Mo17, but levels of gene expression between them varied, with 30% of all genes exhibiting at least a 5-fold difference at any stage. Although presence-absence and copy-number variation might account for much of these differences, fold-changes of expression of a CADa and a FLA11 gene were attributed to polymorphisms in promoter response elements. CONCLUSIONS: Large genetic variation in maize as a species precludes the extrapolation of cell wall-related gene expression networks even from one common inbred line to another. Elucidation of genotype-specific expression patterns and their regulatory controls will be needed for association panels of inbreds and landraces to fully exploit genetic variation in maize and other bioenergy grass species.

Modulation of nonessential amino acid biosynthetic pathways in virulent Hessian fly larvae (Mayetiola destructor), feeding on susceptible host wheat (Triticum aestivum).

Compatible interactions between wheat (Triticum aestivum), and its dipteran pest Hessian fly (Hf, Mayetiola destructor) result in successful establishment of larval feeding sites rendering the host plant susceptible. Virulent larvae employ an effector-based feeding strategy to reprogram the host physiology resulting in formation of a protein- and sugar-rich nutritive tissue beneficial to developing larvae. Previous studies documented increased levels of nonessential amino acids (NAA; that need not be received through insect diet) in the susceptible wheat in response to larval feeding, suggesting importance of plant-derived NAA in larval nutrition. Here, we investigated the modulation of genes from NAA biosynthetic pathways (NAABP) in virulent Hf larvae. Transcript profiling for 16 NAABP genes, annotated from the recently assembled Hf genome, was carried out in the feeding first-, and second-instars and compared with that of the first-instar neonate (newly hatched, migrating, assumed to be non-feeding) larvae. While Tyr, Gln, Glu, and Pro NAABP genes transcript abundance declined in the feeding instars as compared to the neonates, those for Ala, and Ser increased in the feeding larval instars, despite higher levels of these NAA in the susceptible host plant. Asp, Asn, Gly and Cys NAABP genes exhibited variable expression profiles in the feeding first- and second-instars. Our results indicate that while Hf larvae utilize the plant-derived NAA, de novo synthesis of several NAA may be necessary to: (i) provide larvae with the requisite amount for sustaining growth before nutritive tissue formation and, (ii) overcome any inadequate amounts in the host plant, post-nutritive tissue formation.

Fibrinolysis in Trauma: "Myth," "Reality," or "Something in Between".

The emphasis on fibrinolysis as an important contributor to trauma-induced coagulopathy (TIC) has led to a debate regarding the relative clinical significance of fibrinolysis in the setting of trauma. The debate has centered on two camps. The one camp defines fibrinolysis in trauma by standard coagulation tests as well as fibrin split products, D-dimers, and plasmin/antiplasmin levels. This camp favors a more liberal use of tranexamic acid and attributes more significance to hyperfibrinolysis in TIC. The other camp favors a definition of fibrinolysis based on the viscoelastic tests (VET), rotational thromboelastometry (ROTEM), and thromboelastography (TEG). These whole blood assays define hyperfibrinolysis at a higher threshold than plasma-based tests. Therefore, this VET camp reserves antifibrinolytic treatment for patients who demonstrate severe coagulopathy associated with hyperfibrinolysis. This bimodal attribution of the clinical relevance of fibrinolysis in trauma suggests that there may be an underlying "Myth" of the concept of TIC that was historically defined by plasma-based tests and a future "Reality" of the concept of TIC that is grounded on an understanding of TIC based on a VET-defined "fibrinolytic spectrum" of TIC. This narrative review explores this "Myth" and "Reality" of fibrinolysis in TIC and proposes a direction that will allow a "Future" interpretation of TIC that incorporates both the past "Myth" and present "Reality" of fibrinolysis TIC.

Tranexamic Acid for Trauma Resuscitation in the United States of America.

The utilization of tranexamic acid (TXA) for the management of bleeding trauma patients has been a subject of much debate on both sides of the Atlantic and in Australia. As a result of the large randomized controlled study called the Clinical Randomization of an Antifibrinolytic in Severe Hemorrhage (CRASH-2), there was an initial enthusiasm for the use of TXA to treat bleeding patients. However, the adoption of TXA in the United States was delayed by concerns of "knowledge and evidence gaps" of the CRASH-2 study and because of a lack of mechanistic rationale that would explain the survival benefit noted in the study. Subsequent nonrandomized controlled trials questioned the liberal use of TXA in trauma patients. This narrative review explores the historical as well as clinical and theoretical grounds for the more measured use of TXA in the United States and proposes a clinical and point-of-care guided utilization of TXA, blood components, and adjunctive hemostatic agents in bleeding trauma patients.

DNA from Dust: Comparative Genomics of Large DNA Viruses in Field Surveillance Samples.

The intensification of the poultry industry over the last 60 years facilitated the evolution of increased virulence and vaccine breaks in Marek's disease virus (MDV-1). Full-genome sequences are essential for understanding why and how this evolution occurred, but what is known about genome-wide variation in MDV comes from laboratory culture. To rectify this, we developed methods for obtaining high-quality genome sequences directly from field samples without the need for sequence-based enrichment strategies prior to sequencing. We applied this to the first characterization of MDV-1 genomes from the field, without prior culture. These viruses were collected from vaccinated hosts that acquired naturally circulating field strains of MDV-1, in the absence of a disease outbreak. This reflects the current issue afflicting the poultry industry, where virulent field strains continue to circulate despite vaccination and can remain undetected due to the lack of overt disease symptoms. We found that viral genomes from adjacent field sites had high levels of overall DNA identity, and despite strong evidence of purifying selection, had coding variations in proteins associated with virulence and manipulation of host immunity. Our methods empower ecological field surveillance, make it possible to determine the basis of viral virulence and vaccine breaks, and can be used to obtain full genomes from clinical samples of other large DNA viruses, known and unknown. IMPORTANCE Despite both clinical and laboratory data that show increased virulence in field isolates of MDV-1 over the last half century, we do not yet understand the genetic basis of its pathogenicity. Our knowledge of genome-wide variation between strains of this virus comes exclusively from isolates that have been cultured in the laboratory. MDV-1 isolates tend to lose virulence during repeated cycles of replication in the laboratory, raising concerns about the ability of cultured isolates to accurately reflect virus in the field. The ability to directly sequence and compare field isolates of this virus is critical to understanding the genetic basis of rising virulence in the wild. Our approaches remove the prior requirement for cell culture and allow direct measurement of viral genomic variation within and between hosts, over time, and during adaptation to changing conditions.

Viral forensic genomics reveals the relatedness of classic herpes simplex virus strains KOS, KOS63, and KOS79.

Herpes simplex virus 1 (HSV-1) is a widespread global pathogen, of which the strain KOS is one of the most extensively studied. Previous sequence studies revealed that KOS does not cluster with other strains of North American geographic origin, but instead clustered with Asian strains. We sequenced a historical isolate of the original KOS strain, called KOS63, along with a separately isolated strain attributed to the same source individual, termed KOS79. Genomic analyses revealed that KOS63 closely resembled other recently sequenced isolates of KOS and was of Asian origin, but that KOS79 was a genetically unrelated strain that clustered in genetic distance analyses with HSV-1 strains of North American/European origin. These data suggest that the human source of KOS63 and KOS79 could have been infected with two genetically unrelated strains of disparate geographic origins. A PCR RFLP test was developed for rapid identification of these strains.

Identification of a plastidial phenylalanine exporter that influences flux distribution through the phenylalanine biosynthetic network.

In addition to proteins, L-phenylalanine is a versatile precursor for thousands of plant metabolites. Production of phenylalanine-derived compounds is a complex multi-compartmental process using phenylalanine synthesized predominantly in plastids as precursor. The transporter(s) exporting phenylalanine from plastids, however, remains unknown. Here, a gene encoding a Petunia hybrida plastidial cationic amino-acid transporter (PhpCAT) functioning in plastidial phenylalanine export is identified based on homology to an Escherichia coli phenylalanine transporter and co-expression with phenylalanine metabolic genes. Radiolabel transport assays show that PhpCAT exports all three aromatic amino acids. PhpCAT downregulation and overexpression result in decreased and increased levels, respectively, of phenylalanine-derived volatiles, as well as phenylalanine, tyrosine and their biosynthetic intermediates. Metabolic flux analysis reveals that flux through the plastidial phenylalanine biosynthetic pathway is reduced in PhpCAT RNAi lines, suggesting that the rate of phenylalanine export from plastids contributes to regulating flux through the aromatic amino-acid network.

Rapid genome assembly and comparison decode intrastrain variation in human alphaherpesviruses.

UNLABELLED: Herpes simplex virus (HSV) is a widespread pathogen that causes epithelial lesions with recurrent disease that manifests over a lifetime. The lifelong aspect of infection results from latent viral infection of neurons, a reservoir from which the virus reactivates periodically. Recent work has demonstrated the breadth of genetic variation in globally distributed HSV strains. However, the amount of variation or capacity for mutation within one strain has not been well studied. Here we developed and applied a streamlined new approach for assembly and comparison of large DNA viral genomes such as HSV-1. This viral genome assembly (VirGA) workflow incorporates a combination of de novo assembly, alignment, and annotation strategies to automate the generation of draft genomes for large viruses. We applied this approach to quantify the amount of variation between clonal derivatives of a common parental virus stock. In addition, we examined the genetic basis for syncytial plaque phenotypes displayed by a subset of these strains. In each of the syncytial strains, we found an identical DNA change, affecting one residue in the gB (UL27) fusion protein. Since these identical mutations could have appeared after extensive in vitro passaging, we applied the VirGA sequencing and comparison approach to two clinical HSV-1 strains isolated from the same patient. One of these strains was syncytial upon first culturing; its sequence revealed the same gB mutation. These data provide insight into the extent and origin of genome-wide intrastrain HSV-1 variation and present useful methods for expansion to in vivo patient infection studies. IMPORTANCE: Herpes simplex virus (HSV) infects more than 70% of adults worldwide, causing epithelial lesions and recurrent disease that manifests over a lifetime. Prior work has demonstrated that HSV strains vary from country to country and between individuals. However, the amount of variation within one strain has not been well studied. To address this, we developed a new approach for viral genome assembly (VirGA) and analysis. We used this approach to quantify the amount of variation between sister clones of a common parental virus stock and to determine the basis of a unique fusion phenotype displayed by several variants. These data revealed that while sister clones of one HSV stock are more than 98% identical, these variants harbor enough genetic differences to change their observed characteristics. Comparative genomics approaches will allow us to explore the impacts of viral inter- and intrastrain diversity on drug and vaccine efficacy.

Metatranscriptomic profiles of Eastern subterranean termites, Reticulitermes flavipes (Kollar) fed on second generation feedstocks.

BACKGROUND: Second generation lignocellulosic feedstocks are being considered as an alternative to first generation biofuels that are derived from grain starches and sugars. However, the current pre-treatment methods for second generation biofuel production are inefficient and expensive due to the recalcitrant nature of lignocellulose. In this study, we used the lower termite Reticulitermes flavipes (Kollar), as a model to identify potential pretreatment genes/enzymes specifically adapted for use against agricultural feedstocks. RESULTS: Metatranscriptomic profiling was performed on worker termite guts after feeding on corn stover (CS), soybean residue (SR), or 98% pure cellulose (paper) to identify (i) microbial community, (ii) pathway level and (iii) gene-level responses. Microbial community profiles after CS and SR feeding were different from the paper feeding profile, and protist symbiont abundance decreased significantly in termites feeding on SR and CS relative to paper. Functional profiles after CS feeding were similar to paper and SR; whereas paper and SR showed different profiles. Amino acid and carbohydrate metabolism pathways were downregulated in termites feeding on SR relative to paper and CS. Gene expression analyses showed more significant down regulation of genes after SR feeding relative to paper and CS. Stereotypical lignocellulase genes/enzymes were not differentially expressed, but rather were among the most abundant/constitutively-expressed genes. CONCLUSIONS: These results suggest that the effect of CS and SR feeding on termite gut lignocellulase composition is minimal and thus, the most abundantly expressed enzymes appear to encode the best candidate catalysts for use in saccharification of these and related second-generation feedstocks. Further, based on these findings we hypothesize that the most abundantly expressed lignocellulases, rather than those that are differentially expressed have the best potential as pretreatment enzymes for CS and SR feedstocks.