Computational tools for exploring peptide-membrane interactions in gram-positive bacteria

The vital cellular functions in Gram-positive bacteria are controlled by signaling molecules known as quorum sensing peptides (QSPs), considered promising therapeutic interventions for bacterial infections. In the bacterial system QSPs bind to membrane-coupled receptors, which then auto-phosphorylate and activate intracellular response regulators. These response regulators induce target gene expression in bacteria. One of the most reliable trends in drug discovery research for virulence-associated molecular targets is the use of peptide drugs or new functionalities. In this perspective, computational methods act as auxiliary aids for biologists, where methodologies based on machine learning and in silico analysis are developed as suitable tools for target peptide identification. Therefore, the development of quick and reliable computational resources to identify or predict these QSPs along with their receptors and inhibitors is receiving considerable attention. The databases such as Quorumpeps and Quorum Sensing of Human Gut Microbes (QSHGM) provide a detailed overview of the structures and functions of QSPs. The tools and algorithms such as QSPpred, QSPred-FL, iQSP, EnsembleQS and PEPred-Suite have been used for the generic prediction of QSPs and feature representation. The availability of compiled key resources for utilizing peptide features based on amino acid composition, positional preferences, and motifs as well as structural and physicochemical properties, including biofilm inhibitory peptides, can aid in elucidating the QSP and membrane receptor interactions in infectious Gram-positive pathogens. Herein, we present a comprehensive survey of diverse computational approaches that are suitable for detecting QSPs and QS interference molecules. This review highlights the utility of these methods for developing potential biomarkers against infectious Gram-positive pathogens.

A novel image-based machine learning model with superior accuracy and predictability for knee arthroplasty loosening detection and clinical decision making

Background

Loosening is the leading cause of total knee arthroplasty (TKA) revision. This is a heavy burden toward the healthcare system owing to the difficulty in diagnosis and complications occurring from the delay management. Based on automatic analytical model building, machine learning, may potentially help to automatically recognize the risk of loosening based on radiographs alone. The aim of this study was to build an image-based machine-learning model for detecting TKA loosening.

Methods

Image-based machine-learning model was developed based on ImageNet, Xception model and a TKA patient X-ray image dataset. Based on a dataset with TKA patient clinical parameters, another system was then created for developing the clinical-information-based machine learning model with random forest classifier. In addition, the Xception Model was pre-trained on the ImageNet database with python and TensorFlow deep learning library for the prediction of loosening. Class activation maps were also used to interpret the prediction decision made by model. Two senior orthopaedic specialists were invited to assess loosening from X-ray images for 3 attempts in setting up comparison benchmark.

Result

In the image-based machine learning loosening model, the precision rate and recall rate were 0.92 and 0.96, respectively. While for the accuracy rate, 96.3% for visualization classification was observed. However, the addition of clinical-information-based model, with precision rate of 0.71 and recall rate of 0.20, did not further showed improvement on the accuracy. Moreover, as class activation maps showed corresponding signals over bone-implant interface that is loosened radiographically, this confirms that the current model utilized a similar image recognition pattern as that of inspection by clinical specialists.

Conclusion

The image-based machine learning model developed demonstrated high accuracy and predictability of knee arthroplasty loosening. And the class activation heatmap matched well with the radiographic features used clinically to detect loosening, which highlighting its potential role in assisting clinicians in their daily practice. However, addition of clinical-information-based machine-learning model did not offer further improvement in detection. As far as we know, this is the first report of pure image-based machine learning model with high detection accuracy. Importantly, this is also the first model to show relevant class activation heatmap corresponding to loosening location.

Translational potential

The finding in this study indicated image-based machine learning model can detect knee arthroplasty loosening with high accuracy and predictability, which the class activation heatmap can potentially assist surgeons to identify the sites of loosening.

Modelling diagnostics for Echinococcus granulosus surveillance in sheep using Latent Class Analysis: Argentina as a case study

Echinococcus granulosus sensu lato is a globally prevalent zoonotic parasitic cestode leading to cystic echinococcosis (CE) in both humans and sheep with both medical and financial impacts, whose reduction requires the application of a One Health approach to its control. Regarding the animal health component of this approach, lack of accurate and practical diagnostics in livestock impedes the assessment of disease burden and the implementation and evaluation of control strategies. We use of a Bayesian Latent Class Analysis (LCA) model to estimate ovine CE prevalence in sheep samples from the Río Negro province of Argentina accounting for uncertainty in the diagnostics. We use model outputs to evaluate the performance of a novel recombinant B8/2 antigen B subunit (rEgAgB8/2) indirect enzyme-linked immunosorbent assay (ELISA) for detecting E. granulosus in sheep. Necropsy (as a partial gold standard), western blot (WB) and ELISA diagnostic data were collected from 79 sheep within two Río Negro slaughterhouses, and used to estimate individual infection status (assigned as a latent variable within the model). Using the model outputs, the performance of the novel ELISA at both individual and flock levels was evaluated, respectively, using a receiver operating characteristic (ROC) curve, and simulating a range of sample sizes and prevalence levels within hypothetical flocks. The estimated (mean) prevalence of ovine CE was 27.5% (95%Bayesian credible interval (95%BCI): 13.8%-58.9%) within the sample population. At the individual level, the ELISA had a mean sensitivity and specificity of 55% (95%BCI: 46%-68%) and 68% (95%BCI: 63%-92%), respectively, at an optimal optical density (OD) threshold of 0.378. At the flock level, the ELISA had an 80% probability of correctly classifying infection at an optimal cut-off threshold of 0.496. These results suggest that the novel ELISA could play a useful role as a flock-level diagnostic for CE surveillance in the region, supplementing surveillance activities in the human population and thus strengthening a One Health approach. Importantly, selection of ELISA cut-off threshold values must be tailored according to the epidemiological situation.

Serum bilirubin to fetuin-A ratio as a prognostic biomarker in critically ill patients with sepsis

Liver dysfunction during sepsis is associated with increased bilirubin and decreased fetuin-A, a major hepatokine. We aimed to explore the association of bilirubin to fetuin-A (B/F) ratio early in sepsis with severity and outcome in critically ill patients. Based on a previous prospective study, we analyzed data of 90 critically ill patients (52 males, age: 65 ± 15 years, APACHE II: 24 ± 7 and SOFA: 10 ± 3) with sepsis. Bilirubin and fetuin-A increased during the first week of sepsis, (median (IQR) 0.45 (0.32-1) vs 0.55 (0.29-0.78) mg/dL, p = 0.03 and 302 (248-336) vs 358 (307-399) μg/mL, p < 0.001, respectively) while the B/F ratio did not change significantly. However, the B/F ratio at baseline and one week later was significantly higher in patients with septic shock (N = 38) and nonsurvivors (N = 28) compared to patients with sepsis (N = 52) and survivors (N = 62), respectively. The B/F ratio was positively associated with severity scores and outperformed bilirubin as a predictor of mortality in ROC curve analysis (AUC 0.78 (0.69-0.88), p < 0.001 and 0.69 (0.57-0.8), p = 0.003 respectively). The B/F ratio may be a promising sepsis biomarker with possible predictive value in critically ill patients.

Diagnosing Covid-19 chest x-rays with a lightweight truncated DenseNet with partial layer freezing and feature fusion

Due to the unforeseen turn of events, our world has undergone another global pandemic from a highly contagious novel coronavirus named COVID-19. The novel virus inflames the lungs similarly to Pneumonia, making it challenging to diagnose. Currently, the common standard to diagnose the virus's presence from an individual is using a molecular real-time Reverse-Transcription Polymerase Chain Reaction (rRT-PCR) test from fluids acquired through nasal swabs. Such a test is difficult to acquire in most underdeveloped countries with a few experts that can perform the test. As a substitute, the widely available Chest X-Ray (CXR) became an alternative to rule out the virus. However, such a method does not come easy as the virus still possesses unknown characteristics that even experienced radiologists and other medical experts find difficult to diagnose through CXRs. Several studies have recently used computer-aided methods to automate and improve such diagnosis of CXRs through Artificial Intelligence (AI) based on computer vision and Deep Convolutional Neural Networks (DCNN), which some require heavy processing costs and other tedious methods to produce. Therefore, this work proposed the Fused-DenseNet-Tiny, a lightweight DCNN model based on a densely connected neural network (DenseNet) truncated and concatenated. The model trained to learn CXR features based on transfer learning, partial layer freezing, and feature fusion. Upon evaluation, the proposed model achieved a remarkable 97.99 % accuracy, with only 1.2 million parameters and a shorter end-to-end structure. It has also shown better performance than some existing studies and other massive state-of-the-art models that diagnosed COVID-19 from CXRs.

Prediction of the early recurrence in spinal giant cell tumor of bone using radiomics of preoperative CT: Long-term outcome of 62 consecutive patients

•

Characteristics of 62 patients with spinal GCTB who underwent surgery.

•

A prognostic classification model was built based on features selected by SVM.

•

The combined histogram and texture features could predict recurrence of GCTB.

Objectives

To determine if radiomics analysis based on preoperative computed tomography (CT) can predict early postoperative recurrence of giant cell tumor of bone (GCTB) in the spine.

Methods

In a retrospective review, 62 patients with pathologically confirmed spinal GCTB from March 2008 to February 2018, with a minimum follow-up of 24 months, were identified. The mean follow-up was 73.7 months (range, 28.7–152.1 months). The clinical information including age, gender, lesion location, multi-vertebral involvement, and surgical methods, were obtained. CT images acquired before the operation were retrieved for radiomics analysis. For each case, the tumor regions of interest (ROI) was manually outlined, and a total of 107 radiomics features were extracted. The features were selected via the sequential selection process by using the support vector machine (SVM), then used to construct classification models with Gaussian kernels. The differentiation between recurrence and non-recurrence groups was evaluated by ROC analysis, using 10-fold cross-validation.

Results

Of the 62 patients, 17 had recurrence with a recurrence rate of 27.4%. None of the clinical information was significantly different between the two groups. Patients receiving curettage had a higher recurrence rate (6/16 = 37.5%) compared to patients receiving TES (6/26 = 23.1%) or intralesional spondylectomy (5/20 = 25%). The final radiomics model was built using 10 selected features, which achieved an accuracy of 89% with AUC of 0.78.

Conclusions

The radiomics model developed based on pre-operative CT can achieve a high accuracy to predict the recurrence of spinal GCTB. Patients who have a high risk of early recurrence should be treated more aggressively to minimize recurrence.

Dosimetric analysis of Deep Inspiratory Breath-hold technique (DIBH) in left-sided breast cancer radiotherapy and evaluation of pre-treatment predictors of cardiac doses for guiding patient selection for DIBH

Introduction

The risk of radiotherapy-associated cardiovascular disease has been a concern for decades in breast cancer survivors. The objective of our study is to evaluate the dosimetric benefit of Deep Inspiratory Breath-hold technique (DIBH) on organs-at-risk (OAR) sparing in left-sided breast cancer radiotherapy and to find out pre-treatment predictors of cardiac doses for guiding patient selection for DIBH.

Material and methods

Pre-radiotherapy planning CT scans were done in Free Breathing (FB) and in DIBH [using Active Breathing Coordinator system (ABC™)] in 31 left sided breast cancer patients. 3DCRT plans were generated for both scans. Comparison of anatomical and dosimetric variables were done using paired t test and correlation was evaluated using Pearson correlation. Linear regression was used to get independent predictors of cardiac sparing and Receiver Operating Characteristic (ROC) curve analysis was done to find out the specific threshold of the predictors.

Results

There was a 39.15% reduction in mean heart dose in DIBH compared to FB (2.4 Gy vs 4.01 Gy) (p < 0.001), 19% reduction in maximum Left Anterior Descending (LAD) dose and a 9.9% reduction in ipsilateral lung mean dose (p = 0.036) with DIBH. A significant correlation was observed between reduction in Heart Volume in Field (HVIF) and Maximum Heart Depth (MHD) with reduction in mean heart dose. Reduction in HVIF (ΔHVIF) independently predicted cardiac sparing.

Conclusion

DIBH leads to significant reduction in OAR doses and is suggested for all patients of left-sided breast cancer undergoing radiotherapy. However, HVIF and MHD predicted for cardiac sparing and threshold criteria of ΔHVIF and ΔMHD may be used by centres with high workload to select patients for DIBH.

Fast track triage for COVID-19 based on a population study: The soda score

BACKGROUND

Healthcare systems are under prominent stress due to the COVID-19 pandemic. A fast and simple triage is mandatory to screen patients who will benefit from early hospitalization, from those that can be managed as outpatients. There is a lack of all-comers scores, and no score has been proposed for western-world population.

AIMS

To develop a fast-track risk score valid for every COVID-19 patient at diagnosis.

METHODS

Single-center, retrospective study based on all the inhabitants of a healthcare area. Logistic regression was used to identify simple and wide-available risk factors for adverse events (death, intensive care admission, invasive mechanical ventilation, bleeding > BARC3, acute renal injury, respiratory insufficiency, myocardial infarction, acute heart failure, pulmonary emboli, or stroke).

RESULTS

Of the total healthcare area population, 447.979 inhabitants, 965 patients (0.22%), were diagnosed with COVID-19. A total of 124 patients (12.85%) experienced adverse events. The novel SODA score (based on sex, peripheral O2 saturation, presence of diabetes, and age) demonstrated good accuracy for adverse events prediction (area under ROC curve 0.858, CI: 0.82-0.98). A cut-off value of ≤2 points identifies patients with low risk (positive predictive value [PPV] for absence of events: 98.9%) and a cut-off of ≥5 points, high-risk patients (PPV 58.8% for adverse events).

CONCLUSIONS

This quick and easy score allows fast-track triage at the moment of diagnosis for COVID-19 using four simple variables: age, sex, SpO2, and diabetes. SODA score could improve preventive measures taken at diagnosis in high-risk patients and also relieve resources by identifying very low-risk patients.

Machine learning applications in microbial ecology, human microbiome studies, and environmental monitoring

Advances in nucleic acid sequencing technology have enabled expansion of our ability to profile microbial diversity. These large datasets of taxonomic and functional diversity are key to better understanding microbial ecology. Machine learning has proven to be a useful approach for analyzing microbial community data and making predictions about outcomes including human and environmental health. Machine learning applied to microbial community profiles has been used to predict disease states in human health, environmental quality and presence of contamination in the environment, and as trace evidence in forensics. Machine learning has appeal as a powerful tool that can provide deep insights into microbial communities and identify patterns in microbial community data. However, often machine learning models can be used as black boxes to predict a specific outcome, with little understanding of how the models arrived at predictions. Complex machine learning algorithms often may value higher accuracy and performance at the sacrifice of interpretability. In order to leverage machine learning into more translational research related to the microbiome and strengthen our ability to extract meaningful biological information, it is important for models to be interpretable. Here we review current trends in machine learning applications in microbial ecology as well as some of the important challenges and opportunities for more broad application of machine learning to understanding microbial communities.

Standardization of a colorimetric technique for determination of enzymatic activity of diamine oxidase (DAO) and its application in patients with clinical diagnosis of histamine intolerance

Background

Diamine Oxidase (DAO) has an essential role for degradation of exogenous histamine in the intestine; thus, histamine intolerance (HI) mainly has been correlated to a low concentration and/or activity of this enzyme. The objective of the study was to standardize a colorimetric technique to measure the enzymatic activity (function) of hDAO to then apply it to a series of 22 patients with a clinical diagnosis of HI.

Methods

For the standardization variables such as volume and type of sample, incubation time, wavelength of maximum absorption, types of substrates, and concentration of oxidized ascorbate were evaluated. Then the activity and concentration of DAO was determined in 22 patients diagnosed with HI and 22 healthy subjects.

Results

The mean of serum DAO concentration in the 22 patients was of 9.268 ± 1.124 U/mL. The mean of serum DAO concentration in the 22 controls was of 20.710 ± 2.509 U/mL, being significantly higher (P value 0.0002) the mean of the samples. The mean of serum DAO activity of the patients was of 1.143 ± 0.085 U/L and the controls was 1.533 ± 0.119 U/L, significantly greater than the patients (P value 0.011). In addition, the sensitivity of both techniques was 0.63. In the measuring of DAO concentration the specificity was 0.9, constituting a good diagnostic test, especially to rule out the true negatives. The determination of DAO activity had a specificity of 0.68.

Conclusions

Although we used a small number of patients and controls and the absorbance values were lower than expected, statistically significant differences were found in the levels of concentration and DAO activity between the patients with histamine intolerance and the controls. Therefore, the measuring of DAO concentration and DAO activity is a good diagnostic strategy for study suspect cases of HI. The simultaneous use of both assays allows to reduce positive and negative false results, for example, patients with normal DAO levels that could present a dysfunction in the activity of this enzyme.

Improving the serodiagnosis of canine Leishmania infantum infection in geographical areas of Brazil with different disease prevalence

Serodiagnosis of Leishmania infantum infection in dogs relies on the detection of antibodies against leishmanial crude extracts or parasitic defined antigens. The expansion of canine leishmaniasis from geographical areas of Brazil in which the infection is endemic to regions in which the disease is emerging is occurring. This fact makes necessary the analysis of the serodiagnostic capabilities of different leishmanial preparations in distinct geographical locations. In this article sera from dogs infected with Leishmania and showing the clinical form of the disease, were collected in three distinct Brazilian States and were tested against soluble leishmanial antigens or seven parasite individual antigens produced as recombinant proteins. We show that the recognition of soluble leishmanial antigens by sera from these animals was influenced by the geographical location of the infected dogs. Efficacy of the diagnosis based on this crude parasite preparation was higher in newly endemic regions when compared with areas of high disease endemicity. We also show that the use of three of the recombinant proteins, namely parasite surface kinetoplastid membrane protein of 11 kDa (KMP-11), and two members of the P protein family (P2a and P0), can improve the degree of sensitivity without adversely affecting the specificity of the diagnostic assays for canine leishmaniasis, independently of the geographical area of residence. In addition, sera from dogs clinically healthy but infected were also assayed with some of the antigen preparations. We demonstrate that the use of these proteins can help to the serodiagnosis of Leishmania infected animals with subclinical infections. Finally, we propose a diagnostic protocol using a combination of KMP-11, P2a y P0, together with total leishmanial extracts.

Prediction of human-virus protein-protein interactions through a sequence embedding-based machine learning method

The identification of human-virus protein-protein interactions (PPIs) is an essential and challenging research topic, potentially providing a mechanistic understanding of viral infection. Given that the experimental determination of human-virus PPIs is time-consuming and labor-intensive, computational methods are playing an important role in providing testable hypotheses, complementing the determination of large-scale interactome between species. In this work, we applied an unsupervised sequence embedding technique (doc2vec) to represent protein sequences as rich feature vectors of low dimensionality. Training a Random Forest (RF) classifier through a training dataset that covers known PPIs between human and all viruses, we obtained excellent predictive accuracy outperforming various combinations of machine learning algorithms and commonly-used sequence encoding schemes. Rigorous comparison with three existing human-virus PPI prediction methods, our proposed computational framework further provided very competitive and promising performance, suggesting that the doc2vec encoding scheme effectively captures context information of protein sequences, pertaining to corresponding protein-protein interactions. Our approach is freely accessible through our web server as part of our host-pathogen PPI prediction platform (http://zzdlab.com/InterSPPI/). Taken together, we hope the current work not only contributes a useful predictor to accelerate the exploration of human-virus PPIs, but also provides some meaningful insights into human-virus relationships.

RSAT variation-tools: An accessible and flexible framework to predict the impact of regulatory variants on transcription factor binding

Gene regulatory regions contain short and degenerated DNA binding sites recognized by transcription factors (TFBS). When TFBS harbor SNPs, the DNA binding site may be affected, thereby altering the transcriptional regulation of the target genes. Such regulatory SNPs have been implicated as causal variants in Genome-Wide Association Study (GWAS) studies. In this study, we describe improved versions of the programs Variation-tools designed to predict regulatory variants, and present four case studies to illustrate their usage and applications. In brief, Variation-tools facilitate i) obtaining variation information, ii) interconversion of variation file formats, iii) retrieval of sequences surrounding variants, and iv) calculating the change on predicted transcription factor affinity scores between alleles, using motif scanning approaches. Notably, the tools support the analysis of haplotypes. The tools are included within the well-maintained suite Regulatory Sequence Analysis Tools (RSAT, http://rsat.eu), and accessible through a web interface that currently enables analysis of five metazoa and ten plant genomes. Variation-tools can also be used in command-line with any locally-installed Ensembl genome. Users can input personal collections of variants and motifs, providing flexibility in the analysis.

Prediction of survival after surgical management of femoral metastatic bone disease - A comparison of prognostic models

Background

Operative fixation for femoral metastatic bone disease is based on the principles of reducing pain and restoring function. Recent literature has proposed a number of prognostic models for appendicular metastatic bone disease. The aim of this study was to compare the accuracy of proposed soring systems in the setting of femoral metastatic bone disease in order to provide surgeons with information to determine the most appropriate scoring system in this setting.

Methods

A retrospective cohort analysis of patients who underwent surgical management of femoral metastatic bone disease at a single institution were included. A pre-operative predicted survival for all 114 patients was retrospectively calculated utilising the revised Katagiri model, PathFx model, SSG score, Janssen nomogram, OPTModel and SPRING 13 nomogram. Univariate and multivariate Cox regression proportional hazard models were constructed to assess the role of prognostic variables in the patient group. Area under the receiver characteristics and Brier scores were calculated for each prognostic model from comparison of predicted survival and actual survival of patients to quantify the accuracy of each model.

Results

For the femoral metastatic bone disease patients treated with surgical fixation, multivariate analysis demonstrated a number of pre-operative factors associated with survival in femoral metastatic bone disease, consistent with established literature. The OPTIModel demonstrated the highest accuracy at predicting 12-month (Area Under the Curve [AUC] = 0.79) and 24-month (AUC = 0.77) survival after surgical management. PathFx model was the most accurate at predicting 3-month survival (AUC = 0.70) and 6-month (AUC = 0.70) survival. The PathFx model was successfully externally validated in the femoral patient dataset for all time periods.

Conclusions

Among six prognostic models assessed in the setting of femoral metastatic bone disease, the present study observed the most accurate model for 3-month, 6-month, 12-month and 24-month survival. The results of this study may be utilised by the treating surgical team to determine the most accurate model for the required time period and therefore improve decision-making in the care of patients with femoral metastatic bone disease.

A Gene Module-Based eQTL Analysis Prioritizing Disease Genes and Pathways in Kidney Cancer

Clear cell renal cell carcinoma (ccRCC) is the most common and most aggressive form of renal cell cancer (RCC). The incidence of RCC has increased steadily in recent years. The pathogenesis of renal cell cancer remains poorly understood. Many of the tumor suppressor genes, oncogenes, and dysregulated pathways in ccRCC need to be revealed for improvement of the overall clinical outlook of the disease. Here, we developed a systems biology approach to prioritize the somatic mutated genes that lead to dysregulation of pathways in ccRCC. The method integrated multi-layer information to infer causative mutations and disease genes. First, we identified differential gene modules in ccRCC by coupling transcriptome and protein-protein interactions. Each of these modules consisted of interacting genes that were involved in similar biological processes and their combined expression alterations were significantly associated with disease type. Then, subsequent gene module-based eQTL analysis revealed somatic mutated genes that had driven the expression alterations of differential gene modules. Our study yielded a list of candidate disease genes, including several known ccRCC causative genes such as BAP1 and PBRM1, as well as novel genes such as NOD2, RRM1, CSRNP1, SLC4A2, TTLL1 and CNTN1. The differential gene modules and their driver genes revealed by our study provided a new perspective for understanding the molecular mechanisms underlying the disease. Moreover, we validated the results in independent ccRCC patient datasets. Our study provided a new method for prioritizing disease genes and pathways.

Machine learning applications in cancer prognosis and prediction

Cancer has been characterized as a heterogeneous disease consisting of many different subtypes. The early diagnosis and prognosis of a cancer type have become a necessity in cancer research, as it can facilitate the subsequent clinical management of patients. The importance of classifying cancer patients into high or low risk groups has led many research teams, from the biomedical and the bioinformatics field, to study the application of machine learning (ML) methods. Therefore, these techniques have been utilized as an aim to model the progression and treatment of cancerous conditions. In addition, the ability of ML tools to detect key features from complex datasets reveals their importance. A variety of these techniques, including Artificial Neural Networks (ANNs), Bayesian Networks (BNs), Support Vector Machines (SVMs) and Decision Trees (DTs) have been widely applied in cancer research for the development of predictive models, resulting in effective and accurate decision making. Even though it is evident that the use of ML methods can improve our understanding of cancer progression, an appropriate level of validation is needed in order for these methods to be considered in the everyday clinical practice. In this work, we present a review of recent ML approaches employed in the modeling of cancer progression. The predictive models discussed here are based on various supervised ML techniques as well as on different input features and data samples. Given the growing trend on the application of ML methods in cancer research, we present here the most recent publications that employ these techniques as an aim to model cancer risk or patient outcomes.