Explainable Machine Learning for Early Assessment of COVID-19 Risk Prediction in Emergency Departments.

Between January and October of 2020, the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) virus has infected more than 34 million persons in a worldwide pandemic leading to over one million deaths worldwide (data from the Johns Hopkins University). Since the virus begun to spread, emergency departments were busy with COVID-19 patients for whom a quick decision regarding in- or outpatient care was required. The virus can cause characteristic abnormalities in chest radiographs (CXR), but, due to the low sensitivity of CXR, additional variables and criteria are needed to accurately predict risk. Here, we describe a computerized system primarily aimed at extracting the most relevant radiological, clinical, and laboratory variables for improving patient risk prediction, and secondarily at presenting an explainable machine learning system, which may provide simple decision criteria to be used by clinicians as a support for assessing patient risk. To achieve robust and reliable variable selection, Boruta and Random Forest (RF) are combined in a 10-fold cross-validation scheme to produce a variable importance estimate not biased by the presence of surrogates. The most important variables are then selected to train a RF classifier, whose rules may be extracted, simplified, and pruned to finally build an associative tree, particularly appealing for its simplicity. Results show that the radiological score automatically computed through a neural network is highly correlated with the score computed by radiologists, and that laboratory variables, together with the number of comorbidities, aid risk prediction. The prediction performance of our approach was compared to that that of generalized linear models and shown to be effective and robust. The proposed machine learning-based computational system can be easily deployed and used in emergency departments for rapid and accurate risk prediction in COVID-19 patients.

Assessment of Bones Deficient in Fibrillin-1 Microfibrils Reveals Pronounced Sex Differences.

Defects in the extracellular matrix protein fibrillin-1 that perturb transforming growth factor beta (TGFß) bioavailability lead to Marfan syndrome (MFS). MFS is an autosomal-dominant disorder, which is associated with connective tissue and skeletal defects, among others. To date, it is unclear how biological sex impacts the structural and functional properties of bone in MFS. The aim of this study was to investigate the effects of sex on bone microarchitecture and mechanical properties in mice with deficient fibrillin-1, a model of human MFS. Bones of 11-week-old male and female Fbn1mgR/mgR mice were investigated. Three-dimensional micro-computed tomography of femora and vertebrae revealed a lower ratio of trabecular bone volume to tissue volume, reduced trabecular number and thickness, and greater trabecular separation in females vs. males. Three-point bending of femora revealed significantly lower post-yield displacement and work-to-fracture in females vs. males. Mechanistically, we found higher Smad2 and ERK1/2 phosphorylation in females vs. males, demonstrating a greater activation of TGFß signaling in females. In summary, the present findings show pronounced sex differences in the matrix and function of bones deficient in fibrillin-1 microfibrils. Consequently, sex-specific analysis of bone characteristics in patients with MFS may prove useful in improving the clinical management and life quality of these patients, through the development of sex-specific therapeutic approaches.

Improved Diagnosis and Care for Rare Diseases through Implementation of Precision Public Health Framework.

Public health relies on technologies to produce and analyse data, as well as effectively develop and implement policies and practices. An example is the public health practice of epidemiology, which relies on computational technology to monitor the health status of populations, identify disadvantaged or at risk population groups and thereby inform health policy and priority setting. Critical to achieving health improvements for the underserved population of people living with rare diseases is early diagnosis and best care. In the rare diseases field, the vast majority of diseases are caused by destructive but previously difficult to identify protein-coding gene mutations. The reduction in cost of genetic testing and advances in the clinical use of genome sequencing, data science and imaging are converging to provide more precise understandings of the 'person-time-place' triad. That is: who is affected (people); when the disease is occurring (time); and where the disease is occurring (place). Consequently we are witnessing a paradigm shift in public health policy and practice towards 'precision public health'.Patient and stakeholder engagement has informed the need for a national public health policy framework for rare diseases. The engagement approach in different countries has produced highly comparable outcomes and objectives. Knowledge and experience sharing across the international rare diseases networks and partnerships has informed the development of the Western Australian Rare Diseases Strategic Framework 2015-2018 (RD Framework) and Australian government health briefings on the need for a National plan.The RD Framework is guiding the translation of genomic and other technologies into the Western Australian health system, leading to greater precision in diagnostic pathways and care, and is an example of how a precision public health framework can improve health outcomes for the rare diseases population.Five vignettes are used to illustrate how policy decisions provide the scaffolding for translation of new genomics knowledge, and catalyze transformative change in delivery of clinical services. The vignettes presented here are from an Australian perspective and are not intended to be comprehensive, but rather to provide insights into how a new and emerging 'precision public health' paradigm can improve the experiences of patients living with rare diseases, their caregivers and families.The conclusion is that genomic public health is informed by the individual and family needs, and the population health imperatives of an early and accurate diagnosis; which is the portal to best practice care. Knowledge sharing is critical for public health policy development and improving the lives of people living with rare diseases.

The main pulmonary artery in adults: a controlled multicenter study with assessment of echocardiographic reference values, and the frequency of dilatation and aneurysm in Marfan syndrome.

BACKGROUND: Echocardiographic upper normal limits of both main pulmonary artery (MPA) diameters (MPA-d) and ratio of MPA to aortic root diameter (MPA-r) are not defined in healthy adults. Accordingly, frequency of MPA dilatation based on echocardiography remains to be assessed in adults with Marfan syndrome (MFS). METHODS: We enrolled 123 normal adults (72 men, 52 women aged 42 ± 14 years) and 98 patients with MFS (42 men, 56 women aged 39 ± 14 years) in a retrospective cross-sectional observational controlled study in four tertiary care centers. We defined outcome measures including upper normal limits of MPA-d and MPA-r as 95 quantile of normal persons, MPA dilatation as diameters > upper normal limits, MPA aneurysm as diameters >4 cm, and indication for surgery as MPA diameters >6 cm. RESULTS: MPA diameters revealed normal distribution without correlation to age, sex, body weight, body height, body mass index and body surface area. The upper normal limit was 2.6 cm (95% confidence interval (CI) =2.44-2.76 cm) for MPA-d, and 1.05 (95% CI = .86-1.24) for MPA-r. MPA dilatation presented in 6 normal persons (4.9%) and in 68 MFS patients (69.4%; P < .001), MPA aneurysm presented only in MFS (15 patients; 15.3%; P < .001), and no patient required surgery. Mean MPA-r were increased in MFS (P < .001), but ratios >1.05 were equally frequent in 7 normal persons (5%) and in 8 MFS patients (10.5%; P = .161). MPA-r related to aortic root diameters (P = .042), reduced left ventricular ejection fraction (P = .006), and increased pulmonary artery systolic pressures (P = .040). No clinical manifestations of MFS and no FBN1 mutation characteristics related to MPA diameters. CONCLUSIONS: We established 2.6 cm for MPA-d and 1.05 for MPA-r as upper normal limits. MFS exhibits a high prevalence of MPA dilatation and aneurysm. However, patients may require MPA surgery only in scarce circumstances, most likely because formation of marked MPA aneurysm may require LV dysfunction and increased PASP.

Identity-by-descent filtering of exome sequence data for disease-gene identification in autosomal recessive disorders.

MOTIVATION: Next-generation sequencing and exome-capture technologies are currently revolutionizing the way geneticists screen for disease-causing mutations in rare Mendelian disorders. However, the identification of causal mutations is challenging due to the sheer number of variants that are identified in individual exomes. Although databases such as dbSNP or HapMap can be used to reduce the plethora of candidate genes by filtering out common variants, the remaining set of genes still remains on the order of dozens. RESULTS: Our algorithm uses a non-homogeneous hidden Markov model that employs local recombination rates to identify chromosomal regions that are identical by descent (IBD = 2) in children of consanguineous or non-consanguineous parents solely based on genotype data of siblings derived from high-throughput sequencing platforms. Using simulated and real exome sequence data, we show that our algorithm is able to reduce the search space for the causative disease gene to a fifth or a tenth of the entire exome. AVAILABILITY: An R script and an accompanying tutorial are available at http://compbio.charite.de/index.php/ibd2.html.

Clinical diagnostics in human genetics with semantic similarity searches in ontologies.

The differential diagnostic process attempts to identify candidate diseases that best explain a set of clinical features. This process can be complicated by the fact that the features can have varying degrees of specificity, as well as by the presence of features unrelated to the disease itself. Depending on the experience of the physician and the availability of laboratory tests, clinical abnormalities may be described in greater or lesser detail. We have adapted semantic similarity metrics to measure phenotypic similarity between queries and hereditary diseases annotated with the use of the Human Phenotype Ontology (HPO) and have developed a statistical model to assign p values to the resulting similarity scores, which can be used to rank the candidate diseases. We show that our approach outperforms simpler term-matching approaches that do not take the semantic interrelationships between terms into account. The advantage of our approach was greater for queries containing phenotypic noise or imprecise clinical descriptions. The semantic network defined by the HPO can be used to refine the differential diagnosis by suggesting clinical features that, if present, best differentiate among the candidate diagnoses. Thus, semantic similarity searches in ontologies represent a useful way of harnessing the semantic structure of human phenotypic abnormalities to help with the differential diagnosis. We have implemented our methods in a freely available web application for the field of human Mendelian disorders.