Many mortality prediction models have been developed for patients in intensive care units (ICUs); most are based on data available at ICU admission. We investigated whether machine learning methods using analyses of time-series data improved mortality prognostication for patients in the ICU by providing real-time predictions of 90-day mortality. In addition, we examined to what extent such a dynamic model could be made interpretable by quantifying and visualising the features that drive the predictions at different timepoints. Based on the Simplified Acute Physiology Score (SAPS) III variables, we trained a machine learning model on longitudinal data from patients admitted to four ICUs in the Capital Region, Denmark, between 2011 and 2016. We included all patients older than 16 years of age, with an ICU stay lasting more than 1 h, and who had a Danish civil registration number to enable 90-day follow-up. We leveraged static data and physiological time-series data from electronic health records and the Danish National Patient Registry. A recurrent neural network was trained with a temporal resolution of 1 h. The model was internally validated using the holdout method with 20% of the training dataset and externally validated using previously unseen data from a fifth hospital in Denmark. Its performance was assessed with the Matthews correlation coefficient (MCC) and area under the receiver operating characteristic curve (AUROC) as metrics, using bootstrapping with 1000 samples with replacement to construct 95% CIs. A Shapley additive explanations algorithm was applied to the prediction model to obtain explanations of the features that drive patient-specific predictions, and the contributions of each of the 44 features in the model were analysed and compared with the variables in the original SAPS III model. From a dataset containing 15 615 ICU admissions of 12 616 patients, we included 14 190 admissions of 11 492 patients in our analysis. Overall, 90-day mortality was 33·1% (3802 patients). The deep learning model showed a predictive performance on the holdout testing dataset that improved over the timecourse of an ICU stay: MCC 0·29 (95% CI 0·25-0·33) and AUROC 0·73 (0·71-0·74) at admission, 0·43 (0·40-0·47) and 0·82 (0·80-0·84) after 24 h, 0·50 (0·46-0·53) and 0·85 (0·84-0·87) after 72 h, and 0·57 (0·54-0·60) and 0·88 (0·87-0·89) at the time of discharge. The model exhibited good calibration properties. These results were validated in an external validation cohort of 5827 patients with 6748 admissions: MCC 0·29 (95% CI 0·27-0·32) and AUROC 0·75 (0·73-0·76) at admission, 0·41 (0·39-0·44) and 0·80 (0·79-0·81) after 24 h, 0·46 (0·43-0·48) and 0·82 (0·81-0·83) after 72 h, and 0·47 (0·44-0·49) and 0·83 (0·82-0·84) at the time of discharge. The prediction of 90-day mortality improved with 1-h sampling intervals during the ICU stay. The dynamic risk prediction can also be explained for an individual patient, visualising the features contributing to the prediction at any point in time. This explanation allows the clinician to determine whether there are elements in the current patient state and care that are potentially actionable, thus making the model suitable for further validation as a clinical tool. Novo Nordisk Foundation and the Innovation Fund Denmark.

Building a population-specific catalogue of single nucleotide variants (SNVs), indels and structural variants (SVs) with frequencies, termed a national pan-genome, is critical for further advancing clinical and public health genetics in large cohorts. Here we report a Danish pan-genome obtained from sequencing 10 trios to high depth (50 × ). We report 536k novel SNVs and 283k novel short indels from mapping approaches and develop a population-wide de novo assembly approach to identify 132k novel indels larger than 10 nucleotides with low false discovery rates. We identify a higher proportion of indels and SVs than previous efforts showing the merits of high coverage and de novo assembly approaches. In addition, we use trio information to identify de novo mutations and use a probabilistic method to provide direct estimates of 1.27e−8 and 1.5e−9 per nucleotide per generation for SNVs and indels, respectively.

Hundreds of thousands of human genomes are now being sequenced to characterize genetic variation and use this information to augment association mapping studies of complex disorders and other phenotypic traits^1,2,3,4. Genetic variation is identified mainly by mapping short reads to the reference genome or by performing local assembly^2,5,6,7. However, these approaches are biased against discovery of structural variants and variation in the more complex parts of the genome. Hence, large-scale de novo assembly is needed. Here we show that it is possible to construct excellent de novo assemblies from high-coverage sequencing with mate-pair libraries extending up to 20 kilobases. We report de novo assemblies of 150 individuals (50 trios) from the GenomeDenmark project. The quality of these assemblies is similar to those obtained using the more expensive long-read technology^{4,8,9,10,11,12,13}. We use the assemblies to identify a rich set of structural variants including many novel insertions and demonstrate how this variant catalogue enables further deciphering of known association mapping signals. We leverage the assemblies to provide 100 completely resolved major histocompatibility complex haplotypes and to resolve major parts of the Y chromosome. Our study provides a regional reference genome that we expect will improve the power of future association mapping studies and hence pave the way for precision medicine initiatives, which now are being launched in many countries including Denmark.

Current standard treatments for metastatic colorectal cancer (CRC) are based on combination regimens with one of the two chemotherapeutic drugs, irinotecan or oxaliplatin. However, drug resistance frequently limits the clinical efficacy of these therapies. In order to gain new insights into mechanisms associated with chemoresistance, and departing from three distinct CRC cell models, we generated a panel of human colorectal cancer cell lines with acquired resistance to either oxaliplatin or irinotecan. We characterized the resistant cell line variants with regards to their drug resistance profile and transcriptome, and matched our results with datasets generated from relevant clinical material to derive putative resistance biomarkers. We found that the chemoresistant cell line variants had distinctive irinotecan‐ or oxaliplatin‐specific resistance profiles, with non‐reciprocal cross‐resistance. Furthermore, we could identify several new, as well as some previously described, drug resistance‐associated genes for each resistant cell line variant. Each chemoresistant cell line variant acquired a unique set of changes that may represent distinct functional subtypes of chemotherapy resistance. In addition, and given the potential implications for selection of subsequent treatment, we also performed an exploratory analysis, in relevant patient cohorts, of the predictive value of each of the specific genes identified in our cellular models.

Intensive-care units (ICUs) treat the most critically ill patients, which is complicated by the heterogeneity of the diseases that they encounter. Severity scores based mainly on acute physiology measures collected at ICU admission are used to predict mortality, but are non-specific, and predictions for individual patients can be inaccurate. We investigated whether inclusion of long-term disease history before ICU admission improves mortality predictions. Registry data for long-term disease histories for more than 230 000 Danish ICU patients were used in a neural network to develop an ICU mortality prediction model. Long-term disease histories and acute physiology measures were aggregated to predict mortality risk for patients for whom both registry and ICU electronic patient record data were available. We compared mortality predictions with admission scores on the Simplified Acute Physiology Score (SAPS) II, the Acute Physiologic Assessment and Chronic Health Evaluation (APACHE) II, and the best available multimorbidity score, the Multimorbidity Index. An external validation set from an additional hospital was acquired after model construction to confirm the validity of our model. During initial model development data were split into a training set (85%) and an independent test set (15%), and a five-fold cross-validation was done during training to avoid overfitting. Neural networks were trained for datasets with disease history of 1 month, 3 months, 6 months, 1 year, 2·5 years, 5 years, 7·5 years, 10 years, and 23 years before ICU admission. Mortality predictions with a model based solely on disease history outperformed the Multimorbidity Index (Matthews correlation coefficient 0·265 vs 0·065), and performed similarly to SAPS II and APACHE II (Matthews correlation coefficient with disease history, age, and sex 0·326 vs 0·347 and 0·300 for SAPS II and APACHE II, respectively). Diagnoses up to 10 years before ICU admission affected current mortality prediction. Aggregation of previous disease history and acute physiology measures in a neural network yielded the most precise predictions of in-hospital mortality (Matthews correlation coefficient 0·391 for in-hospital mortality compared with 0·347 with SAPS II and 0·300 with APACHE II). These results for the aggregated model were validated in an external independent dataset of 1528 patients (Matthews correlation coefficient for prediction of in-hospital mortality 0·341). Longitudinal disease-spectrum-wide data available before ICU admission are useful for mortality prediction. Disease history can be used to differentiate mortality risk between patients with similar vital signs with more precision than SAPS II and APACHE II scores. Machine learning models can be deconvoluted to generate novel understandings of how ICU patient features from long-term and short-term events interact with each other. Explainable machine learning models are key in clinical settings, and our results emphasise how to progress towards the transformation of advanced models into actionable, transparent, and trustworthy clinical tools. Novo Nordisk Foundation and Innovation Fund Denmark.

Klinefelter syndrome (KS) (47,XXY) is the most common male sex chromosome aneuploidy. Diagnosis and clinical supervision remain a challenge due to varying phenotypic presentation and insufficient characterization of the syndrome. Here we combine health data-driven epidemiology and molecular level systems biology to improve the understanding of KS and the molecular interplay influencing its comorbidities. In total, 78 overrepresented KS comorbidities were identified using in- and out-patient registry data from the entire Danish population covering 6.8 million individuals. The comorbidities extracted included both clinically well-known (e.g. infertility and osteoporosis) and still less established KS comorbidities (e.g. pituitary gland hypofunction and dental caries). Several systems biology approaches were applied to identify key molecular players underlying KS comorbidities: Identification of co-expressed modules as well as central hubs and gene dosage perturbed protein complexes in a KS comorbidity network build from known disease proteins and their protein–protein interactions. The systems biology approaches together pointed to novel aspects of KS disease phenotypes including perturbed Jak-STAT pathway, dysregulated genes important for disturbed immune system (IL4), energy balance (POMC and LEP) and erythropoietin signalling in KS. We present an extended epidemiological study that links KS comorbidities to the molecular level and identify potential causal players in the disease biology underlying the identified comorbidities.