Interrupted time series analysis on first cardiovascular disease hospitalization for adherence to lipid-lowering therapy

Powerful and accurate detection of temporal gene expression patterns from multi-sample multi-stage single-cell transcriptomics data with TDEseq

We present a non-parametric statistical method called TDEseq that takes full advantage of smoothing splines basis functions to account for the dependence of multiple time points in scRNA-seq studies, and uses hierarchical structure linear additive mixed models to model the correlated cells within an individual. As a result, TDEseq demonstrates powerful performance in identifying four potential temporal expression patterns within a specific cell type. Extensive simulation studies and the analysis of four published scRNA-seq datasets show that TDEseq can produce well-calibrated p-values and up to 20% power gain over the existing methods for detecting temporal gene expression patterns.

Medication adherence prediction through temporal modelling in cardiovascular disease management

Background

Chronic conditions place a considerable burden on modern healthcare systems. Within New Zealand and worldwide cardiovascular disease (CVD) affects a significant proportion of the population and it is the leading cause of death. Like other chronic diseases, the course of cardiovascular disease is usually prolonged and its management necessarily long-term. Despite being highly effective in reducing CVD risk, non-adherence to long-term medication continues to be a longstanding challenge in healthcare delivery. The study investigates the benefits of integrating patient history and assesses the contribution of explicitly temporal models to medication adherence prediction in the context of lipid-lowering therapy.

Methods

Data from a CVD risk assessment tool is linked to routinely collected national and regional data sets including pharmaceutical dispensing, hospitalisation, lab test results and deaths. The study extracts a sub-cohort from 564,180 patients who had primary CVD risk assessment for analysis. Based on community pharmaceutical dispensing record, proportion of days covered (PDC) $$\ge$$ ≥  80 is used as the threshold for adherence. Two years (8 quarters) of patient history before their CVD risk assessment is used as the observation window to predict patient adherence in the subsequent 5 years (20 quarters). The predictive performance of temporal deep learning models long short-term memory (LSTM) and simple recurrent neural networks (Simple RNN) are compared against non-temporal models multilayer perceptron (MLP), ridge classifier (RC) and logistic regression (LR). Further, the study investigates the effect of lengthening the observation window on the task of adherence prediction.

Results

Temporal models that use sequential data outperform non-temporal models, with LSTM producing the best predictive performance achieving a ROC AUC of 0.805. A performance gap is observed between models that can discover non-linear interactions between predictor variables and their linear counter parts, with neural network (NN) based models significantly outperforming linear models. Additionally, the predictive advantage of temporal models become more pronounced when the length of the observation window is increased.

Conclusion

The findings of the study provide evidence that using deep temporal models to integrate patient history in adherence prediction is advantageous. In particular, the RNN architecture LSTM significantly outperforms all other model comparators.

A prediction model for nonpersistence or nonadherence to direct oral anticoagulants in hospitalized patients with atrial fibrillation

BACKGROUND: Medication adherence and persistence is fundamental for drug effectiveness, which is also true for the prevention of strokes in patients with atrial fibrillation (AF). Adherence to direct oral anticoagulants (DOACs) as first-line agents is often high in the early posthospital period. However, adherence often sharply declines (or eventually leads to nonpersistence) in the post-discharge ambulatory period, rendering stroke prevention ineffective. If patients at high risk of nonpersistence or nonadherence could be identified early, they could be offered early intervention measures to improve adherence and/or persistence. OBJECTIVE: To develop and internally validate a predictive model for medication nonadherence and nonpersistence to DOAC treatment in patients with AF after discharge using health insurance claims data. METHODS: We selected health insurance claims data between 2011 and 2016 from 8,055 patients with AF who were newly treated with rivaroxaban or apixaban after a hospital admission for stroke, transient ischemic attack, or AF. In the post-discharge ambulatory period, medication adherence was derived as the proportion of days covered, calculated from drug dispensation data. A maximum permissible 90-day gap between the end of a prescription and the next dispensation was used to estimate persistence. Candidate predictors were either derived from the index hospital admission or summarized from the previous year (eg, comorbidities or medication adherence to long-term treatments, such as ß-blockers, renin-angiotensin system inhibitors, statins, and thyroid hormones). A regularized logistic regression model was fitted using the least absolute shrinkage and selection operator in a split-sample approach (66.7% training data; 33.3% test data) to predict a composite of medication nonadherence/nonpersistence. Discrimination performance was assessed using the area under the receiver operating characteristic curve, the maximum sensitivity/specificity, and the scaled Brier score. A calibration curve fitted by linear regression was used to evaluate model calibration. RESULTS: The average age of the study participants was 79.7 years, 62% were female, and 3,515 patients (44%) were adherent and persistent (median follow-up of 185 days). Medication adherence to previous long-term treatments showed strong predictive properties. The developed model discriminated well (concordance statistic: 0.755), was well calibrated, and showed a scaled Brier score of 0.202 for identification of patients at risk. CONCLUSIONS: The model successfully predicted medication non-adherence/nonpersistence to DOAC treatment after discharge. Such a model could help ensure that targeted interventions are already in place at the time of hospital discharge, potentially preventing strokes and reducing costs. DISCLOSURES: Mr Wirbka is funded by the German Innovation Funds according to § 92a (2) Volume V of the Social Insurance Code (§ 92a Abs. 2, SGBV-Fünftes Buch Sozialgesetzbuch), grant number: 01VSF18019. Dr Haefeli received financial support from Daiichi-Sankyo, app development (https://www.easydoac.de/), and Bayer. He also received personal speaker fees from Bristol Myers-Squibb and Daiichi-Sankyo Online Seminar. Dr Meid is funded by the Physician-Scientist Programme of the Medical Faculty of Heidelberg University.

Identify differential genes and cell subclusters from time-series scRNA-seq data using scTITANS

Time-series single-cell RNA sequencing (scRNA-seq) provides a breakthrough in modern biology by enabling researchers to profile and study the dynamics of genes and cells based on samples obtained from multiple time points at an individual cell resolution. However, cell asynchrony and an additional dimension of multiple time points raises challenges in the effective use of time-series scRNA-seq data for identifying genes and cell subclusters that vary over time. However, no effective tools are available. Here, we propose scTITANS (https://github.com/ZJUFanLab/scTITANS), a method that takes full advantage of individual cells from all time points at the same time by correcting cell asynchrony using pseudotime from trajectory inference analysis. By introducing a time-dependent covariate based on time-series analysis method, scTITANS performed well in identifying differentially expressed genes and cell subclusters from time-series scRNA-seq data based on several example datasets. Compared to current attempts, scTITANS is more accurate, quantitative, and capable of dealing with heterogeneity among cells and making full use of the timing information hidden in biological processes. When extended to broader research areas, scTITANS will bring new breakthroughs in studies with time-series single cell RNA sequencing data.