Radiomics analysis using stability selection supervised component analysis for right-censored survival data.

Radiomics is a newly emerging field that involves the extraction of massive quantitative features from biomedical images by using data-characterization algorithms. Distinctive imaging features identified from biomedical images can be used for prognosis and therapeutic response prediction, and they can provide a noninvasive approach for personalized therapy. So far, many of the published radiomics studies utilize existing out of the box algorithms to identify the prognostic markers from biomedical images that are not specific to radiomics data. To better utilize biomedical images, we propose a novel machine learning approach, stability selection supervised principal component analysis (SSSuperPCA) that identifies stable features from radiomics big data coupled with dimension reduction for right-censored survival outcomes. The proposed approach allows us to identify a set of stable features that are highly associated with the survival outcomes in a simple yet meaningful manner, while controlling the per-family error rate. We evaluate the performance of SSSuperPCA using simulations and real data sets for non-small cell lung cancer and head and neck cancer, and compare it with other machine learning algorithms. The results demonstrate that our method has a competitive edge over other existing methods in identifying the prognostic markers from biomedical imaging data for the prediction of right-censored survival outcomes.

An enhanced machine learning tool for cis-eQTL mapping with regularization and confounder adjustments.

Many expression quantitative trait loci (eQTL) studies have been conducted to investigate the biological effects of variants in gene regulation. However, these eQTL studies may suffer from low or moderate statistical power and overly conservative false-discovery rate. In practice, most algorithms for eQTL identification do not model the joint effects of multiple genetic variants with weak or moderate influence. Here we present a novel machine-learning algorithm, lasso least-squares kernel machine (LSKM-LASSO) that model the association between multiple genetic variants and phenotypic traits simultaneously with the existence of nongenetic and genetic confounding. With a more general and flexible framework for the estimation of genetic confounding, LSKM-LASSO is able to provide a more accurate evaluation of the joint effects of multiple genetic variants. Our simulations demonstrate that our approach outperforms three state-of-the-art alternatives in terms of eQTL identification and phenotype prediction. We then apply our method to genotype and gene expression data of 11 tissues obtained from the Genotype-Tissue Expression project. Our algorithm was able to identify more genes with eQTL than other algorithms. By incorporating a regularization term and combining it with least-squares kernel machine, LSKM-LASSO provides a powerful tool for eQTL mapping and phenotype prediction.

Ten-year employment patterns of patients with first-episode schizophrenia-spectrum disorders: comparison of early intervention and standard care services.

BACKGROUND: Little is known about long-term employment outcomes for patients with first-episode schizophrenia-spectrum (FES) disorders who received early intervention services. AIMS: We compared the 10-year employment trajectory of patients with FES who received early intervention services with those who received standard care. Factors differentiating the employment trajectories were explored. METHOD: Patients with FES (N = 145) who received early intervention services in Hong Kong between 1 July 2001 and 30 June 2002 were matched with those who entered standard care 1 year previously. We used hierarchical clustering analysis to explore the 10-year employment clusters for both groups. We used the mixed model test to compare cluster memberships and piecewise regression analysis to compare the employment trajectories of the two groups. RESULTS: There were significantly more patients who received the early intervention service in the good employment cluster (early intervention: N = 98 [67.6%]; standard care: N = 76 [52.4%]; P = 0.009). In the poor employment cluster, there was a significant difference in the longitudinal pattern between early intervention and standard care for years 1-5 (P < 0.0001). The number of relapses during the first 3 years, months of full-time employment during the first year and years of education were significant in differentiating the clusters of the early intervention group. CONCLUSIONS: Results suggest there was an overall long-term benefit of early intervention services on employment. However, the benefit was not sustained for all patients. Personalisation of the duration of the early intervention service with a focus on relapse prevention and early vocational reintegration should be considered for service enhancement.

Association of an Early Intervention Service for Psychosis With Suicide Rate Among Patients With First-Episode Schizophrenia-Spectrum Disorders.

Importance: Patients with schizophrenia have a substantially higher suicide rate than the general public. Early intervention (EI) services improve short-term outcomes. However, little is known about the association of EI with suicide reduction in the long term. Objective: To examine the association of a 2-year EI service with suicide reduction in patients with first-episode schizophrenia-spectrum (FES) disorders during 12 years and the risk factors for early and late suicide. Design, Setting, and Participants: This historical control study compared 617 consecutive patients with FES who received the 2-year EI service in Hong Kong between July 1, 2001, and June 30, 2003, with 617 patients with FES who received standard care (SC) between July 1, 1998, and June 30, 2001, matched individually. Clinical information was systematically retrieved for the first 3 years of clinical care for both groups. The details of death were collected up to 12 years from presentation to the services. Data analysis was performed from October 30, 2016, to August 18, 2017. Main Outcomes and Measures: Suicide rate during 12 years was the primary measure. The association of the EI service with the suicide rates during years 1 through 3 and years 4 through 12 were explored separately. Results: The main analysis included 1234 patients, with 617 in each group (mean [SD] age at baseline, 21.2 [3.4] years in the EI group and 21.3 [3.4] years in the SC group; 318 male [51.5%] in the EI group and 322 [52.2%] in the SC group). The suicide rates were 7.5% in the SC group and 4.4% in the EI group (McNemar χ2 = 5.55, P = .02). Patients in the EI group had significantly better survival (propensity score-adjusted hazard ratio, 0.57; 95% CI, 0.36-0.91; P = .02), with the maximum association observed in the first 3 years. The number of suicide attempts was an indicator of early suicide (1-3 years). Premorbid occupational impairment, number of relapses, and poor adherence during the initial 3 years were indicators of late suicide (4-12 years). Conclusions and Relevance: This study suggests that the EI service may be associated with reductions in the long-term suicide rate. Suicide at different stages of schizophrenia was associated with unique risk factors, highlighting the importance of a phase-specific service.

A comparison of graph- and kernel-based -omics data integration algorithms for classifying complex traits.

BACKGROUND: High-throughput sequencing data are widely collected and analyzed in the study of complex diseases in quest of improving human health. Well-studied algorithms mostly deal with single data source, and cannot fully utilize the potential of these multi-omics data sources. In order to provide a holistic understanding of human health and diseases, it is necessary to integrate multiple data sources. Several algorithms have been proposed so far, however, a comprehensive comparison of data integration algorithms for classification of binary traits is currently lacking. RESULTS: In this paper, we focus on two common classes of integration algorithms, graph-based that depict relationships with subjects denoted by nodes and relationships denoted by edges, and kernel-based that can generate a classifier in feature space. Our paper provides a comprehensive comparison of their performance in terms of various measurements of classification accuracy and computation time. Seven different integration algorithms, including graph-based semi-supervised learning, graph sharpening integration, composite association network, Bayesian network, semi-definite programming-support vector machine (SDP-SVM), relevance vector machine (RVM) and Ada-boost relevance vector machine are compared and evaluated with hypertension and two cancer data sets in our study. In general, kernel-based algorithms create more complex models and require longer computation time, but they tend to perform better than graph-based algorithms. The performance of graph-based algorithms has the advantage of being faster computationally. CONCLUSIONS: The empirical results demonstrate that composite association network, relevance vector machine, and Ada-boost RVM are the better performers. We provide recommendations on how to choose an appropriate algorithm for integrating data from multiple sources.