Machine Learning for Postoperative Continuous Recovery Scores of Oncology Patients in Perioperative Care with Data from Wearables.

Assessing post-operative recovery is a significant component of perioperative care, since this assessment might facilitate detecting complications and determining an appropriate discharge date. However, recovery is difficult to assess and challenging to predict, as no universally accepted definition exists. Current solutions often contain a high level of subjectivity, measure recovery only at one moment in time, and only investigate recovery until the discharge moment. For these reasons, this research aims to create a model that predicts continuous recovery scores in perioperative care in the hospital and at home for objective decision making. This regression model utilized vital signs and activity metrics measured using wearable sensors and the XGBoost algorithm for training. The proposed model described continuous recovery profiles, obtained a high predictive performance, and provided outcomes that are interpretable due to the low number of features in the final model. Moreover, activity features, the circadian rhythm of the heart, and heart rate recovery showed the highest feature importance in the recovery model. Patients could be identified with fast and slow recovery trajectories by comparing patient-specific predicted profiles to the average fast- and slow-recovering populations. This identification may facilitate determining appropriate discharge dates, detecting complications, preventing readmission, and planning physical therapy. Hence, the model can provide an automatic and objective decision support tool.

Using a priori knowledge to align sequencing reads to their exact genomic position.

The use of a priori knowledge in the alignment of targeted sequencing data is investigated using computational experiments. Adapting a Needleman-Wunsch algorithm to incorporate the genomic position information from the targeted capture, we demonstrate that alignment can be done to just the target region of interest. When in addition use is made of direct string comparison, an improvement of up to a factor of 8 in alignment speed compared to the fastest conventional aligner (Bowtie) is obtained. This results in a total alignment time in targeted sequencing of around 7 min for aligning approximately 56 million captured reads. For conventional aligners such as Bowtie, BWA or MAQ, alignment to just the target region is not feasible as experiments show that this leads to an additional 88% SNP calls, the vast majority of which are false positives (≈ 92%).

Equivalence testing in microarray analysis: similarities in the transcriptome of human atherosclerotic and nonatherosclerotic macrophages.

We focus on similarities in the transcriptome of human Kupffer cells and alveolar, splenic, and atherosclerotic plaque-residing macrophages. We hypothesized that these macrophages share a common expression signature. We performed microarray analysis on mRNA from these subsets (4 patients) and developed a novel statistical method to identify genes with significantly similar expression levels. Phenotypic and functional diversity between macrophage subpopulations reflects their plasticity to respond to microenvironmental signals. Apart from detecting differences in expression profiles, the comparison of the transcriptomes of different macrophage populations may also allow the definition of molecular similarities between these subsets. This new method calculates the maximum difference in gene expression level, based on the estimated confidence interval on that gene's expression variance. We listed the genes by equivalence ranking relative to expression level. FDR estimation was used to determine significance. We identified 500 genes with significantly equivalent expression levels in the macrophage subsets at 5.5% FDR using a confidence level of α = 0.05 for equivalence. Among these are the established macrophage marker CD68, IL1 receptor antagonist, and MHC-related CD1C. These 500 genes were submitted to IPA and GO clustering using DAVID. Additionally, hierarchical clustering of these genes in the Novartis human gene expression atlas revealed a subset of 200 genes specifically expressed in macrophages. Equivalently expressed genes, identified by this new method, may not only help to dissect common molecular mechanisms, but also to identify cell- or condition-specific sets of marker genes that can be used for drug targeting and molecular imaging.

ER and PI3K Pathway Activity in Primary ER Positive Breast Cancer Is Associated with Progression-Free Survival of Metastatic Patients under First-Line Tamoxifen.

: Estrogen receptor positive (ER+) breast cancer patients are eligible for hormonal treatment, but only around half respond. A test with higher specificity for prediction of endocrine therapy response is needed to avoid hormonal overtreatment and to enable selection of alternative treatments. A novel testing method was reported before that enables measurement of functional signal transduction pathway activity in individual cancer tissue samples, using mRNA levels of target genes of the respective pathway-specific transcription factor. Using this method, 130 primary breast cancer samples were analyzed from non-metastatic ER+ patients, treated with surgery without adjuvant hormonal therapy, who subsequently developed metastatic disease that was treated with first-line tamoxifen. Quantitative activity levels were measured of androgen and estrogen receptor (AR and ER), PI3K-FOXO, Hedgehog (HH), NFκB, TGFß, and Wnt pathways. Based on samples with known pathway activity, thresholds were set to distinguish low from high activity. Subsequently, pathway activity levels were correlated with the tamoxifen treatment response and progression-free survival. High ER pathway activity was measured in 41% of the primary tumors and was associated with longer time to progression (PFS) of metastases during first-line tamoxifen treatment. In contrast, high PI3K, HH, and androgen receptor pathway activity was associated with shorter PFS, and high PI3K and TGFß pathway activity with worse treatment response. Potential clinical utility of assessment of ER pathway activity lies in predicting response to hormonal therapy, while activity of PI3K, HH, TGFß, and AR pathways may indicate failure to respond, but also opens new avenues for alternative or complementary targeted treatments.

A comprehensive sensitivity analysis of microarray breast cancer classification under feature variability.

BACKGROUND: Large discrepancies in signature composition and outcome concordance have been observed between different microarray breast cancer expression profiling studies. This is often ascribed to differences in array platform as well as biological variability. We conjecture that other reasons for the observed discrepancies are the measurement error associated with each feature and the choice of preprocessing method. Microarray data are known to be subject to technical variation and the confidence intervals around individual point estimates of expression levels can be wide. Furthermore, the estimated expression values also vary depending on the selected preprocessing scheme. In microarray breast cancer classification studies, however, these two forms of feature variability are almost always ignored and hence their exact role is unclear. RESULTS: We have performed a comprehensive sensitivity analysis of microarray breast cancer classification under the two types of feature variability mentioned above. We used data from six state of the art preprocessing methods, using a compendium consisting of eight different datasets, involving 1131 hybridizations, containing data from both one and two-color array technology. For a wide range of classifiers, we performed a joint study on performance, concordance and stability. In the stability analysis we explicitly tested classifiers for their noise tolerance by using perturbed expression profiles that are based on uncertainty information directly related to the preprocessing methods. Our results indicate that signature composition is strongly influenced by feature variability, even if the array platform and the stratification of patient samples are identical. In addition, we show that there is often a high level of discordance between individual class assignments for signatures constructed on data coming from different preprocessing schemes, even if the actual signature composition is identical. CONCLUSION: Feature variability can have a strong impact on breast cancer signature composition, as well as the classification of individual patient samples. We therefore strongly recommend that feature variability is considered in analyzing data from microarray breast cancer expression profiling experiments.

An evaluation protocol for subtype-specific breast cancer event prediction.

In recent years increasing evidence appeared that breast cancer may not constitute a single disease at the molecular level, but comprises a heterogeneous set of subtypes. This suggests that instead of building a single monolithic predictor, better predictors might be constructed that solely target samples of a designated subtype, which are believed to represent more homogeneous sets of samples. An unavoidable drawback of developing subtype-specific predictors, however, is that a stratification by subtype drastically reduces the number of samples available for their construction. As numerous studies have indicated sample size to be an important factor in predictor construction, it is therefore questionable whether the potential benefit of subtyping can outweigh the drawback of a severe loss in sample size. Factors like unequal class distributions and differences in the number of samples per subtype, further complicate comparisons. We present a novel experimental protocol that facilitates a comprehensive comparison between subtype-specific predictors and predictors that do not take subtype information into account. Emphasis lies on careful control of sample size as well as class and subtype distributions. The methodology is applied to a large breast cancer compendium involving over 1500 arrays, using a state-of-the-art subtyping scheme. We show that the resulting subtype-specific predictors outperform those that do not take subtype information into account, especially when taking sample size considerations into account.