Man vs. Machine: Comparing Physician vs. Electronic Health Record-Based Model Predictions for 30-Day Hospital Readmissions.

BACKGROUND: Electronic health record (EHR)-based readmission risk prediction models can be automated in real-time but have modest discrimination and may be missing important readmission risk factors. Clinician predictions of readmissions may incorporate information unavailable in the EHR, but the comparative usefulness is unknown. We sought to compare clinicians versus a validated EHR-based prediction model in predicting 30-day hospital readmissions. METHODS: We conducted a prospective survey of internal medicine clinicians in an urban safety-net hospital. Clinicians prospectively predicted patients' 30-day readmission risk on 5-point Likert scales, subsequently dichotomized into low- vs. high-risk. We compared human with machine predictions using discrimination, net reclassification, and diagnostic test characteristics. Observed readmissions were ascertained from a regional hospitalization database. We also developed and assessed a "human-plus-machine" logistic regression model incorporating both human and machine predictions. RESULTS: We included 1183 hospitalizations from 106 clinicians, with a readmission rate of 20.8%. Both clinicians and the EHR model had similar discrimination (C-statistic 0.66 vs. 0.66, p = 0.91). Clinicians had higher specificity (79.0% vs. 48.9%, p < 0.001) but lower sensitivity (43.9 vs. 75.2%, p < 0.001) than EHR model predictions. Compared with machine, human was better at reclassifying non-readmissions (non-event NRI + 30.1%) but worse at reclassifying readmissions (event NRI - 31.3%). A human-plus-machine approach best optimized discrimination (C-statistic 0.70, 95% CI 0.67-0.74), sensitivity (65.5%), and specificity (66.7%). CONCLUSION: Clinicians had similar discrimination but higher specificity and lower sensitivity than EHR model predictions. Human-plus-machine was better than either alone. Readmission risk prediction strategies should incorporate clinician assessments to optimize the accuracy of readmission predictions.

Cancer Gene Discovery: Past to Present.

Cancer is a complex disease that originates from genetic changes leading to multiple phenotypic manifestations that ultimately result in suffering and death from cancer. Attempts have been made to define the phenotypic and genetic "hallmarks" of cancer, but many of these "hallmarks" remain descriptive, while the underlying mechanisms responsible for these hallmarks remain elusive. For decades, cancer researchers have been methodically identifying the molecular mechanisms that result in tumor initiation, growth, metastases, and resistance to therapy. Great strides forward have been made and we are entering an era of "precision medicine" with the goal of treating each cancer based on its unique etiology. Increasingly, the decision to use targeted therapies and immunotherapies in the clinic is based on the genotype of the cancer being treated. For example, specific tyrosine kinase inhibitors are only prescribed to patients that express the tyrosine kinase protein on their cancer cells. Likewise, a genetically unstable cancer is predictive for successful immunotherapy. Knowledge of the specific genetic changes that result in overproduction of oncogenes and reduced production of tumor suppressors is crucial for advancing therapeutic options for cancer. The first chapter of this book presents a brief history of cancer gene discovery. In the remaining chapters of this book, we present protocols using in silico, in vitro, and in vivo techniques for identifying genetic drivers of cancer, in the hope that these protocols will be used to increase our knowledge of the molecular mechanisms driving cancer.

Transposon mutagenesis screen in mice identifies TM9SF2 as a novel colorectal cancer oncogene.

New therapeutic targets for advanced colorectal cancer (CRC) are critically needed. Our laboratory recently performed an insertional mutagenesis screen in mice to identify novel CRC driver genes and, thus, potential drug targets. Here, we define Transmembrane 9 Superfamily 2 (TM9SF2) as a novel CRC oncogene. TM9SF2 is an understudied protein, belonging to a well conserved protein family characterized by their nine putative transmembrane domains. Based on our transposon screen we found that TM9SF2 is a candidate progression driver in digestive tract tumors. Analysis of The Cancer Genome Atlas (TCGA) data revealed that approximately 35% of CRC patients have elevated levels of TM9SF2 mRNA, data we validated using an independent set of CRC samples. RNAi silencing of TM9SF2 reduced CRC cell growth in an anchorage-independent manner, a hallmark of cancer. Furthermore, CRISPR/Cas9 knockout of TM9SF2 substantially diminished CRC tumor fitness in vitro and in vivo. Transcriptome analysis of TM9SF2 knockout cells revealed a potential role for TM9SF2 in cell cycle progression, oxidative phosphorylation, and ceramide signaling. Lastly, we report that increased TM9SF2 expression correlates with disease stage and low TM9SF2 expression correlate with a more favorable relapse-free survival. Taken together, this study provides evidence that TM9SF2 is a novel CRC oncogene.

Mouse models for the discovery of colorectal cancer driver genes.

Colorectal cancer (CRC) constitutes a major public health problem as the third most commonly diagnosed and third most lethal malignancy worldwide. The prevalence and the physical accessibility to colorectal tumors have made CRC an ideal model for the study of tumor genetics. Early research efforts using patient derived CRC samples led to the discovery of several highly penetrant mutations (e.g., APC, KRAS, MMR genes) in both hereditary and sporadic CRC tumors. This knowledge has enabled researchers to develop genetically engineered and chemically induced tumor models of CRC, both of which have had a substantial impact on our understanding of the molecular basis of CRC. Despite these advances, the morbidity and mortality of CRC remains a cause for concern and highlight the need to uncover novel genetic drivers of CRC. This review focuses on mouse models of CRC with particular emphasis on a newly developed cancer gene discovery tool, the Sleeping Beauty transposon-based mutagenesis model of CRC.

Apoptosis-inducing factor (AIF) is targeted in IFN-α2a-induced Bid-mediated apoptosis through Bak activation in ovarian cancer cells.

Previously we have shown that interferon (IFN)-α induced apoptosis is predominantly mediated by the upregulation of tumor necrosis factor related apoptosis-inducing ligand (TRAIL) via the caspase-8 pathway. It was also shown that recruitment of mitochondria in IFN-α induced apoptosis involves the cleavage of BH3 interacting domain death agonist (Bid) to truncated Bid (tBid). In the present study, we demonstrate that tBid induced by IFN-α2a activates mitochondrial Bak to trigger the loss of mitochondrial membrane integrity, consequently causing release of apoptosis-inducing factor (AIF) in ovarian cancer cells, OVCAR3. AIF translocates from the mitochondria to the nucleus and induces nuclear fragmentation and cell death. Both a small molecule Bid inhibitor (BI-6C9) or Bid-RNA interference (RNAi) preserved mitochondrial membrane potential, prevented nuclear translocation of AIF, and abrogated IFN-α2a-induced cell death. Cell death induced by tBid was inhibited by AIF-RNAi, indicating that caspase-independent AIF signaling is the main pathway through which Bid mediates cell death. This was further supported by experiments showing that BI-6C9 did not prevent the release of cytochrome c from mitochondria to cytosol, while the release of AIF was prevented. In conclusion, IFN-α2a-induced apoptosis is mediated via the mitochondria-associated pathway involving the cleavage of Bid followed by AIF release that involves Bak activation and translocation of AIF from the mitochondria to the nucleus in OVCAR3 cells.

Potent antitumor effects of combination therapy with IFNs and monocytes in mouse models of established human ovarian and melanoma tumors.

Interferon-activated monocytes are known to exert cytocidal activity against tumor cells in vitro. Here, we have examined whether a combination of IFN-α2a and IFN-γ and human monocytes mediate significant antitumor effects against human ovarian and melanoma tumor xenografts in mouse models. OVCAR-3 tumors were treated i.t. with monocytes alone, IFN-α2a and IFN-γ alone or combination of all three on day 0, 15 or 30 post-tumor implantation. Mice receiving combination therapy beginning day 15 showed significantly reduced tumor growth and prolonged survival including complete regression in 40% mice. Tumor volumes measured on day 80 in mice receiving combination therapy (206 mm(3)) were significantly smaller than those of mice receiving the IFNs alone (1,041 mm(3)), monocytes alone (1,111 mm(3)) or untreated controls (1,728 mm(3)). Similarly, combination therapy with monocytes and IFNs of much larger tumor also inhibited OVCAR-3 tumor growth. Immunohistochemistry studies showed a large number of activated macrophages (CD31(+)/CD68(+)) infiltrating into OVCAR-3 tumors and higher densities of IL-12, IP10 and NOS2, markers of M1 (classical) macrophages in tumors treated with combination therapy compared to the controls. Interestingly, IFNs-activated macrophages induced apoptosis of OVCAR-3 tumor cells as monocytes alone or IFNs alone did not mediate significant apoptosis. Similar antitumor activity was observed in the LOX melanoma mouse model, but not as profound as seen with the OVCAR-3 tumors. Administration of either mixture of monocytes and IFN-α2a or monocytes and IFN-γ did not inhibit Lox melanoma growth; however, a significant inhibition was observed when tumors were treated with a mixture of monocytes, IFN-α2a and IFN-γ. These results indicate that monocytes and both IFN-α2a and IFN-γ may be required to mediate profound antitumor effect against human ovarian and melanoma tumors in mouse models.

Plasma concentrations of buprenorphine after epidural administration in conscious cats.

Buprenorphine plasma concentrations were measured after administering buprenorphine (20 µg/kg) into the lumbosacral epidural space of conscious cats chronically instrumented with an epidural catheter. Blood was collected from a jugular vein before injection and 15, 30, 45 and 60 min and 2, 3, 4, 5, 6, 8, 12 and 24 h after administration. Plasma buprenorphine concentrations were measured using ELISA. Background concentration (before injection) was 1.27 ± 0.27 ng/mL (mean ± SD). Including background concentration, the mean peak plasma concentration was obtained 15 min after injection (5.82 ± 3.75 ng/mL), and ranged from 3.79 to 2.20 ng/mL (30 min-3 h), remaining between 1.93 and 1.77 ng/mL (4-12 h), and declined to 1.40 ± 0.62 ng/mL at 24h. Elimination half-life was 58.8 ± 40.2 min and clearance 56.7 ± 21.5 mL/min. Results indicate early rapid systemic uptake of buprenorphine from epidural administration with plasma concentrations similar to using buccal or IM routes by 15 min postinjection.

Putative biomarker of working memory systems development during childhood and adolescence.

The study aimed to identify brain functional indicators of working memory systems development between 6 and 18 years. Event-related potentials (ERPs) were recorded from 251 normally developing children to stimuli requiring the updating of working memory. Cluster analysis of event-related potential componentry divided the sample into three clusters (mean ages 9, 12 and 16 years), with ascending cluster membership independently associated with improved task performance. The clusters correspond to periods of grey matter loss and white matter increase observed in developing children, supporting the view that the clusters delineate three key qualitative stages in advancing cognitive capability during the maturation of higher brain systems function. This outcome identifies a biomarker with the potential for assessing abnormalities in the rate of brain development.

The impact of early life stress on psychophysiological, personality and behavioral measures in 740 non-clinical subjects.

Early Life Stress (ELS) has been associated with a range of adverse outcomes in adults, including abnormalities in electrical brain activity [1], personality dimensions [40], increased vulnerability to substance abuse and depression [14]. The present study seeks to quantify these proposed effects in a large sample of non-clinical subjects. Data for the study was obtained from The Brain Resource International Database (six laboratories: two in USA, two in Europe, two in Australia). This study analyzed scalp electrophysiological data (EEG eyes open, closed and target auditory oddball data) and personality (NEO-FFI), history of addictive substance use and ELS) data that was acquired from 740 healthy volunteers. The ELS measures were collected via a self-report measure and covered a broad range of events from childhood sexual and physical abuse, to first-hand experience of traumatizing accidents and sustained domestic conflict [41]. Analysis of covariance, controlling for age and gender, compared EEG data from subjects exposed to ELS with those who were unexposed. ELS was associated with significantly decreased power across the EEG spectrum. The between group differences were strongest in the eyes closed paradigm, where subjects who experienced ELS showed significantly reduced beta (F1,405=12.37, p=.000), theta (F1,405=20.48, p=.000), alpha (F1,405=9.65, p=.002) and delta power (F1,450=36.22, p=.000). ELS exposed subjects also showed a significantly higher alpha peak frequency (F1,405=6.39, p=.012) in the eyes closed paradigm. Analysis of covariance on ERP components revealed that subjects who experienced ELS had significantly decreased N2 amplitude (F1,405=7.73, p=.006). Analyses of variance conducted on measures of personality revealed that subjects who experienced ELS had significantly higher levels of neuroticism (F1,264=13.39, p=.000) and openness (F1,264=17.11, p=.000), but lower levels of conscientiousness, than controls (F1,264=4.08, p=.044). The number of ELS events experienced was shown to be a significant predictor of scores on the DASS questionnaire [27], which rates subjects on symptoms of depression (F3,688=16.44, p=.000, R2=.07), anxiety (F3,688=14.32, p=.000, R2=.06) and stress (F3,688=20.02, p=.000, R2=.08). Each additional early life stressor was associated with an increase in these scores independent of age, gender and the type of stressor. Furthermore, the number of ELS experiences among smokers was also found to be a positive predictor of the nicotine dependency score (Faegstrom Test For Nicotine Dependence, [19]) (F3,104=10.99, p=.000, R2=.24), independent of age, gender and type of stressor. In conclusion, we highlight the impact of a history of ELS showed significant effects on brain function (EEG and ERP activity), personality dimensions and nicotine dependence.

Effect of hemi-circumferential periosteal transection and elevation in foals with experimentally induced angular limb deformities.

OBJECTIVE: To evaluate the effect of hemi-circumferential periosteal transection and elevation (HCPTE) in foals with, experimentally induced angular limb deformities. DESIGN: Prospective study. ANIMALS: 10 healthy foals. PROCEDURE: When foals were 30 days old, transphyseal bridge implants were placed on the lateral aspects of both distal radial physes. At 90 days of age (or when 15 degrees of angulation had developed), implants were removed, and HCPTE was performed on 1 limb. Foals were confined in small pens after surgery; the front feet of the foals were rasped weekly to maintain medial-to-lateral hoof wall balance. Dorsopalmar radiographic projections of the carpi were obtained before HCPTE and 2, 4, 6, 8, and 48 weeks later. RESULTS: At the time of transphyseal bridge removal and HCPTE, both treated and control limbs were observed to have a significantly greater carpal valgus, compared with the initial degree of angulation at 30 days of age. Following HCPTE or sham surgery, all limbs straightened over the subsequent 2 months of the study. Median angulation was not significantly different between treated and control limbs at any time during the study. CONCLUSIONS AND CLINICAL RELEVANCE: Results suggest that in foals with experimentally induced limb deformities, HCPTE was no more effective than stall confinement and hoof trimming alone for correction of the deformity.