Optimized approach to decision fusion of heterogeneous data for breast cancer diagnosis.

As more diagnostic testing options become available to physicians, it becomes more difficult to combine various types of medical information together in order to optimize the overall diagnosis. To improve diagnostic performance, here we introduce an approach to optimize a decision-fusion technique to combine heterogeneous information, such as from different modalities, feature categories, or institutions. For classifier comparison we used two performance metrics: The receiving operator characteristic (ROC) area under the curve [area under the ROC curve (AUC)] and the normalized partial area under the curve (pAUC). This study used four classifiers: Linear discriminant analysis (LDA), artificial neural network (ANN), and two variants of our decision-fusion technique, AUC-optimized (DF-A) and pAUC-optimized (DF-P) decision fusion. We applied each of these classifiers with 100-fold cross-validation to two heterogeneous breast cancer data sets: One of mass lesion features and a much more challenging one of microcalcification lesion features. For the calcification data set, DF-A outperformed the other classifiers in terms of AUC (p < 0.02) and achieved AUC=0.85 +/- 0.01. The DF-P surpassed the other classifiers in terms of pAUC (p < 0.01) and reached pAUC=0.38 +/- 0.02. For the mass data set, DF-A outperformed both the ANN and the LDA (p < 0.04) and achieved AUC=0.94 +/- 0.01. Although for this data set there were no statistically significant differences among the classifiers' pAUC values (pAUC=0.57 +/- 0.07 to 0.67 +/- 0.05, p > 0.10), the DF-P did significantly improve specificity versus the LDA at both 98% and 100% sensitivity (p < 0.04). In conclusion, decision fusion directly optimized clinically significant performance measures, such as AUC and pAUC, and sometimes outperformed two well-known machine-learning techniques when applied to two different breast cancer data sets.

Comparative performance of multiview stereoscopic and mammographic display modalities for breast lesion detection.

PURPOSE: Mammography is known to be one of the most difficult radiographic exams to interpret. Mammography has important limitations, including the superposition of normal tissue that can obscure a mass, chance alignment of normal tissue to mimic a true lesion and the inability to derive volumetric information. It has been shown that stereomammography can overcome these deficiencies by showing that layers of normal tissue lay at different depths. If standard stereomammography (i.e., a single stereoscopic pair consisting of two projection images) can significantly improve lesion detection, how will multiview stereoscopy (MVS), where many projection images are used, compare to mammography? The aim of this study was to assess the relative performance of MVS compared to mammography for breast mass detection. METHODS: The MVS image sets consisted of the 25 raw projection images acquired over an arc of approximately 45 degrees using a Siemens prototype breast tomosynthesis system. The mammograms were acquired using a commercial Siemens FFDM system. The raw data were taken from both of these systems for 27 cases and realistic simulated mass lesions were added to duplicates of the 27 images at the same local contrast. The images with lesions (27 mammography and 27 MVS) and the images without lesions (27 mammography and 27 MVS) were then postprocessed to provide comparable and representative image appearance across the two modalities. All 108 image sets were shown to five full-time breast imaging radiologists in random order on a state-of-the-art stereoscopic display. The observers were asked to give a confidence rating for each image (0 for lesion definitely not present, 100 for lesion definitely present). The ratings were then compiled and processed using ROC and variance analysis. RESULTS: The mean AUC for the five observers was 0.614 +/- 0.055 for mammography and 0.778 +/- 0.052 for multiview stereoscopy. The difference of 0.164 +/- 0.065 was statistically significant with a p-value of 0.0148. CONCLUSIONS: The differences in the AUCs and the p-value suggest that multiview stereoscopy has a statistically significant advantage over mammography in the detection of simulated breast masses. This highlights the dominance of anatomical noise compared to quantum noise for breast mass detection. It also shows that significant lesion detection can be achieved with MVS without any of the artifacts associated with tomosynthesis.

The cost of biopharmaceutical R&D: Is biotech different?

The costs of developing the types of new drugs that have been pursued by traditional pharmaceutical firms have been estimated in a number of studies. However, similar analyses have not been published on the costs of developing the types of molecules on which biotech firms have focused. This study represents a first attempt to get a sense for the magnitude of the R&D costs associated with the discovery and development of new therapeutic biopharmaceuticals (specifically, recombinant proteins and monoclonal antibodies [mAbs]). We utilize drug-specific data on cash outlays, development times, and success in obtaining regulatory marketing approval to estimate the average pre-tax R&D resource cost for biopharmaceuticals up to the point of initial US marketing approval (in year 2005 dollars). We found average out-of-pocket (cash outlay) cost estimates per approved biopharmaceutical of $198 million, $361 million, and $559 million for the preclinical period, the clinical period, and in total, respectively. Including the time costs associated with biopharmaceutical R&D, we found average capitalized cost estimates per approved biopharmaceutical of $615 million, $626 million, and $1241 million for the preclinical period, the clinical period, and in total, respectively. Adjusting previously published estimates of R&D costs for traditional pharmaceutical firms by using past growth rates for pharmaceutical company costs to correspond to the more recent period to which our biopharmaceutical data apply, we found that total out-of-pocket cost per approved biopharmaceutical was somewhat lower than for the pharmaceutical company data ($559 million vs $672 million). However, estimated total capitalized cost per approved new molecule was nearly the same for biopharmaceuticals as for the adjusted pharmaceutical company data ($1241 million versus $1318 million). The results should be viewed with some caution for now given a limited number of biopharmaceutical molecules with data on cash outlays, different therapeutic class distributions for biopharmaceuticals and for pharmaceutical company drugs, and uncertainty about whether recent growth rates in pharmaceutical company costs are different from immediate past growth rates. Copyright © 2007 John Wiley & Sons, Ltd.

Returns to R&D on new drug introductions in the 1980s.

This study finds that the mean IRR for 1980-84 U.S. new drug introductions is 11.1%, and the mean NPV is 22 million (1990 dollars). The distribution of returns is highly skewed. The results are robust to plausible changes in the baseline assumptions. Our work is also compared with a 1993 study by the OTA. Despite some important differences in assumptions, both studies imply that returns for the average NCE are within one percentage point of the industry's cost of capital. This is much less than what is typically observed in analyses based on accounting data.

Where does energy R&D come from? Examining crowding out from energy R&D

Recent efforts to endogenize technological change in climate policy models demonstrate the importance of accounting for the opportunity cost of climate R&D investments. Because the social returns to R&D investments are typically higher than the social returns to other types of investment, any new climate mitigation R&D that comes at the expense of other R&D investment may dampen the overall gains from induced technological change. Unfortunately, there has been little empirical work to guide modelers as to the potential magnitude of such crowding out effects. This paper considers both the private and social opportunity costs of climate R&D. Addressing private costs, we ask whether an increase in climate R&D represents new R&D spending, or whether some (or all) of the additional climate R&D comes at the expense of other R&D. Addressing social costs, we use patent citations to compare the social value of alternative energy research to other types of R&D that may be crowded out. Beginning at the industry level, we find no evidence of crowding out across sectors-that is, increases in energy R&D do not draw R&D resources away from sectors that do not perform R&D. Given this, we proceed with a detailed look at alternative energy R&D. Linking patent data and financial data by firm, we ask whether an increase in alternative energy patents leads to a decrease in other types of patenting activity. While we find that increases in alternative energy patents do result in fewer patents of other types, the evidence suggests that this is due to profit-maximizing changes in research effort, rather than financial constraints that limit the total amount of R&D possible. Finally, we use patent citation data to compare the social value of alternative energy patents to other patents by these firms. Alternative energy patents are cited more frequently, and by a wider range of other technologies, than other patents by these firms, suggesting that their social value is higher. © 2011 Elsevier B.V.

R&D Costs and Returns to New Drug Development: A Review of the Evidence

© 2012 by Oxford University Press. All rights reserved.This article reviews the extensive literature on R&D costs and returns. The first section focuses on R&D costs and the various factors that have affected the trends in real R&D costs over time. The second section considers economic studies on the distribution of returns in pharmaceuticals for different cohorts of new drug introductions. It also reviews the use of these studies to analyze the impact of policy actions on R&D costs and returns. The final section concludes and discusses open questions for further research.

R Markdown: Integrating A Reproducible Analysis Tool into Introductory Statistics

Nolan and Temple Lang argue that “the ability to express statistical computations is an es- sential skill.” A key related capacity is the ability to conduct and present data analysis in a way that another person can understand and replicate. The copy-and-paste workflow that is an artifact of antiquated user-interface design makes reproducibility of statistical analysis more difficult, especially as data become increasingly complex and statistical methods become increasingly sophisticated. R Markdown is a new technology that makes creating fully-reproducible statistical analysis simple and painless. It provides a solution suitable not only for cutting edge research, but also for use in an introductory statistics course. We present experiential and statistical evidence that R Markdown can be used effectively in introductory statistics courses, and discuss its role in the rapidly-changing world of statistical computation.

Drug-drug interaction studies of cardiovascular drugs involving P-glycoprotein, an efflux transporter, on the pharmacokinetics of edoxaban, an oral factor Xa inhibitor.

BACKGROUND: Edoxaban, an oral direct factor Xa inhibitor, is in development for thromboprophylaxis, including prevention of stroke and systemic embolism in patients with atrial fibrillation (AF). P-glycoprotein (P-gp), an efflux transporter, modulates absorption and excretion of xenobiotics. Edoxaban is a P-gp substrate, and several cardiovascular (CV) drugs have the potential to inhibit P-gp and increase drug exposure. OBJECTIVE: To assess the potential pharmacokinetic interactions of edoxaban and 6 cardiovascular drugs used in the management of AF and known P-gp substrates/inhibitors. METHODS: Drug-drug interaction studies with edoxaban and CV drugs with known P-gp substrate/inhibitor potential were conducted in healthy subjects. In 4 crossover, 2-period, 2-treatment studies, subjects received edoxaban 60 mg alone and coadministered with quinidine 300 mg (n = 42), verapamil 240 mg (n = 34), atorvastatin 80 mg (n = 32), or dronedarone 400 mg (n = 34). Additionally, edoxaban 60 mg alone and coadministered with amiodarone 400 mg (n = 30) or digoxin 0.25 mg (n = 48) was evaluated in a single-sequence study and 2-cohort study, respectively. RESULTS: Edoxaban exposure measured as area under the curve increased for concomitant administration of edoxaban with quinidine (76.7 %), verapamil (52.7 %), amiodarone (39.8 %), and dronedarone (84.5 %), and exposure measured as 24-h concentrations for quinidine (11.8 %), verapamil (29.1 %), and dronedarone (157.6 %) also increased. Administration of edoxaban with amiodarone decreased the 24-h concentration for edoxaban by 25.7 %. Concomitant administration with digoxin or atorvastatin had minimal effects on edoxaban exposure. CONCLUSION: Coadministration of the P-gp inhibitors quinidine, verapamil, and dronedarone increased edoxaban exposure. Modest/minimal effects were observed for amiodarone, atorvastatin, and digoxin.