Decision-theoretic designs for dose-finding clinical trials with multiple outcomes

A decision-theoretic framework is proposed for designing sequential dose-finding trials with multiple outcomes. The optimal strategy is solvable theoretically via backward induction. However, for dose-finding studies involving k doses, the computational complexity is the same as the bandit problem with k-dependent arms, which is computationally prohibitive. We therefore provide two computationally compromised strategies, which is of practical interest as the computational complexity is greatly reduced: one is closely related to the continual reassessment method (CRM), and the other improves CRM and approximates to the optimal strategy better. In particular, we present the framework for phase I/II trials with multiple outcomes. Applications to a pediatric HIV trial and a cancer chemotherapy trial are given to illustrate the proposed approach. Simulation results for the two trials show that the computationally compromised strategy can perform well and appear to be ethical for allocating patients. The proposed framework can provide better approximation to the optimal strategy if more extensive computing is available.

An extension of the continual reassessment method using decision theory

The primary goal of a phase I trial is to find the maximally tolerated dose (MTD) of a treatment. The MTD is usually defined in terms of a tolerable probability, q*, of toxicity. Our objective is to find the highest dose with toxicity risk that does not exceed q*, a criterion that is often desired in designing phase I trials. This criterion differs from that of finding the dose with toxicity risk closest to q*, that is used in methods such as the continual reassessment method. We use the theory of decision processes to find optimal sequential designs that maximize the expected number of patients within the trial allocated to the highest dose with toxicity not exceeding q*, among the doses under consideration. The proposed method is very general in the sense that criteria other than the one considered here can be optimized and that optimal dose assignment can be defined in terms of patients within or outside the trial. It includes as an important special case the continual reassessment method. Numerical study indicates the strategy compares favourably with other phase I designs.

Recipient v russkom razgovore : o raspredelenii funkcij mezhdu otvetami da, nu i tak

The study investigates actions by recipients in spontaneous Russian conversations by focusing on DA, NU and TAK, when they are used as responses to the main speaker's larger on-going turn. The database for the study consists of some 7 hours of spontaneous conversations. The participants of the conversations come from different parts of Russia. The use of DA, NU and TAK was analyzed by applying the method of ethnomethodological conversation analysis from the point of view of the type of the context, the sequential placement of the response and its manner of production. The particles were analyzed both in contexts in which they responded to an informing and in affective contexts. The particles NU and TAK were used by the speakers almost exclusively in informing contexts, whereas DA was the central response type in affective contexts. DA was also the most common response to information with affective implications. The information, to which the particle NU provided as response, was often unspesific and projected a spesification or explanation by its speaker as the next action. DA and TAK, by contrast, treated the information as one that could be followed and was sufficient in its local context. As a response to parenthetical information NU responded to information that was only loosely connected with the mainline of talk. The particle DA, by contrast, was used as a response to such parenthetical information, which was more crucial for the larger on-going activity. Only NU was used as a response that invited the main speaker to continue a turn that she or he had offered as possibly complete. NU was also used by the recipient after her or his own contribution as a continuer. In affective contexts, DA expressed, depending on its more spesific context, not only agreement but also other functions, such as giving up arguing or prior knowledge on the topic being discussed. In addition DA responses were used to display empathy and identification with the state of affairs expressed by the co-participant. NU, by contrast, was seldom used as a response to a turn that expressed affect. When it was used in affective contexts, it displayed agreement with the co-participant or just registered an assessment by her or him.

Mauwake reference grammar

This dissertation provides a synchronic grammatical description of Mauwake, a Papuan (Trans-New Guinea) language of about 2000 speakers on the North Coast of the Madang Province in Papua New Guinea. The theoretical background is that of Basic Linguistic Theory (BLT), used extensively in analysing and writing descriptive grammars. The chapters from morphology to clause level are described from form to function; in the later chapters the function is taken more often as the starting point. Any theory-specific terminology is kept to the minimum and formalisms have been avoided in accordance with BLT principles. Mauwake has a classic 5-vowel system and 14 consonant phonemes. With its simple phonology it is a typical representative of the Madang North Coast languages. For a Papuan language there are relatively few morphophonological alternations. Nouns are either alienably or inalienably possessed. There is no obligatory number marking in nouns or noun phrases. Pronouns have several different forms: five for case and three for other functions. The dative pronouns are treated as [+human] locatives, and they have also grammaticalised as possessives. The verbal morphology is agglutinative and mainly suffixal. Unusual features include two distributive suffixes, and the interaction of the derivational benefactive and the inflectional beneficiary suffixes. The applicative suffix has either transitivising or causative but not benefactive function. The switch-reference system distinguishes between simultaneous and sequential action, as well as same or different subject in relation to the following clause. There are several verbs denoting coming and going, and they may combine with one of three prefixes to indicate bringing and taking. Mauwake is a nominative-accusative type language, and the basic constituent order in a clause is SOV. Subject and object are the only syntactic arguments. There is no indirect object, but a clause can have two or even three objects. A nominalised clause with a finite verb functions as a relative clause or a complement clause; one with a nominalised verb has several different functions. Functional domains described include modality, negation, deixis, quantification, possession and comparison. As there are four negators, Mauwake has more variation in negative expressions than is usual in Papuan languages. Clause chaining is the preferred strategy for joining clauses into sentences, but coordination and subordination of finite clauses are also common. The form of a complement clause depends on whether it is of the fact, action or potential type. Tail-head linkage is used as a cohesive device between sentences. The discourse-level features described are topic and focus.

A Bayesian decision approach for sample size determination in phase II trials

Stallard (1998, Biometrics 54, 279-294) recently used Bayesian decision theory for sample-size determination in phase II trials. His design maximizes the expected financial gains in the development of a new treatment. However, it results in a very high probability (0.65) of recommending an ineffective treatment for phase III testing. On the other hand, the expected gain using his design is more than 10 times that of a design that tightly controls the false positive error (Thall and Simon, 1994, Biometrics 50, 337-349). Stallard's design maximizes the expected gain per phase II trial, but it does not maximize the rate of gain or total gain for a fixed length of time because the rate of gain depends on the proportion: of treatments forwarding to the phase III study. We suggest maximizing the rate of gain, and the resulting optimal one-stage design becomes twice as efficient as Stallard's one-stage design. Furthermore, the new design has a probability of only 0.12 of passing an ineffective treatment to phase III study.

Optimal designs for evaluating a series of treatments

Several articles in this journal have studied optimal designs for testing a series of treatments to identify promising ones for further study. These designs formulate testing as an ongoing process until a promising treatment is identified. This formulation is considered to be more realistic but substantially increases the computational complexity. In this article, we show that these new designs, which control the error rates for a series of treatments, can be reformulated as conventional designs that control the error rates for each individual treatment. This reformulation leads to a more meaningful interpretation of the error rates and hence easier specification of the error rates in practice. The reformulation also allows us to use conventional designs from published tables or standard computer programs to design trials for a series of treatments. We illustrate these using a study in soft tissue sarcoma.

Bias reduction using stochastic approximation

The paper studies stochastic approximation as a technique for bias reduction. The proposed method does not require approximating the bias explicitly, nor does it rely on having independent identically distributed (i.i.d.) data. The method always removes the leading bias term, under very mild conditions, as long as auxiliary samples from distributions with given parameters are available. Expectation and variance of the bias-corrected estimate are given. Examples in sequential clinical trials (non-i.i.d. case), curved exponential models (i.i.d. case) and length-biased sampling (where the estimates are inconsistent) are used to illustrate the applications of the proposed method and its small sample properties.

Error bounds for calculation of the Gittins indices

For a wide class of semi-Markov decision processes the optimal policies are expressible in terms of the Gittins indices, which have been found useful in sequential clinical trials and pharmaceutical research planning. In general, the indices can be approximated via calibration based on dynamic programming of finite horizon. This paper provides some results on the accuracy of such approximations, and, in particular, gives the error bounds for some well known processes (Bernoulli reward processes, normal reward processes and exponential target processes).

Gittins indices and constrained allocation in clinical trials

We explore the use of Gittins indices to search for near optimality in sequential clinical trials. Some adaptive allocation rules are proposed to achieve the following two objectives as far as possible: (i) to reduce the expected successes lost, (ii) to minimize the error probability at the end. Simulation results indicate the merits of the rules based on Gittins indices for small trial sizes. The rules are generalized to the case when neither of the response densities is known. Asymptotic optimality is derived for the constrained rules. A simple allocation rule is recommended for one-stage models. The simulation results indicate that it works better than both equal allocation and Bather's randomized allocation. We conclude with a discussion of possible further developments.

Quantifying progressive anterior overgrowth in the thoracic vertebrae of adolescent idiopathic scoliosis patients

Study design Anterior and posterior vertebral body heights were measured from sequential MRI scans of adolescent idiopathic scoliosis (AIS) patients and healthy controls. Objective To measure changes in vertebral body height over time during scoliosis progression to assess how vertebral body height discrepancies change during growth. Summary of background data Relative anterior overgrowth has been proposed as a potential driver for AIS initiation and progression. This theory proposes that the anterior column grows faster, and the posterior column slower, in AIS patients when compared to healthy controls. There is disagreement in the literature as to whether the anterior vertebral body heights are proportionally greater than posterior vertebral body heights in AIS patients when compared to healthy controls. To some extent, these discrepancies may be attributed to methodological differences. Methods MRI scans of the major curve of 21 AIS patients (mean age 12.5 ± 1.4 years, mean Cobb 32.2 ± 12.8º) and between T4 and T12 of 21 healthy adolescents (mean age 12.1 ± 0.5 years) were captured for this study. Of the 21 AIS patients, 14 had a second scan on average 10.8 ± 4.7 months after the first. Anterior and posterior vertebral body heights were measured from the true sagittal plane of each vertebra such that anterior overgrowth could be quantified. Results The difference between anterior and posterior vertebral body height in healthy, non-scoliotic children was significantly greater than in AIS patients with mild to moderate scoliosis. However there was no significant relationship between the overall anterior-posterior vertebral body height difference in AIS and either severity of the curve or its progression over time. Conclusions Whilst AIS patients have a proportionally longer anterior column than non-scoliotic controls, the degree of anterior overgrowth was not related to the rate of progression or the severity of the scoliotic curve.

Focused Prefetching: Performance Oriented Prefetching Based On Commit Stalls

Loads that miss in L1 or L2 caches and waiting for their data at the head of the ROB cause significant slow down in the form of commit stalls. We identify that most of these commit stalls are caused by a small set of loads, referred to as LIMCOS (Loads Incurring Majority of COmmit Stalls). We propose simple history-based classifiers that track commit stalls suffered by loads to help us identify this small set of loads. We study an application of these classifiers to prefetching. The classifiers are used to train the prefetcher to focus on the misses suffered by LIMCOS. This, referred to as focused prefetching, results in a 9.8% gain in IPC over naive GHB based delta correlation prefetcher along with a 20.3% reduction in memory traffic for a set of 17 memory-intensive SPEC2000 benchmarks. Another important impact of focused prefetching is a 61% improvement in the accuracy of prefetches. We demonstrate that the proposed classification criterion performs better than other existing criteria like criticality and delinquent loads. Also we show that the criterion of focusing on commit stalls is robust enough across cache levels and can be applied to any prefetcher without any modifications to the prefetcher.

A thermometer for interdependence: Exploring patterns of interdependence using networks of affordances

Interdependence is a central concept in systems and organizations, yet our methods for measuring it are not well developed. Here, we report on a novel method for transforming digital trace data into networks of events that can be used to visualize and measure interdependence. The edges in the network represent sequential flow and the vertices represent actors, actions and artifacts. We refer to this representation as an affordance network. As with conventional approaches such as process mining, our method uses input from a stream of time-stamped occurrences, but the representation is simpler and more appropriate for exploration and theory building. As digital trace data becomes more widely available, this method may become more useful in information systems research and practice. Like a thermometer, it helps us measure a basic property of a system that would otherwise be difficult to see.

Standardization and validation of an induced ovulation model system in buffalo cows: Characterization of gene expression changes in the periovulatory follicle

In bovines characterization of biochemical and molecular determinants of the dominant follicle before and during different time intervals after gonadotrophin surge requires precise identification of the dominant follicle from a follicular wave. The objectives of the present study were to standardize an experimental model in buffalo cows for accurately identifying the dominant follicle of the first wave of follicular growth and characterize changes in follicular fluid hormone concentrations as well as expression patterns of various genes associated with the process of ovulation. From the day of estrus (day 0), animals were subjected to blood sampling and ultrasonography for monitoring circulating progesterone levels and follicular growth. On day 7 of the cycle, animals were administered a PGF2α analogue (Tiaprost Trometamol, 750 μg i.m.) followed by an injection of hCG (2000 IU i.m.) 36 h later. Circulating progesterone levels progressively increased from day 1 of the cycle to 2.26 ± 0.17 ng/ml on day 7 of the cycle, but declined significantly after PGF2α injection. A progressive increase in the size of the dominant follicle was observed by ultrasonography. The follicular fluid estradiol and progesterone concentrations in the dominant follicle were 600 ± 16.7 and 38 ± 7.6 ng/ml, respectively, before hCG injection and the concentration of estradiol decreased to 125.8 ± 25.26 ng/ml, but concentration of progesterone increased to 195 ± 24.6 ng/ml, 24 h post-hCG injection. Inh-α and Cyp19A1 expressions in granulosa cells were maximal in the dominant follicle and declined in response to hCG treatment. Progesterone receptor, oxytocin and cycloxygenase-2 expressions in granulosa cells, regarded as markers of ovulation, were maximal at 24 h post-hCG. The expressions of genes belonging to the super family of proteases were also examined; Cathepsin L expression decreased, while ADAMTS 3 and 5 expressions increased 24 h post-hCG treatment. The results of the current study indicate that sequential treatments of PGF2α and hCG during early estrous cycle in the buffalo cow leads to follicular growth that culminates in ovulation. The model system reported in the present study would be valuable for examining temporo-spatial changes in the periovulatory follicle immediately before and after the onset of gonadotrophin surge.

A hierarchical system of learning automata that can learn die globally optimal path

Systems of learning automata have been studied by various researchers to evolve useful strategies for decision making under uncertainity. Considered in this paper are a class of hierarchical systems of learning automata where the system gets responses from its environment at each level of the hierarchy. A classification of such sequential learning tasks based on the complexity of the learning problem is presented. It is shown that none of the existing algorithms can perform in the most general type of hierarchical problem. An algorithm for learning the globally optimal path in this general setting is presented, and its convergence is established. This algorithm needs information transfer from the lower levels to the higher levels. Using the methodology of estimator algorithms, this model can be generalized to accommodate other kinds of hierarchical learning tasks.

Soil sorption-desorption of phosphorus from piggery effluent compared with inorganic sources

The leaching of phosphorus (P) within soils can be a limiting consideration for the sustainable operation of intensive livestock enterprises. Sorption curves are widely used to assist estimation of P retention, though the effect of effluent constituents on their accuracy is not well understood. We conducted a series of P-sorption-desorption batch experiments with an Oxic Haplustalf (soil 1), Haplusterts (soils 2 and 3), and a Natrustalf (soil 4). Phosphorus sources included effluent, orthophosphate-P in a matrix replicating the effluent's salt constituents (the reference solution), and an orthophosphate-P solution. Treated soils were incubated for up to 193 days before sequential desorption extraction. Effluent constituents, probably the organic or particulate components, temporarily increased the vulnerability of sorbed-P to desorption. The increase in vulnerability was removed by 2-113 days of incubation (25 degrees C). Despite vigorous extraction for 20 consecutive days, some P sorbed as part of the treatments of soils 1 and 2 was not desorbed. The increased vulnerability due to effluent constituents lasted a maximum of about one cropping season and, for all other treatments, adsorption curves overestimated vulnerability to desorption. Therefore, adsorption curves provide a conservative estimate of vulnerability to desorption where effluent is used in continued crop production in these soils.