Strategies for sequential prediction of stationary time series

We present simple procedures for the prediction of a real valued sequence. The algorithms are based on a combinationof several simple predictors. We show that if the sequence is a realization of a bounded stationary and ergodic random process then the average of squared errors converges, almost surely, to that of the optimum, given by the Bayes predictor. We offer an analog result for the prediction of stationary gaussian processes.

A sequential model for older workers’ labor transitions after a health shock

In this work we study older workers (50 64) labor force transitions after a health/disability shock. We find that the probability of keeping working decreases with both age and severity of the shock. Moreover, we find strong interactions between age and severity in the 50 64 age range and none in the 30 49 age range. Regarding demographics we find that being female and married reduce the probability of keeping work. On the contrary, being main breadwinner, education and skill levels increase it. Interestingly, the effect of some demographics changes its sign when we look at transitions from inactivity to work. This is the case of being married or having a working spouse. Undoubtedly, leisure complementarities should play a role in the latter case. Since the data we use contains a very detailed information on disabilities, we are able to evaluate the marginal effect of each type of disability either in the probability of keeping working or in returning back to work. Some of these results may have strong policy implications.

Choice by sequential procedures

We propose a rule of decision-making, the sequential procedure guided byroutes, and show that three influential boundedly rational choice models can be equivalentlyunderstood as special cases of this rule. In addition, the sequential procedure guidedby routes is instrumental in showing that the three models are intimately related. We showthat choice with a status-quo bias is a refinement of rationalizability by game trees, which, inturn, is also a refinement of sequential rationalizability. Thus, we provide a sharp taxonomyof these choice models, and show that they all can be understood as choice by sequentialprocedures.

Valuation bubbles and sequential bubbles

Price bubbles in an Arrow-Debreu valuation equilibrium in infinite-timeeconomy are a manifestation of lack of countable additivity of valuationof assets. In contrast, known examples of price bubbles in sequentialequilibrium in infinite time cannot be attributed to the lack of countableadditivity of valuation. In this paper we develop a theory of valuation ofassets in sequential markets (with no uncertainty) and study the nature ofprice bubbles in light of this theory. We consider an operator, calledpayoff pricing functional, that maps a sequence of payoffs to the minimumcost of an asset holding strategy that generates it. We show that thepayoff pricing functional is linear and countably additive on the set ofpositive payoffs if and only if there is no Ponzi scheme, and providedthat there is no restriction on long positions in the assets. In the knownexamples of equilibrium price bubbles in sequential markets valuation islinear and countably additive. The presence of a price bubble indicatesthat the asset's dividends can be purchased in sequential markers at acost lower than the asset's price. We also present examples of equilibriumprice bubbles in which valuation is nonlinear but not countably additive.

A characterization of sequential rationalizability

A choice function is sequentially rationalizable if there is an ordered collection of asymmetric binary relations that identifies the selected alternative in every choice problem. We propose a property, F-consistency, and show that it characterizes the notion of sequential rationalizability. F-consistency is a testable property that highlights the behavioral aspects implicit in sequentially rationalizable choice. Further, our characterization result provides a novel tool with which to study how other behavioral concepts are related to sequential rationalizability, and establish a priori unexpected implications. In particular, we show that the concept of rationalizability by game trees, which, in principle, had little to do with sequential rationalizability, is a refinement of the latter. Every choice function that is rationalizable by a game tree is also sequentially rationalizable. Finally, we show that some prominent voting mechanisms are also sequentially rationalizable.

A simple randomized algorithm for consistent sequential prediction of ergodic time series

We present a simple randomized procedure for the prediction of a binary sequence. The algorithm uses ideas from recent developments of the theory of the prediction of individual sequences. We show that if thesequence is a realization of a stationary and ergodic random process then the average number of mistakes converges, almost surely, to that of the optimum, given by the Bayes predictor.

Sequential screening

We present a model of price discrimination where a monopolistfaces a consumer who is privately informed about thedistribution of his valuation for an indivisible unit ofgood but has yet to learn privately the actual valuation.The monopolist sequentially screens the consumer with amenu of contracts:the consumer self-selects once by choosing a contract andthen self-selects again when he learns the actual valuation. A deterministic sequential mechanism is a menu of refundcontracts, each consisting of an advance payment and a refundamount in case of no consumption, but sequential mechanismsmay involve randomization.We characterize the optimal sequential mechanism when someconsumer types are more eager in the sense of first-orderstochastic dominance, and when some types face greatervaluation uncertainty in the sense of mean-preserving-spread.We show that it can be optimal to subsidize consumer typeswith smaller valuation uncertainty (through low refund, as inairplane ticket pricing) in order to reduce the rent to thosewith greater uncertainty. The size of distortion depends bothon the type distribution and on how informative the consumer'sinitial private knowledge is about his valuation, but noton how much he initially knows about the valuation per se.

A zero-delay sequential scheme for lossy coding of individual sequences

We consider adaptive sequential lossy coding of bounded individual sequences when the performance is measured by the sequentially accumulated mean squared distortion. Theencoder and the decoder are connected via a noiseless channel of capacity $R$ and both are assumed to have zero delay. No probabilistic assumptions are made on how the sequence to be encoded is generated. For any bounded sequence of length $n$, the distortion redundancy is defined as the normalized cumulative distortion of the sequential scheme minus the normalized cumulative distortion of the best scalarquantizer of rate $R$ which is matched to this particular sequence. We demonstrate the existence of a zero-delay sequential scheme which uses common randomization in the encoder and the decoder such that the normalized maximum distortion redundancy converges to zero at a rate $n^{-1/5}\log n$ as the length of the encoded sequence $n$ increases without bound.

Phase 3 randomized trial on larynx preservation comparing sequential vs alternating chemotherapy and radiotherapy.

BACKGROUND: Both induction chemotherapy followed by irradiation and concurrent chemotherapy and radiotherapy have been reported as valuable alternatives to total laryngectomy in patients with advanced larynx or hypopharynx cancer. We report results of the randomized phase 3 trial 24954 from the European Organization for Research and Treatment of Cancer. METHODS: Patients with resectable advanced squamous cell carcinoma of the larynx (tumor stage T3-T4) or hypopharynx (T2-T4), with regional lymph nodes in the neck staged as N0-N2 and with no metastasis, were randomly assigned to treatment in the sequential (or control) or the alternating (or experimental) arm. In the sequential arm, patients with a 50% or more reduction in primary tumor size after two cycles of cisplatin and 5-fluorouracil received another two cycles, followed by radiotherapy (70 Gy total). In the alternating arm, a total of four cycles of cisplatin and 5-fluorouracil (in weeks 1, 4, 7, and 10) were alternated with radiotherapy with 20 Gy during the three 2-week intervals between chemotherapy cycles (60 Gy total). All nonresponders underwent salvage surgery and postoperative radiotherapy. The Kaplan-Meier method was used to obtain time-to-event data. RESULTS: The 450 patients were randomly assigned to treatment (224 to the sequential arm and 226 to the alternating arm). Median follow-up was 6.5 years. Survival with a functional larynx was similar in sequential and alternating arms (hazard ratio of death and/or event = 0.85, 95% confidence interval = 0.68 to 1.06), as were median overall survival (4.4 and 5.1 years, respectively) and median progression-free interval (3.0 and 3.1 years, respectively). Grade 3 or 4 mucositis occurred in 64 (32%) of the 200 patients in the sequential arm who received radiotherapy and in 47 (21%) of the 220 patients in the alternating arm. Late severe edema and/or fibrosis was observed in 32 (16%) patients in the sequential arm and in 25 (11%) in the alternating arm. CONCLUSIONS: Larynx preservation, progression-free interval, and overall survival were similar in both arms, as were acute and late toxic effects.

The Sequential Organ Failure Assessment score and copeptin for predicting survival in ventilator-associated pneumonia.

INTRODUCTION: Ventilator-associated pneumonia remains the most common nosocomial infection in the critically ill and contributes to significant morbidity. Eventual decisions regarding withdrawal or maximal therapy are demanding and rely on physicians' experience. Additional objective tools for risk assessment may improve medical judgement. Copeptin, reflecting vasopressin release, as well as the Sequential Organ Failure Assessment (SOFA) score, reflecting the individual degree of organ dysfunction, might qualify for survival prediction in ventilator-associated pneumonia. We investigated the predictive value of the SOFA score and copeptin in ventilator-associated pneumonia. METHODS: One hundred one patients with ventilator-associated pneumonia were prospectively assessed. Death within 28 days after ventilator-associated pneumonia onset was the primary end point. RESULTS: The SOFA score and the copeptin levels at ventilator-associated pneumonia onset were significantly elevated in nonsurvivors (P = .002 and P = .017, respectively). Both markers had different time courses in survivors and nonsurvivors (P < .001 and P = .006). Mean SOFA (average SOFA of 10 days after VAP onset) was superior in predicting 28-day survival as compared with SOFA and copeptin at ventilator-associated pneumonia onset (area under the curve, 0.90 vs 0.73 and 0.67, respectively). CONCLUSIONS: The predictive value of serial-measured SOFA significantly exceeds those of single SOFA and copeptin measurements. Serial SOFA scores accurately predict outcome in ventilator-associated pneumonia.

Nitrogen application timing and soil inorganic nitrogen dynamics under no-till oat/maize sequential cropping

The timing of N application to maize is a key factor to be considered in no-till oat/maize sequential cropping. This study aimed to evaluate the influence of pre-planting, planting and sidedress N application on oat residue decomposition, on soil N immobilisation and remineralisation and on N uptake by maize plants in no-till oat/maize sequential cropping. Undisturbed soil cores of 10 and 20 cm diameter were collected from the 0-15 cm layer of a no-till Red Latossol, when the oat cover crop was in the milk-grain stage. Two greenhouse experiments were conducted simultaneously. Experiment A, established in the 10 cm diameter cores and without plant cultivation, was used to asses N dynamics in soil and oat residues. Experiment B, established in the 20 cm diameter cores and with maize cultivation, was used to assess plant growth and N uptake. An amount of 6.0 Mg ha-1 dry matter of oat residues was spread on the surface of the cores. A rate of 90 kg N ha-1 applied as ammonium sulphate in both experiments was split in pre-planting, planting and sidedress applications as follows: (a) 00-00-00 (control), (b) 90-00-00 (pre-planting application, 20 days before planting), (c) 00-90-00 (planting application), (d) 00-30-60 (split in a planting and a sidedress application 31 days after emergence), (e) 00-00-00* (control, without oat residue) and (f) 90-00-00* (pre-planting application, without oat residue). The N concentration and N content in oat residues were not affected during decomposition by N fertilisation. Most of the fertiliser NH4+-N was converted into NO3--N within 20 days after application. A significant decrease in NO3--N contents in the 0-4 cm layer was observed in all treatments between 40 and 60 days after the oat residue placement on the soil surface, suggesting the occurrence of N immobilisation in this period. Considering that most of the inorganic N was converted into NO3- and that no immobilisation of the pre planting fertiliser N occurred at the time of its application, it was possible to conclude that pre-planting applied N was prone to losses by leaching. On the other hand, with split N applications, maize plants showed N deficiency symptoms before sidedress application. Two indications for fertiliser-N management in no-till oat/maize sequential cropping could be suggested: (a) in case of split application, the sidedress should be earlier than 30 days after emergence, and (b) if integral application is preferred to save field operations, this should be done at planting.

Sequential or concomitant chemotherapy in limited stage small-cell lung cancer.

PURPOSE: Chemotherapy (CT) combined with radiation therapy (RT) is the standard treatment for limited disease small-cell lung cancer (LDSCLC). Many questions including RT dose, fractionation, and sequence of RT/CT administration remain controversial. In this paper, we retrospectively assessed the outcome of patients with LDSCLC treated with radiation of at least 50 Gy.METHODS AND MATERIALS: From December 1997 to January 2006, 69 consecutive patients with LDSCLC were treated at our institutions. Treatment consisted of at least 4 cycles of CT, and 3D conformal thoracic RT. The median age was 61 years (range, 37-78 years). Sequential or concomitant CT/RT was given in 47 (68%) and 22 (32%) of the patients, respectively. The median RT dose was 60 Gy. Prophylactic cranial irradiation (PCI) was administered in 47 (68%) patients.RESULTS: With a median follow-up of 36 months (range, 6-107), 16 patients were alive without disease. The median overall survival time was 24 months, with a 3-year survival rate of 29%. The 3-year disease-free survival (DFS) and loco-regional control (LRC) rates were 23% and 60%, respectively. A better DFS was significantly associated with performance status (PS) 0 (p = 0.004), complete response to treatment (p = 0.03), and PCI group (p = 0.03). A trend towards improved overall survival (OS) was observed for patients who underwent PCI (p = 0.07). Patients treated with sequential CT/RT had a better outcome than those treated with concomitant treatment (3-year DFS rate 27% vs. 13%; p = 0.04). However, PCI was delivered more frequently for the sequential group. No significant dose-response relationship was found in terms of LRC. The multivariate analysis showed that complete response to treatment was the only significant factor for OS.CONCLUSION: Complete response to treatment was the most important factor for OS. A better DFS was significantly associated with the PCI group. We did not find a significant difference in outcome between patients receiving doses of 60 Gy or more and patients receiving 60 Gy or less.