Summary Background Emergency contraception can prevent unintended pregnancies, but current methods are only effective if used as soon as possible after sexual intercourse and before ovulation. We ...compared the efficacy and safety of ulipristal acetate with levonorgestrel for emergency contraception. Methods Women with regular menstrual cycles who presented to a participating family planning clinic requesting emergency contraception within 5 days of unprotected sexual intercourse were eligible for enrolment in this randomised, multicentre, non-inferiority trial. 2221 women were randomly assigned to receive a single, supervised dose of 30 mg ulipristal acetate (n=1104) or 1·5 mg levonorgestrel (n=1117) orally. Allocation was by block randomisation stratified by centre and time from unprotected sexual intercourse to treatment, with allocation concealment by identical opaque boxes labelled with a unique treatment number. Participants were masked to treatment assignment whereas investigators were not. Follow-up was done 5–7 days after expected onset of next menses. The primary endpoint was pregnancy rate in women who received emergency contraception within 72 h of unprotected sexual intercourse, with a non-inferiority margin of 1% point difference between groups (limit of 1·6 for odds ratio). Analysis was done on the efficacy-evaluable population, which excluded women lost to follow-up, those aged over 35 years, women with unknown follow-up pregnancy status, and those who had re-enrolled in the study. Additionally, we undertook a meta-analysis of our trial and an earlier study to assess the efficacy of ulipristal acetate compared with levonorgestrel. This trial is registered with ClinicalTrials.gov , number NCT00551616. Findings In the efficacy-evaluable population, 1696 women received emergency contraception within 72 h of sexual intercourse (ulipristal acetate, n=844; levonorgestrel, n=852). There were 15 pregnancies in the ulipristal acetate group (1·8%, 95% CI 1·0–3·0) and 22 in the levonorgestrel group (2·6%, 1·7–3·9; odds ratio OR 0·68, 95% CI 0·35–1·31). In 203 women who received emergency contraception between 72 h and 120 h after sexual intercourse, there were three pregnancies, all of which were in the levonorgestrel group. The most frequent adverse event was headache (ulipristal acetate, 213 events 19·3% in 1104 women; levonorgestrel, 211 events 18·9% in 1117 women). Two serious adverse events were judged possibly related to use of emergency contraception; a case of dizziness in the ulipristal acetate group and a molar pregnancy in the levonorgestrel group. In the meta-analysis (0–72 h), there were 22 (1·4%) pregnancies in 1617 women in the ulipristal acetate group and 35 (2·2%) in 1625 women in the levonorgestrel group (OR 0·58, 0·33–0·99; p=0·046). Interpretation Ulipristal acetate provides women and health-care providers with an effective alternative for emergency contraception that can be used up to 5 days after unprotected sexual intercourse. Funding HRA Pharma.
In the framework of Markov Decision Processes, we consider linear \emph{off-policy} learning, that is the problem of learning a linear approximation of the value function of some fixed policy from ...one trajectory possibly generated by some other policy. We briefly review \emph{on-policy} learning algorithms of the literature (gradient-based and least-squares-based), adopting a unified algorithmic view. Then, we highlight a systematic approach for adapting them to \emph{off-policy} learning \emph{with eligibility traces}. This leads to some known algorithms -- off-policy LSTD($\lambda$), LSPE($\lambda$), TD($\lambda$), TDC/GQ($\lambda$) -- and suggests new extensions -- off-policy FPKF($\lambda$), BRM($\lambda$), gBRM($\lambda$), GTD2($\lambda$). We describe a comprehensive algorithmic derivation of all algorithms in a recursive and memory-efficent form, discuss their known convergence properties and illustrate their relative empirical behavior on Garnet problems. Our experiments suggest that the most standard algorithms on and off-policy LSTD($\lambda$)/LSPE($\lambda$) -- and TD($\lambda$) if the feature space dimension is too large for a least-squares approach -- perform the best.
Given a Markov decision process (MDP) with
n
states and a total number
m
of actions, we study the number of iterations needed by policy iteration (PI) algorithms to
converge to the optimal
γ
...-discounted policy. We consider two variations of PI: Howard’s PI that changes the actions in all states with a positive advantage, and
Simplex-PI that only changes the action in the state with maximal advantage. We show that Howard’s PI terminates after at
most
O
((
m
/(1 −
γ
))log(1/(1 −
γ
))) iterations, improving by a factor
O
(log
n
) a result by Hansen et al. Hansen TD, Miltersen PB, Zwick U (2013) Strategy iteration is strongly polynomial for two-player turn-based stochastic games with a
constant discount factor.
J. ACM
60(1):1:1–1:16., whereas Simplex-PI terminates after at most
O
((
n
m
/(1 −
γ
))log(1/(1 −
γ
))) iterations, improving by a factor
O
(log
n
) a result by Ye Ye Y (2011) The simplex and
policy-iteration methods are strongly polynomial for the Markov decision problem with a fixed discount rate.
Math. Oper. Res.
36(4):593–603.. Under some structural properties of
the MDP, we then consider bounds that are independent of the discount factor
γ
: quantities of interest are bounds
τ
t
and
τ
r
—uniform on all states and policies—respectively, on the
expected time spent in transient states
and
the inverse of the
frequency of visits in recurrent states
given that the process starts from the uniform distribution. Indeed, we show that Simplex-PI terminates after at
most
Õ
(
n
3
m
2
τ
t
τ
r
) iterations.
This extends a recent result for deterministic MDPs by Post and Ye Post I, Ye Y (2013) The simplex method is strongly polynomial for deterministic Markov decision processes. Khanna S, ed.
Proc. 24th
ACM-SIAM Sympos. Discrete Algorithms, SODA '13
(SIAM, Philadelphia), 1465–1473. in which
τ
t
≤ 1 and
τ
r
≤
n
; in particular it shows that Simplex-PI is strongly polynomial for a much larger class of MDPs. We explain why similar results seem hard to derive for Howard’s PI. Finally, under the additional (restrictive) assumption that the state space is partitioned in two sets, respectively, states that are transient and recurrent for all policies, we show that both Howard’s PI and Simplex-PI terminate after at most
Õ
(
m
(
n
2
τ
t
+
n
τ
r
)) iterations.
Modified policy iteration (MPI) is a dynamic programming (DP) algorithm that contains the two celebrated policy and value iteration methods. Despite its generality, MPI has not been thoroughly ...studied, especially its approximation form which is used when the state and/or action spaces are large or infinite. In this paper, we propose three implementations of approximate MPI (AMPI) that are extensions of the well-known approximate DP algorithms:~fitted-value iteration, fitted-Q iteration, and classification-based policy iteration. We provide error propagation analysis that unify those for approximate policy and value iteration. We develop the finite-sample analysis of these algorithms, which highlights the influence of their parameters. In the classification-based version of the algorithm (CBMPI), the analysis shows that MPI's main parameter controls the balance between the estimation error of the classifier and the overall value function approximation. We illustrate and evaluate the behavior of these new algorithms in the Mountain Car and Tetris problems. Remarkably, in Tetris, CBMPI outperforms the existing DP approaches by a large margin, and competes with the current state-of-the-art methods while using fewer samples.
We consider the discrete-time infinite-horizon optimal control problem formalized by Markov decision processes (Puterman, 1994; Bertsekas and Tsitsiklis, 1996). We revisit the work of Bertsekas and ...Ioffe (1996), that introduced $\lambda$ policy iteration-a family of algorithms parametrized by a parameter $\lambda$-that generalizes the standard algorithms value and policy iteration, and has some deep connections with the temporal-difference algorithms described by Sutton and Barto (1998). We deepen the original theory developed by the authors by providing convergence rate bounds which generalize standard bounds for value iteration described for instance by Puterman (1994). Then, the main contribution of this paper is to develop the theory of this algorithm when it is used in an approximate form. We extend and unify the separate analyzes developed by Munos for approximate value iteration (Munos, 2003) and approximate policy iteration (Munos, 2003), and provide performance bounds in the discounted and the undiscounted situations. Finally, we revisit the use of this algorithm in the training of a Tetris playing controller as originally done by Bertsekas and Ioffe (1996).. Our empirical results are different from those of Bertsekas and Ioffe (which were originally qualified as "paradoxical" and "intriguing"). We track down the reason to be a minor implementation error of the algorithm, which suggests that, in practice, $\lambda$ policy iteration may be more stable than previously thought.
Abstract Background Emergency contraception (EC) does not always work. Clinicians should be aware of potential risk factors for EC failure. Study Design Data from a meta-analysis of two randomized ...controlled trials comparing the efficacy of ulipristal acetate (UPA) with levonorgestrel were analyzed to identify factors associated with EC failure. Results The risk of pregnancy was more than threefold greater for obese women compared with women with normal body mass index (odds ratio (OR), 3.60; 95% confidence interval (CI), 1.96–6.53; p<.0001), whichever EC was taken. However, for obese women, the risk was greater for those taking levonorgestrel (OR, 4.41; 95% CI, 2.05–9.44, p=.0002) than for UPA users (OR, 2.62; 95% CI, 0.89–7.00; ns ). For both ECs, pregnancy risk was related to the cycle day of intercourse. Women who had intercourse the day before estimated day of ovulation had a fourfold increased risk of pregnancy (OR, 4.42; 95% CI, 2.33–8.20; p<.0001) compared with women having sex outside the fertile window. For both methods, women who had unprotected intercourse after using EC were more likely to get pregnant than those who did not (OR, 4.64; 95% CI, 2.22–8.96; p=.0002). Conclusions Women who have intercourse around ovulation should ideally be offered a copper intrauterine device. Women with body mass index >25 kg/m2 should be offered an intrauterine device or UPA. All women should be advised to start effective contraception immediately after EC.
Few therapeutic options are available to treat the late-stage of human African trypanosomiasis, a neglected tropical disease, caused by Trypanosoma brucei gambiense (g-HAT). The firstline treatment ...is a combination therapy of oral nifurtimox and intravenous eflornithine that needs to be administered in a hospital setting by trained personnel, which is not optimal given that patients often live in remote areas with few health resources. Therefore, we aimed to assess the safety and efficacy of an oral regimen of fexinidazole (a 2-substituted 5-nitroimidazole with proven trypanocidal activity) versus nifurtimox eflornithine combination therapy in patients with late-stage g-HAT.
In this randomised, phase 2/3, open-label, non-inferiority trial, we recruited patients aged 15 years and older with late-stage g-HAT from g-HAT treatment centres in the Democratic Republic of the Congo (n=9) and the Central African Republic (n=1). Patients were randomly assigned (2:1) to receive either fexinidazole or nifurtimox eflornithine combination therapy according to a predefined randomisation list (block size six). The funder, data management personnel, and study statisticians were masked to treatment. Oral fexinidazole was given once a day (days 1–4: 1800 mg, days 5–10: 1200 mg). Oral nifurtimox was given three times a day (days 1–10: 15 mg/kg per day) with eflornithine twice a day as 2 h infusions (days 1–7: 400 mg/kg per day). The primary endpoint was success at 18 months (ie, deemed as patients being alive, having no evidence of trypanosomes in any body fluid, not requiring rescue medication, and having a cerebrospinal fluid white blood cell count ≤20 cells per μL). Safety was assessed through routine monitoring. Primary efficacy analysis was done in the modified intention-to-treat population and safety analyses in the intention-to-treat population. The acceptable margin for the difference in success rates was defined as 13%. This study has been completed and is registered with ClinicalTrials.gov, number NCT01685827.
Between October, 2012, and November, 2016, 419 patients were pre-screened. Of the 409 eligible patients, 14 were not included because they did not meet all inclusion criteria (n=12) or for another reason (n=2). Therefore, 394 patients were randomly assigned, 264 to receive fexinidazole and 130 to receive nifurtimox eflornithine combination therapy. Success at 18 months was recorded in 239 (91%) patients given fexinidazole and 124 (98%) patients given nifurtimox eflornithine combination therapy, within the margin of acceptable difference of −6·4% (97·06% CI −11·2 to −1·6; p=0·0029). We noted no difference in the proportion of patients who experienced treatment-related adverse events (215 81% in the fexinidazole group vs 102 79% in the nifurtimox eflornithine combination therapy group). Treatment discontinuations were unrelated to treatment (n=2 1% in the fexinidazole group). Temporary nifurtimox eflornithine combination therapy interruption occurred in three (2%) patients. 11 patients died during the study (nine 3% in the fexinidazole group vs two 2% in the nifurtimox eflornithine combination therapy group).
Our findings show that oral fexinidazole is effective and safe for the treatment of T b gambiense infection compared with nifurtimox eflornithine combination therapy in late-stage HAT patients. Fexinidazole could be a key asset in the elimination of this fatal neglected disease.
Drugs for Neglected Diseases initiative.
In some situations, practice guidelines do not provide firm evidence-based guidance regarding COPD treatment choices, especially when large trials have failed to identify subgroups of particularly ...good or poor responders to available medications.
This observational cross-sectional study explored the yield of four types of multidimensional analyses to assess the associations between the clinical characteristics of COPD patients and pharmacological and non-pharmacological treatments prescribed by lung specialists in a real-life context.
Altogether, 2494 patients were recruited by 515 respiratory physicians. Multiple correspondence analysis and hierarchical clustering identified 6 clinical subtypes and 6 treatment subgroups. Strong bi-directional associations were found between clinical subtypes and treatment subgroups in multivariate logistic regression. However, although the overall frequency of prescriptions varied from one clinical subtype to the other for all types of pharmacological treatments, clinical subtypes were not associated with specific prescription profiles. When canonical analysis of redundancy was used, the proportion of variation in pharmacological treatments that was explained by clinical characteristics remained modest: 6.23%. This proportion was greater (14.29%) for non-pharmacological components of care.
This study shows that, although pharmacological treatments of COPD are quantitatively very well related to patients' clinical characteristics, there is no particular patient profile that could be qualitatively associated to prescriptions. This underlines uncertainties perceived by physicians for differentiating the respective effects of available pharmacological treatments. The methodology applied here is useful to identify areas of uncertainty requiring further research and/or guideline clarification.
Summary Background Spinal muscular atrophy (SMA) is a progressive motor neuron disease causing loss of motor function and reduced life expectancy, for which limited treatment is available. We ...investigated the safety and efficacy of olesoxime in patients with type 2 or non-ambulatory type 3 SMA. Methods This randomised, double-blind, placebo-controlled, phase 2 study was done in 22 neuromuscular care centres in Belgium, France, Germany, Italy, Netherlands, Poland, and the UK. Safety and efficacy of olesoxime were assessed in patients aged 3–25 years with genetically confirmed type 2 or non-ambulatory type 3 SMA. A centralised, computerised randomisation process allocated patients (2:1 with stratification by SMA type and centre) to receive olesoxime (10 mg/kg per day) in an oral liquid suspension or placebo for 24 months. Patients, investigators assessing outcomes, and sponsor study personnel were masked to treatment assignment. The primary outcome measure was change from baseline compared with 24 months between the two treatment groups in functional domains 1 and 2 of the Motor Function Measure (MFM D1 + D2) assessed in the full analysis population. A shorter, 20-item version of the MFM, which was specifically adapted for young children, was used to assess patients younger than 6 years. Safety was assessed in the intention-to-treat population. The trial is registered with ClinicalTrials.gov , number NCT01302600. Findings The trial was done between Nov 18, 2010, and Oct 9, 2013. Of 198 patients screened, 165 were randomly assigned to olesoxime (n=108) or placebo (n=57). Five patients in the olesoxime group were not included in the primary outcome analysis because of an absence of post-baseline assessments. The change from baseline to month 24 on the primary outcome measure was 0·18 for olesoxime and −1·82 for placebo (treatment difference 2·00 points, 96% CI −0·25 to 4·25, p=0·0676). Olesoxime seemed to be safe and generally well tolerated, with an adverse event profile similar to placebo. The most frequent adverse events in the olesoxime group were pyrexia (n=34), cough (n=32), nasopharyngitis (n=25), and vomiting (n=25). There were two patient deaths (one in each group), but these were not deemed to be related to the study treatment. Interpretation Olesoxime was safe at the doses studied, for the duration of the trial. Although the primary endpoint was not met, secondary endpoints and sensitivity analyses suggest that olesoxime might maintain motor function in patients with type 2 or type 3 SMA over a period of 24 months. Based on these results, olesoxime might provide meaningful clinical benefits for patients with SMA and, given its mode of action, might be used in combination with other drugs targeting other mechanisms of disease, although additional evidence is needed. Funding AFM Téléthon and Trophos SA.
PDF
