Deep advantage learning for optimal dynamic treatment regime

Recently deep learning has successfully achieved state-of-the-art performance on many difficulttasks. Deep neural networks allow for model flexibility and process features without the needof domain knowledge. Advantage learning (A-learning) is a popular method in dynamic treatment regime (DTR). It models the advantage function, which is of direct relevance to optimaltreatment decision. No assumptions on baseline function are made. However, there is a paucityof literature on deep A-learning. In this paper, we present a deep A-learning approach to estimate optimal DTR. We use an inverse probability weighting method to estimate the differencebetween potential outcomes. Parameter sharing of convolutional neural networks (CNN) greatlyreduces the amount of parameters in neural networks, which allows for high scalability. Convexified convolutional neural networks (CCNN) relax the constraints of CNN for optimisation purpose.Different architectures of CNN and CCNN are implemented for contrast function estimation.Both simulation results and application to the STAR*D (Sequenced Treatment Alternatives toRelieve Depression) trial indicate that the proposed methods outperform penalised least squareestimator.

Effects of water regime, crop residues, and application rates on control of Fusarium oxysporum f.sp.cubense

Biological soil disinfestation is an effective method to control soil-borne disease by flooding and incorporating with organic amendments, but field conditions and resources sometimes limited its practical application. A laboratory experiment was conducted to develop practice guidelines on controlling Fusarium wilt, a widespread banana disease caused by Fusarium oxysporum f. sp. cubense（FOC）. FOC infested soil incorporated with rice or maize straw at rates of 1.5 tons/ha and 3.0 tons/ha was incubated under flooded or water-saturated（100% water holding capacity） conditions at 30℃ for 30 days. Results showed that FOC populations in the soils incorporated with either rice or maize straw rapidly reduced more than 90% in the first 15 days and then fluctuated till the end of incubation, while flooding alone without organic amendment reduced FOC populations slightly. The rapid and dramatic decrease of redox potential（down to- 350 m V） in straw-amended treatments implied that both anaerobic condition and strongly reductive soil condition would contribute to pathogen inactivation. Water-saturation combined with straw amendments had the comparable effects on reduction of FOC, indicating that flooding was not indispensable for inactivating FOC. There was no significant difference in the reduction of FOC observed in the straw amendments at between 1.5 and 3 tons/ha. Therefore,incorporating soil with straw（rice or maize straw） at a rate of 3.0 tons/ha under 100%water holding capacity or 1.5 tons/ha under flooding, would effectively alleviate banana Fusarium wilt caused by FOC after 15-day treating under 30℃.

Efficacy and safety of human chorionic gonadotropin combined with human menopausal gonadotropin and a gonadotropin-releasing hormone pump for male adolescents with congenital hypogonadotropic hypogonadism

Background:Compared to adult studies,studies which involve the treatment of pediatric congenital hypogonadotropic hypogonadism(CHH)are limited and no universal treatment regimen is available.The aim of this study was to evaluate the feasibility of human chorionic gonadotropin(hCG)/human menopausal gonadotropin(hMG)therapy for treating male adolescents with CHH.Methods:Male adolescent CHH patients were treated with hCG/hMG(n=20)or a gonadotropin-releasing hormone(GnRH)pump(n=21).The treatment was divided into a study phase(0-3 months)and a follow-up phase(3-12 months).The testicular volume(TV),penile length(PL),penis diameter(PD),and sex hormone levels were compared between the two groups.The TV and other indicators between the groups were analyzed using a t-test(equal variance)or a rank sum test(unequal variance).Results:Before treatment,there was no statistical difference between the two groups in terms of the biochemistry,hormones,and other demographic indicators.After 3 months of treatment,the TV of the hCG/hMG and GnRH groups increased to 5.1±2.3 mL and 4.1±1.8 mL,respectively;however,the difference was not statistically significant(P>0.05,t=1.394).The PL reached 6.9±1.8 cm and 5.1±1.6 cm(P<0.05,t=3.083),the PD reached 2.4±0.5 cm and 2.0±0.6 cm(P<0.05,t=2.224),respectively,in the two groups.At the end of 6 months of treatment,biomarkers were in normal range in the two groups.Compared with the GnRH group,the testosterone(T)level and growth of PL and PD were significantly greater in the hCG/hMG group(all P<0.05).While the TV of both groups increased,the difference was not statistically significant(P>0.05,t=0.314).After 9 to 12 months of treatment,the T level was higher in the hCG/hMG group.Other parameters did not exhibit a statistical difference.Conclusions:The hCG/hMG regimen is feasible and effective for treating male adolescents with CHH.The initial 3 months of treatment may be a window to optimally observe the strongest effects of therapy.Furthermore,results from the extended time-period showed positive outcomes at the 1-year mark;however,the long-term effectiveness,strengths,and weaknesses of the hCG/hMG regimen require further research.Trial Registration:ClinicalTrials.gov,NCT02880280;https://clinicaltrials.gov/ct2/show/NCT02880280.