No abstract

There is increasing interest in discovering individualized treatment rules for patients who have heterogeneous responses to treatment. In particular, one aims to find an optimal individualized treatment rule which is a deterministic function of patient specific characteristics maximizing expected clinical outcome. In this paper, we first show that estimating such an optimal treatment rule is equivalent to a classification problem where each subject is weighted proportional to his or her clinical outcome. We then propose an outcome weighted learning approach based on the support vector machine framework. We show that the resulting estimator of the treatment rule is consistent. We further obtain a finite sample bound for the difference between the expected outcome using the estimated individualized treatment rule and that of the optimal treatment rule. The performance of the proposed approach is demonstrated via simulation studies and an analysis of chronic depression data.

Although the screened group had a higher proportion of patients with pancreatic insufficiency, their growth indices were significantly better than those of the control group during the 13-year follow-up evaluation and, therefore, this randomized clinical trial of early CF diagnosis must be interpreted as unequivocally positive. Our conclusions did not change when the height and weight data before 4 years of age for the controls detected by unblinding were included in the analysis. Also, comparison of growth outcomes after 4 years of age in all subjects showed persistence of the significant differences. Therefore, selection bias has been eliminated as a potential explanation. In addition, the results show that severe malnutrition persists after delayed diagnosis of CF and that catch-up may not be possible. We conclude that early diagnosis of CF through neonatal screening combined with aggressive nutritional therapy can result

BackgroundPrognosis is of critical interest in breast cancer research. Biomedical studies suggest that genomic measurements may have independent predictive power for prognosis. Gene profiling studies have been conducted to search for predictive genomic measurements. Genes have the inherent pathway structure, where pathways are composed of multiple genes with coordinated functions. The goal of this study is to identify gene pathways with predictive power for breast cancer prognosis. Since our goal is fundamentally different from that of existing studies, a new pathway analysis method is proposed.ResultsThe new method advances beyond existing alternatives along the following aspects. First, it can assess the predictive power of gene pathways, whereas existing methods tend to focus on model fitting accuracy only. Second, it can account for the joint effects of multiple genes in a pathway, whereas existing methods tend to focus on the marginal effects of genes. Third, it can accommodate multiple heterogeneous datasets, whereas existing methods analyze a single dataset only. We analyze four breast cancer prognosis studies and identify 97 pathways with significant predictive power for prognosis. Important pathways missed by alternative methods are identified.ConclusionsThe proposed method provides a useful alternative to existing pathway analysis methods. Identified pathways can provide further insights into breast cancer prognosis.

Lactating dairy cows have poor reproductive efficiency because of low fertility and low rates of estrus detection. To eliminate the dependence on detection of estrus, we have recently developed a timed artificial insemination (AI) protocol that synchronized the time of ovulation using GnRH and PGF2 alpha. The effectiveness of this protocol as a management tool was compared with standard reproductive management. Lactating dairy cows (n = 333) from three herds were randomly assigned at parturition to two groups. Control cows were managed according to the typical reproductive strategy of the farm that relied on detection of estrus, the a.m.-p.m. breeding rule, and periodic use of PGF2 alpha. Treated cows had timed AI after synchronization of ovulation with GnRH and PGF2 alpha. For both groups, the voluntary waiting period was 50 d postpartum. Pregnancy diagnosis was performed by ultrasound between 32 and 38 d post-AI. Nonpregnant cows were inseminated again using the original treatment until diagnosed as pregnant or until culled from the herd. Median days to first AI (54 vs. 83) and days open (99 vs. 118) were lower for treated cows than for control cows, respectively. Pregnancy rates for the first AI were similar (37% vs. 39%) for the two groups even though treated cows were bred at an earlier time postpartum. More treated cows than control cows were pregnant at 60 d (37% vs. 5%) and at 100 d (53% vs. 35%) after calving. Thus, this protocol allowed effective management of AI in lactating dairy cows without the need for estrus detection.