Classification using ensemble learning under weighted misclassification loss

Binary classification rules based on covariates typically depend on simple loss functions such as zero-one misclassification. Some cases may require more complex loss functions. For example, individual-level monitoring of HIV-infected individuals on antiretroviral therapy requires periodic assessment of treatment failure, defined as having a viral load (VL) value above a certain threshold. In some resource limited settings, VL tests may be limited by cost or technology, and diagnoses are based on other clinical markers. Depending on scenario, higher premium may be placed on avoiding false-positives, which brings greater cost and reduced treatment options. Here, the optimal rule is determined by minimizing a weighted misclassification loss/risk. We propose a method for finding and cross-validating optimal binary classification rules under weighted misclassification loss. We focus on rules comprising a prediction score and an associated threshold, where the score is derived using an ensemble learner. Simulations and examples show that our method, which derives the score and threshold jointly, more accurately estimates overall risk and has better operating characteristics compared with methods that derive the score first and the cutoff conditionally on the score especially for finite samples.     

Quantification of print,radio and television exposure among previous blood donors in Kenya: an opportunity for encouraging repeat donation in a resource‐limited setting?

Blood services in sub‐Saharan Africa experience blood shortages and low retention of voluntary, non‐remunerated donors. To boost collections by encouraging repeat donations, the Kenya National Blood Transfusion Service is exploring the likelihood of reaching previous donors through targeted print, radio and television advertising. We analysed data from a national AIDS Indicator Survey to determine whether previous donors have significant exposure to media. Respondents reporting history of blood donation had significantly higher exposure to print, radio and television media than those without history of blood donation. Targeted media campaigns encouraging repeat donation are likely to reach previous donors even in resource‐limited settings.     

Y染色体或将在进化中消失

美国宾夕法尼亚州立大学的2位科学家在性染色体进化变异机制的研究上取得进展。研究发现,Y染色体比X染色体的演化速率快得多,这将导致Y染色体上的基因急剧丢失,如此,Y染色体将会完全消失,人类的传宗接代将受威胁。     

Early recognition of heart failure in patients with diabetes type 2 in primary care. A prospective diagnostic efficiency study. (UHFO-DM2)

Background  

We hypothesize that the prevalence of unknown heart failure in diabetic patients aged 60 years and over is relatively high (15% or more) and that a cost-effective strategy can be developed to detect heart failure in these patients. The strategy is expected to include some signs and symptoms (such as dyspnoea, orthopnoea, pulmonary crepitations and laterally displaced apical beat), natriuretic peptide measurements (Amino-terminal B-type natriuretic peptide) and possibly electrocardiography. In a subset of patients straightforward echocardiography may show to be cost-effective. With information from our study the detection of previously unknown heart failure in diabetic patients could be improved and enable the physician to initiate beneficial morbidity and mortality reducing heart failure treatment more timely.     

Intradermal Vaccinations With RNA Coding for TAA Generate CD8+ and CD4+ Immune Responses and Induce Clinical Benefit in Vaccinated Patients

The aim of this phase I/II nonrandomized trial was to assess feasibility, safety as well as immunological and clinical responses of a mRNA-based vaccination in patients with stage IV renal cell cancer using granulocyte-macrophage colony stimulating factor (GM-CSF) as adjuvant. Intradermal injections of in vitro transcribed naked mRNA, which was generated using plasmids coding for the tumor-associated antigens mucin 1(MUC1), carcinoembryonic (CEA), human epidermal growth factor receptor 2 (Her-2/neu), telomerase, survivin, and melanoma-associated antigen 1 (MAGE-A1) were performed in 30 enrolled patients. In the first 14 patients (cohort A) vaccinations were administered on days 0, 14, 28, and 42 (20 µg/antigen) while in the consecutive 16 patients (cohort B) an intensified protocol consisting of injections at days 0–3, 7–10, 28, and 42 (50 µg/antigen) was used. In both cohorts, after this induction period, vaccinations were repeated monthly until tumor progression analyzed by Response Evaluation Criteria In Solid Tumors criteria (RECIST). Vaccinations were well tolerated with no severe side effects and induced clinical responses [six stable diseases (SD) and one partial response in cohort A and nine SD in cohort B]. In cohort A, 35.7% survived 4 years (median survival 24 months) compared to 31.25% in cohort B (median survival 29 months). Induction of CD4⁺ and CD8⁺ T cell responses was shown for several tumor-associated antigens (TAA) using interferon-γ (IFN-γ) enzyme-linked immunosorbent spot (ELISpot) and Cr-release assays.     

Quality of HIV laboratory testing in Tanzania: a situation analysis

Tanzania is scaling up prevention, treatment, care and support of individuals affected with HIV. There is therefore a need for high quality and reliable HIV infection testing and AIDS staging. The objective of this study was to assess laboratories capacities of services in terms of HIV testing and quality control. A baseline survey was conducted from December 2004 to February 2005 in 12 laboratories which were conveniently selected to represent all the zones of Tanzania. The questionnaires comprised of questions on laboratory particulars, internal and external quality control for HIV testing and quality control of reagents. Source and level of customer satisfaction of HIV test kits supply was established. Of 12 laboratories, nine used rapid tests for screening and two used rapid tests for diagnosis. In the 12 laboratories, four used double ELISA and five used single ELISA and three did not use ELISA. Confirmatory tests observed were Western Blot in three laboratories, DNA PCR in two laboratories, CD4 counting in seven laboratories, and viral load in two laboratories. Although all laboratories conducted quality control (QC) of the HIV kits, only two laboratories had Standard Operating Procedures (SOPs). Internal and external quality control (EQC) was done at varied proportions with the highest frequency of 55.6% (5/9) for internal quality control (IQC) for rapid tests and EQC for ELISA, and the lowest frequency of 14.3% (1/ 7) for IQC for CD4 counting. None of the nine laboratories which conducted QC for reagents used for rapid tests and none of the five which performed IQC and EQC had SOPs. HIV kits were mainly procured by the Medical Store Department and most of laboratories were not satisfied with the delay in procurement procedures. Most of the laboratories used rapid tests only, while some used both rapid tests and ELISA method for HIV testing. In conclusion, the survey revealed inadequacy in Good Laboratory Practice and poor laboratory quality control process for HIV testing reagents, internal and external quality control.     

Chagas' disease diagnosis: evaluation of several tests in blood bank screening

Blood transfusion is one of the principal routes of transmission of Chagas' disease, a major endemic disease in Latin America. Methods for blood screening are not accurate and may yield false results that lead to high social and economic costs. This study compares two methods of diagnosing Chagas' disease (indirect immunofluorescence and hemagglutination) and several enzyme-linked immunosorbent assays (ELISAs) with regard to specificity and sensitivity, by using human sera with known serologic and parasitologic characteristics, as well as samples with discrepant results on conventional serologic tests. An ELISA using recombinant antigens showed no cross-reactivity with sera that were positive for other diseases. All evaluated ELISAs performed well, and their use may lead to a reduction of more than 50 percent in the number of discordant sera. Further improvements are needed in view of the complexity of the serologic diagnosis of Chagas' disease.     

Virtual colonoscopy with magnetic resonance imaging: In vitro evaluation of a new concept

BACKGROUND & AIMS: Screening for colonic polyps is desirable. A new concept based on cross-sectional and endoscopic analysis of a magnetic resonance (MR) data set is presented. METHODS: Ex vivo autopsy colonic specimens, containing artificially placed polyps, were obtained and filled with a gadolinium-containing solution. Forty-four thin-section MR images were obtained in a 1.5-T MR scanner in 28 seconds. A three- dimensional endoscopic fly-through of these images was rendered. Fly- throughs and two-dimensional cross-sectional images were analyzed by two observers for the presence of polyps. RESULTS: The average sensitivity and specificity for the detection of polyps based on three- dimensional endoscopic MR colon imaging were 87% and 96%, respectively. Analysis of cross-sectional images showed an overall sensitivity and specificity of merely 57% and 84%, respectively. The difference in the interpretation of three-dimensional MR colonoscopy and two-dimensional cross-sections was statistically significant (P < 0.001). With three- dimensional MR colonoscopy, overall sensitivity for detection of polyps measuring < or =5 mm in length and diameter was 70%; for larger polyps, it increased to 95% (P < 0.01). CONCLUSIONS: The feasibility of an MR- based endoluminal assessment of the colon is shown. Minimal invasiveness, lack of radiation exposure, and high in vitro diagnostic accuracy warrant further investigation of this novel concept. (Gastroenterology 1997 Jun;112(6):1863-70)     

Phase II trial of 2-chlorodeoxyadenosine for the treatment of cutaneous T-cell lymphoma [see comments]

We investigated the efficacy of 2-chlorodeoxyadenosine (2-CdA) therapy in patients with mycosis fungoides (MF) and the Sezary syndrome (SS). Between February 1991 and November 1993, 21 patients with relapsed or refractory MF/SS were treated with 2-CdA. 2-CdA was administered by continuous intravenous infusion at a dose of 0.1 mg/kg/d for 7 days initially (13 patients), but was subsequently reduced to 5 days (nine patients) due to hematologic toxicity. All patients had failed to respond to at least one prior treatment for MF/SS (median number of total prior therapies, five; median number of systemic prior therapies, three) and had an Eastern Cooperative Oncology Group performance status of two or better. Cycles were administered at 28-day intervals. Assessable patients received at least 5 days of 2-CdA. Fourteen patients received more than one cycle of 2-CdA. An overall response rate of 28% was achieved. Three patients (14%) had a complete response with a median duration of 4.5 months (range, 2.5 to 16). Three (14%) had a partial response with a median duration of 2 months (range, 2 to 4). Fifteen patients (72%) had no response. The most significant toxicities encountered were bone marrow suppression (62% of patients) and infectious complications (62% of patients). Thirty-eight percent of patients experienced no toxicity from 2-CdA. 2-CdA has activity as a single agent in patients with previously treated relapsed MF/SS. Studies in less heavily pretreated individuals with 2-CdA alone or in combination will be undertaken.     

Systematic mechanism-orientated approach to chronic pancreatitis pain

Pain in chronic pancreatitis(CP) shows similarities with other visceral pain syndromes(i.e.,inflammatory bowel disease and esophagitis),which should thus be managed in a similar fashion.Typical causes of CP pain include increased intrapancreatic pressure,pancreatic inflammation and pancreatic/extrapancreatic complications.Unfortunately,CP pain continues to be a major clinical challenge.It is recognized that ongoing pain may induce altered central pain processing,e.g.,central sensitization or pro-nociceptive pain modulation.When this is present conventional pain treatment targeting the nociceptive focus,e.g.,opioid analgesia or surgical/endoscopic intervention,often fails even if technically successful.If central nervous system pain processing is altered,specific treatment targeting these changes should be instituted(e.g.,gabapentinoids,ketamine or tricyclic antidepressants).Suitable tools are now available to make altered central processing visible,including quantitative sensory testing,electroencephalograpy and(functional) magnetic resonance imaging.These techniques are potentially clinically useful diagnostic tools to analyze central pain processing and thus define optimum management approaches for pain in CP and other visceral pain syndromes.The present review proposes a systematic mechanism-orientated approach to pain management in CP based on a holistic view of the mechanisms involved.Future research should address the circumstances under which central nervous system pain processing changes in CP,and how this is influenced by ongoing nociceptive input and therapies.Thus we hope to predict which patients are at risk for developing chronic pain or not responding to therapy,leading to improved treatment of chronic pain in CP and other visceral pain disorders.