A distributed approach to enabling privacy-preserving model-based classifier training

This paper proposes a novel approach for privacy-preserving distributed model-based classifier training. Our approach is an important step towards supporting customizable privacy modeling and protection. It consists of three major steps. First, each data site independently learns a weak concept model (i.e., local classifier) for a given data pattern or concept by using its own training samples. An adaptive EM algorithm is proposed to select the model structure and estimate the model parameters simultaneously. The second step deals with combined classifier training by integrating the weak concept models that are shared from multiple data sites. To reduce the data transmission costs and the potential privacy breaches, only the weak concept models are sent to the central site and synthetic samples are directly generated from these shared weak concept models at the central site. Both the shared weak concept models and the synthetic samples are then incorporated to learn a reliable and complete global concept model. A computational approach is developed to automatically achieve a good trade off between the privacy disclosure risk, the sharing benefit and the data utility. The third step deals with validating the combined classifier by distributing the global concept model to all these data sites in the collaboration network while at the same time limiting the potential privacy breaches. Our approach has been validated through extensive experiments carried out on four UCI machine learning data sets and two image data sets.

The ObReco-360°: a new ecological tool to memory assessment using 360° immersive technology

Virtual Reality - One important feature of a neuropsychological test is its ecological validity, which defines how much patients’ test scores are linked to real-life functioning. However,...     

A rapid interview protocol supporting patient-centered quality improvement: hearing the parent's voice in a pediatric cancer unit

BACKGROUND: The Institute of Medicine's 2001 report on quality delimits six dimensions of optimal care: safety, effectiveness, efficiency, timeliness, patient centeredness, and equity. In fall 2001 parents of pediatric cancer patients were interviewed to determine how well they thought these dimensions were addressed with respect to medication administration. Immediate goals were to identify system weaknesses and devise strategies to prevent future errors. A higher-order goal was to develop and demonstrate a model protocol for rapid-cycle interview assessments. METHODS: Hematology/oncology directors worked with a research expert to develop a semistructured interview protocol. After training, which included directed reading, oral instruction, and role-playing, a convenience sample of 20 English- and Spanish-speaking parents of inpatients was recruited. Parents were asked to characterize current medication administration practices and to describe problems that they had experienced or witnessed. Rapid content analysis techniques were used to identify issues of importance to the parents. FINDINGS: Parents' medication concerns centered on their children's comfort. Parents called for communication improvements, standardization of all nursing procedures and techniques, and a guide or an outline providing a clear understanding of what to expect when and from whom. Viewing these concerns in relation to the Institute of Medicine's quality domains allowed the department to frame an improvement action plan aligned with organizational and national priorities. IMPLICATIONS: With good supervision and limited focused training, inexperienced staff can successfully administer semistructured qualitative interviews and help analyze findings for rapid cycle improvement purposes. The protocol can be adapted for use in organizations interested in rapid qualitative assessments of patient and parent preferences.     

Impact assessment of odours emitted by a wastewater treatment plant

Complaints from the Domingos Martins population about sewage odours in the city made the district attorney order an impact assessment of the odours emitted by the city wastewater treatment plant (WWTP). This study comprised various techniques, models and population surveys. In 2007, an odour emission model proved that the main hydrogen sulphide emitter was the aeration tank of the WWTP (13.5 g h(-1)) and such emissions, according to CALPUFF model, should be perceived in the whole Domingos Martins city centre area. In this area, 58% of those interviewed were annoyed by the WWTP odours. However, in 2009, the odour monitoring panel recorded few odour occurrences. A second population survey showed that hereafter only 20% of those interviewed were annoyed by the WWTP emissions. Odour emission and dispersion models run with 2010 data proved a drastic reduction of the WWTP aeration tank emissions and consequently the city centre was not bothered by WWTP emissions anymore. The odour emission reduction was due to the modification of the WWTP aeration tank system. Despite the odour emission reduction, houses located southeast of the WWTP were still annoyed by sewage odours. However, in this part of the town, other sources of sewage odours have been found.     

Toll road crashes of commercial and passenger motor vehicles

Revenue-collection data from toll roads allow for accurate estimates of miles driven by vehicle type and, when combined with crash data, valid estimates of crash involvements per mile driven. Data on vehicle-miles traveled and collisions were obtained from toll road authorities in Florida. Kansas, and New York. In addition, state crash files and published vehicle-miles of travel were obtained for toll roads in Illinois. Indiana, Ohio, and Pennsylvania. Large commercial motor vehicles were significantly underinvolved in single-vehicle crashes on all state toll roads. In five states, commercial motor vehicles were significantly overinvolved in multiple-vehicle crashes relative to passenger vehicles; the exceptions were Kansas, where they had significantly lower multiple-vehicle involvement rates, and Indiana. where there were no significant differences in multiple-vehicle involvements by vehicle type. The risk of commercial motor vehicle involvement in multiple-vehicle crashes resulting in deaths or serious injuries was double that of passenger vehicles in the two states (Ohio and Pennsylvania) that identified serious injuries. Whether crash rates, on toll roads of commercial motor vehicles are higher or lower than those of passenger vehicles appears to depend on the type of crash, specific toll road. and traffic density.     

Simultaneous determination of bisphenol A,octylphenol, and nonylphenol by pressurised liquid extraction and liquid chromatography–tandem mass spectrometry in powdered milk and infant formulas

A new analytical method, using pressurised liquid extraction (PLE) and liquid chromatography–tandem mass spectrometry (LC–MS/MS), was developed for the simultaneous determination of bisphenol A (BPA), octylphenol (OP) and nonylphenol (NP) in powdered infant formulas (IF) and powdered skimmed milk (PM). The analytes were extracted by PLE, using this optimised conditions: ethyl acetate as solvent, 70 °C of temperature, reversed-phase silica C₁₈ as dispersing agent and three cycles of extraction. The extracts were then injected in LC–MS/MS using a Gemini C₁₈ column and a mixture of 5% water and 95% methanol/acetonitrile, both with 0.1% ammonia, as a mobile phase. Recoveries at different fortification levels (0.5 and 0.05 mg kg⁻¹), were between 89% and 92% for BPA, 84 and 98% for OP, and 93% and 101% for NP. The method was applied to the analysis of samples of PM and IF, bought in Italian and Spanish markets. In positive samples, phenols concentration ranged from 0.07 to 1.29 mg kg⁻¹ for BPA, from 0.028 to 1.55 mg kg⁻¹ for OP and from 0.026 to 1.47 mg kg⁻¹ for NP.     

Intratumoral Thermal Reading During Photo‐Thermal Therapy by Multifunctional Fluorescent Nanoparticles

The tremendous development of nanotechnology is bringing us closer to the dream of clinical application of nanoparticles in photothermal therapies of tumors. This requires the use of specific nanoparticles that must be highly biocompatible, efficient light‐to‐heat converters and fluorescent markers. Temperature reading by the heating nanoparticles during therapy appears of paramount importance to keep at a minimum the collateral damage that could arise from undesirable excessive heating. In this work, this thermally controlled therapy is possible by using Nd³⁺ ion‐doped LaF₃ nanocrystals. Because of the particular optical features of Nd³⁺ ions at high doping concentrations, these nanoparticles are capable of in vivo photothermal heating, fluorescent tumor localization and intratumoral thermal sensing. The successful photothermal therapy experiments here presented highlight the importance of controlling therapy parameters based on intratumoral temperature measurements instead of on the traditionally used skin temperature measurements. In fact, significant differences between intratumoral and skin temperatures do exist and could lead to the appearance of excessive collateral damage. These results open a new avenue for the real application of nano­particle‐based photothermal therapy at clinical level.