Distributed learning on 20 000+ lung cancer patients - The Personal Health Train.

BACKGROUND AND PURPOSE: Access to healthcare data is indispensable for scientific progress and innovation. Sharing healthcare data is time-consuming and notoriously difficult due to privacy and regulatory concerns. The Personal Health Train (PHT) provides a privacy-by-design infrastructure connecting FAIR (Findable, Accessible, Interoperable, Reusable) data sources and allows distributed data analysis and machine learning. Patient data never leaves a healthcare institute. MATERIALS AND METHODS: Lung cancer patient-specific databases (tumor staging and post-treatment survival information) of oncology departments were translated according to a FAIR data model and stored locally in a graph database. Software was installed locally to enable deployment of distributed machine learning algorithms via a central server. Algorithms (MATLAB, code and documentation publicly available) are patient privacy-preserving as only summary statistics and regression coefficients are exchanged with the central server. A logistic regression model to predict post-treatment two-year survival was trained and evaluated by receiver operating characteristic curves (ROC), root mean square prediction error (RMSE) and calibration plots. RESULTS: In 4 months, we connected databases with 23 203 patient cases across 8 healthcare institutes in 5 countries (Amsterdam, Cardiff, Maastricht, Manchester, Nijmegen, Rome, Rotterdam, Shanghai) using the PHT. Summary statistics were computed across databases. A distributed logistic regression model predicting post-treatment two-year survival was trained on 14 810 patients treated between 1978 and 2011 and validated on 8 393 patients treated between 2012 and 2015. CONCLUSION: The PHT infrastructure demonstrably overcomes patient privacy barriers to healthcare data sharing and enables fast data analyses across multiple institutes from different countries with different regulatory regimens. This infrastructure promotes global evidence-based medicine while prioritizing patient privacy.

Predicting adsorptive removal of chlorophenol from aqueous solution using artificial intelligence based modeling approaches.

The research aims to develop artificial intelligence (AI)-based model to predict the adsorptive removal of 2-chlorophenol (CP) in aqueous solution by coconut shell carbon (CSC) using four operational variables (pH of solution, adsorbate concentration, temperature, and contact time), and to investigate their effects on the adsorption process. Accordingly, based on a factorial design, 640 batch experiments were conducted. Nonlinearities in experimental data were checked using Brock-Dechert-Scheimkman (BDS) statistics. Five nonlinear models were constructed to predict the adsorptive removal of CP in aqueous solution by CSC using four variables as input. Performances of the constructed models were evaluated and compared using statistical criteria. BDS statistics revealed strong nonlinearity in experimental data. Performance of all the models constructed here was satisfactory. Radial basis function network (RBFN) and multilayer perceptron network (MLPN) models performed better than generalized regression neural network, support vector machines, and gene expression programming models. Sensitivity analysis revealed that the contact time had highest effect on adsorption followed by the solution pH, temperature, and CP concentration. The study concluded that all the models constructed here were capable of capturing the nonlinearity in data. A better generalization and predictive performance of RBFN and MLPN models suggested that these can be used to predict the adsorption of CP in aqueous solution using CSC.

Levels and distribution of persistent organochlorine pesticide residues in water and sediments of Gomti River (India)--a tributary of the Ganges River.

This study reports the concentration levels and distribution pattern of the persistent organochlorine pesticide (OCPs) residues in the water and bed-sediments of the Gomti River collected seasonally over a period of 2 years. The water and bed-sediment samples were collected from eight different sites and analyzed for aldrin, dieldrin, endrin, HCB, HCH isomers, DDT isomers/metabolites, endosulfan isomers (alpha and beta), endosulfan sulfate, heptachlor and its metabolites, alpha-chlordane, gamma-chlordane and methoxychlor. In the river water and sediments SigmaOCPs residues ranged between 2.16 and 567.49 ng l(-1) and 0.92 and 813.59 ng g(-1), respectively. The results, further, suggested that source of DDT contamination is from the aged and weathered agricultural soils with signature of recently used DDT in the river catchments. To assess any adverse effect of OCPs contamination on river's biological component, the threshold effect level (TEL) was used. The results revealed that bed-sediments of the Gomti River are contaminated with lindane, endrin, heptachlor epoxides and DDT, particularly at site-4 and may contribute to sediment toxicity in the freshwater ecosystem of the river.

Distribution of polycyclic aromatic hydrocarbons in edible fish from Gomti river, India.

This study reports the levels and distribution patterns of selected polycyclic aromatic hydrocarbons (PAHs) in fish samples of the Gomti river, India, collected from three sites during the pre- and post-monsoon seasons of the years 2004-2005. In the fish muscles, SigmaPAHs ranged between 12.85 and 34.89 ng g(-1) wet wt (mean value: 23.98+/-6.70 ng g(-1)). Naphthalene was the most prevalent compound both in terms of detection as well as levels, while, benzo[k]fluoranthene, benzo(a)pyrene, and indeno(123-cd)pyrene + benzo(ghi)perylene could not be detected in any of the sample. Low-molecular weight PAHs were observed dominating over the high molecular weight PAHs.

Liquid-phase adsorption of phenols using activated carbons derived from agricultural waste material.

Physical and chemical properties of activated carbons prepared from coconut shells (SAC and ATSAC) were studied. The adsorption equilibria and kinetics of phenol and 2,4-dichlorophenol from aqueous solutions on such carbons were then examined at three different temperatures (10, 25 and 40 degrees C). Adsorption of both phenol and 2,4-dichlorophenol increased with an increase in temperature. The experimental data were analyzed using the Langmuir and Freundlich isotherm models. Both the isotherm models adequately fit the adsorption data for both the phenols. The carbon developed through the acid treatment of coconut shells (ATSAC) exhibited relatively higher monolayer adsorption capacity for phenol (0.53 mmol g(-1)) and 2,4-dichlorophenol (0.31 mmol g(-1)) as compared to that developed by thermal activation (SAC) with adsorption capacity of 0.36 and 0.20 mmol g(-1), for phenol and 2,4-dichlorophenol, respectively. The equilibrium sorption and kinetics model parameters and thermodynamic functions were estimated and discussed. The thermodynamic parameters (free energy, enthalpy and entropy changes) exhibited the feasibility and spontaneous nature of the adsorption process. The sorption kinetics was studied using the pseudo-first-order and second-order kinetics models. The adsorption kinetics data for both the phenol and 2,4-dichlorophenol fitted better to the second-order model. An attempt was also made to identify the rate-limiting step involved in the adsorption process. Results of mass transfer analysis suggested the endothermic nature of the reaction and change in the mechanism with time and initial concentration of the adsorbate. The results of the study show that the activated carbons derived from coconut shells can be used as potential adsorbent for phenols in water/wastewater.

Vertical characterization of soil contamination using multi-way modeling--a case study.

This study describes application of chemometric multi-way modeling approach to analyze the dataset pertaining to soils of industrial area with a view to assess the soil/sub-soil contamination, accumulation pathways and mobility of contaminants in the soil profiles. The three-way (sampling depths, chemical variables, sampling sites) dataset on heavy metals in soil samples collected from three different sites in an industrial area, up to a depth of 60 m each was analyzed using three-way Tucker3 model validated for stability and goodness of fit. A two component Tucker3 model, explaining 66.6% of data variance, allowed interpretation of the data information in all the three modes. The interpretation of core elements revealing interactions among the components of different modes (depth, variables, sites) allowed inferring more realistic information about the contamination pattern of soils both along the horizontal and vertical coordinates, contamination pathways, and mobility of contaminants through soil profiles, as compared to the traditional data analysis techniques. It concluded that soils at site-1 and site-2 are relatively more contaminated with heavy metals of both the natural as well as anthropogenic origins, as compared to the soil of site-3. Moreover, the accumulation pathways of metals for upper shallow layers and deeper layers of soils in the area were differentiated. The information generated would be helpful in developing strategies for remediation of the contaminated soils for reducing the subsequent risk of ground-water contamination in the study region.

Residues of organochlorine pesticides in fish from the Gomti river, India.

This study reports the levels and distribution patterns of some organochlorine pesticides (OCPs) in fish samples of the Gomti river, India, collected from three sites. In the fish muscles Sigma OCPs ranged between 2.58-22.56 ng g(-1) (mean value: 9.66 +/- 5.60 ng g(-1)). Neither spatial nor temporal trends could be observed in distribution of the OCPs. Aldrin was the predominant OCP, whereas, HCB and methoxychlor could not be detected. alpha-HCH and beta-HCH among the isomers of HCH and pp-DDE among the metabolites of DDT were the most frequently detected OCPs. The results revealed that the fish of the Gomti river are contaminated with various OCPs.

Influence of physico-chemical factors on leaching of chemical additives from aluminium foils used for packaging of food materials.

In recent years, the use of aluminium foils to wrap foodstuff and commodities has been increased to a great extent. Aluminium was found to leach out from the foil in different simulants particularly in distilled water, acidic and alkaline medium at 60 +/- 2 degrees C for 2 hours and 40 +/- 2 degrees C for 24 hours. The migration was found to be above the permissible limit as laid down by WHO guidelines, that is of 0.2 mg/L of water. The protocol used for this study was based on the recommendation of Bureau of Indian Standard regarding the migration of chemical additives from packaging materials used to pack food items. Migration of the aluminium metal was found significantly higher in acidic and aqueous medium in comparison to alcoholic and saline medium. Higher temperature conditions also enhanced the rate of migration of aluminium in acidic and aqueous medium. Leaching of aluminium metal occurred in double distilled water, acetic acid 3%, normal saline and sodium carbonate, except ethanol 8%, in which aluminium migration was below the detection limit of the instrument where three brands of the aluminium foil samples studied.