MPBA PREDICTIVE MODELS FOR BIOMEDICINE AND ENVIRONMENT Cesare - PowerPoint PPT Presentation

NTTS2015 – Special Session on Visualization for Data Dissemination- Bruxelles March 10 MPBA PREDICTIVE MODELS FOR BIOMEDICINE AND ENVIRONMENT Cesare Furlanello FBK-ICT / Complex Data Analytics Assembling Information: dynamic dashboards for actionable data analytics Ack: E. Arbitrio, A. Gobbi, C. Dolci, G. Jurman

MPBA: a Data Science Lab SCIENTIFIC & TECHNICAL ASSETS • Predictive Models for solving complex problems on very high-dimensionality/throughput data • Machine learning, Network Analysis Spatial data and time series, WebGIS, Data Analytics platforms Complex Data Analytics Research Line (CODA) Predictive Models (MPBA) Dynamic processes in Remote Sensing for the complex societies (DPCA) Digital Earth (RSDE) Mobile and Social Neuroinformatics Computing ( MobS ) Laboratory (NILab) C. Furlanello – MPBA Mar 2015

Outline 1. Dissemination of data within organizations Interactive Visualization Data Analytics in strategic marketing (experience) 2. Platforms for public access and strategic decisions From Graphs Statistics for Environmental Risk to Models 3. Personalized data analytics Transfer Learning e.g. Personal Proteomics & Autonomy for Nutrition/Food Quality Crowdsourcing C. Furlanello – MPBA Mar 2015

ACTIONABLE GRAPHICS FOR STRATEGIC MARKETING FLEXIBLE MULTI-SCALE PLATFORMS PRODUCT: DataExpress - Central Strategic Analysis + Internal Fidelity Programme Unilife A SINGLE INTERACTIVE DASHBOARD CONCEPT: for central strategic decision ( > 600 ML Eur/y) + associates’ network (2 -15 ML Euro / y) Predictive Models for marginality Seasonality / Geographic Analysis / Network 4

C. Furlanello – MPBA Mar 2015

Categories Covariates Interactive target function on selected samples and categories Samples Interactive list of selected samples Interactive Biclustering Interactive (from bioinformatics) WebGIS panel C. Furlanello – MPBA Mar 2015

support data exploration: • Dendrogram structures: click on nodes to select subsamples • Here: cluster on sales, target indicator is incidence C. Furlanello – MPBA Mar 2015

Integration of several solutions • Javascript: D3js, highcharts, highstock, sigma.js, heatmap.js, leaflet, inchlib, • R: RJSONIO + ad hoc scripts • PostgreSQL: PLR, PostGIS C. Furlanello – MPBA Mar 2015

From genomics: managing big taxonomies • “Sunburst” plots • Integrated with time series C. Furlanello – MPBA Mar 2015

Add-ons for taxonomies • Visualize cascades • Each node can be rooted • Absolute/Relative importance C. Furlanello – MPBA Mar 2015

Where should interaction features be added? C. Furlanello – MPBA Mar 2015

2002-2014: MPBA WEBGIS PLATFORMS UXB-Trentino: WebGIS & UXO MITRIS: High res mapping and MITRIS-FVG: High res mapping Risk Map of WWII - Unexploded surveillance of Traffic Risk (PAT) and surveillance of Traffic Risk ( Aerial Bombs (Civil Defense PAT) (Reg. Friuli Venezia Giulia) FAUNATN: Wildlife ENVIROCHANGE: Modeling impacts of CancerAtlas-TN: Oncology Risk management & deer crash Climate Change on Alpine Agriculture associated to environmental factors monitoring (research, dissemination, decision ) (Forest Dept PAT)

Interface to Economics & Territorial data: a complete integration of the asset of statistical and environmental data of the autonomous province Official • Serving the reorganization of municipalities Statistics • 200 technical users/decision makers trained • System fully transferred to PAT IET , 2008-2013

IET: Harmonization + Tools 120 thematisms URBAN PLANNING • Plans current ENVIRONMENT Metadata cartography • Reference to Geo province or INDUSTRY township levels Catalogue ENERGY SLD PROTEZIONE HUB-1: Geo CIVILE data WebGIS IET Tools for public studies and shared 67 indicators Indicators decisions • society (22) • Graphics economy (21) • environment (4) Reporting • Landscape (20) Statistical service New data HUB-2: ISTAT from user session statistical data Metadata SISVSP2012 - Rome

THE INTERSOS HR-WEBGIS A pilot Open-Source WebGIS platform developed by INTERSOS , with the technical support of FBK and its spin-off MPA Solutions, to support returnee monitoring in West Darfur (Sudan) for the United Nations High Commissioner for Refugees (UNHCR ). Similar infrastructures: Chad and Yemen • Standardization of movement information from the tables ( PHP ) 2008-2010 • Calculation of nearest border points for movt. to Chad border ( SQL ) • PostGreSQL query [ RdbiPgSQL ] • >2500 movt. in 2002-2007; total population involved: >1 M 15

WEBGIS: GAP ANALYTICS THE PLATFORM access to safe water resources by UNICEF standards The MITRIS client-server Gap analysis architecture & internal • maps engine + geostatistical • graphs indicator functions • indicators • shapefiles sp grass6 spat stat Rdbi PgSQL PL/R: integration of R Data functions within PostgreSQL/PostGIS Prototype Gap Analysis + module: compute and display space-time indicators about gaps 16

Data Analytics for Safety and Security SICURSKIWEB fro from SKI SKI RES RESCUE rep reports A web/mobile architecture for alpine ski safety First a FBK project: now funded as a regional infrastructure for 25 Ski-Areas, Ski-School association (1500) Enabled for Personal Wearable Sensors Esecurity EU-SEC project 2013-15 (wit (w ith eCrim eCrime UN UNITN TN & Qu Questura di di Tre rento) A “predictive policing” framework to model the risk of of crim crimes (re (real l and per ercei eived) in in Tren rento. Framework: WebGIS + database Support the right to safety

eSecGIS Structure Report eSecDB Criminology in GeoDB Moduli di data mining predittivo Algoritmi criminologici Predictive Policing Maps Identification of Emerging patterns eSecGIS Spatial Data Infrastructure with smart city Big Data wit ith eCr eCrim ime UN UNITN TN & Qu Questura di di Tre rento

COMPARING CRIMES AND VICTIMIZATION C. Furlanello – MPBA Mar 2015

ENVIRONMENT Breast cancer incidence, RT1999/2002 1-3 3-5 5-10 10-25 > 25 CANCER ATLAS TRENTINO Georeferenced version of Cancer Registry, in a WebGIS interface, complete of statistical functions (e.g. small area analysis) for epidemiology Assess the impact of climate change on plant and pathogens dynamic in NanoWebGIS: an open source GeoICT platform for cell agriculture at the regional scale (short-term biological, environmental image analysis based on spatial features and economic); spatialization and clustering of time series on HPC.

ENVIRO an innovative web mapping tool to monitor and forecast plant and pests dynamics based on climate data Future IPM in Europe Riva del Garda , 19 Marzo 2013 MPBA Agroecosistemi Sostenibili e Biorisorse

ENVIRO an innovative web mapping tool to monitor and forecast plant and pests dynamics based on climate data

Some example of questions ENVIRO can answer: 1. What will be the effect of climate on the phenology of the chardonnay? 2. What is the dynamics of risk powdery mildew? 3. What is the influence of climate on copper treatments in organic agriculture and how they will change? 4. How to optimize the monitoring of the territory according risk of new diseases? 5. How to optimize lab analysis in case of risk of mycotoxin contamination of grape?

Long term decisions R Model used: ( Development and evaluation of a warning model for the optimal use of copper in organic viticulture. Pellegrini et al, 2010) 1. EU normative and restrictions 2. How will it change in the future? Need for Interactive simulation

How will the use of copper change in R organic agriculture in the future? 2003 2008

R Based on climate change models: how will the use of copper change in organic agriculture in the future? 2021/2050 2071/2099

NEW FRONTIER: BIG DATA FROM PERSONAL PROTEOMICS NEAR INFRARED SPECTROMETRY (NIRS) • Pocket molecular sensor by Consumer Physics (20g) • Pledged: $2.7 M of $200K goal • DLP4500NIR TI MEMS technology also available to open use of NIR for mobile use Start: 2015 • Partner: OPBG Rome Children Hospital excellence center in clinical proteomics and metabolomics Consumer Physics, Israel SCIENTIFIC CROWD-SOURCING IN QUALITY & SAFETY VALIDATION HEALTH – WELL BEING OF NUTRITION

Engaging the new colleagues Young People • WebValley is the FBK summer school for dissemination of interdisciplinary scientific research • The WebValley formula, since 2001, is to create a team of enthusiast and motivated high school students (18 y) tutored by experienced researchers. The project activities are developed mostly working in groups, in a high-tech lab located in a small Innovative Alpine town. ~ 300 students – international from 2011. ICT tools • The team accepts a challenge by a top scientist collaborator from Ecology, Biology or Social Sciences, and develops in three weeks a new web-based prototype for data analysis and management. C. Furlanello – MPBA Feb 2015