79296 - ADVANCED ANALYSIS OF DATA

Conoscenze e abilità da conseguire

By the end of the course the student will develop advanced expertise in formulating and implementing statistical approaches to practical problems in a wide variety of subject areas. To integrate material covered in various lecture courses with skills developed through practical work in order to solve real-world problems. By the end of this course students will be able to: - formulate questions of interest and identify relevant informal and formal statistical methodology for a wide variety of practical contexts; - implement the various stages of advanced statistical analysis appropriately in R; - interpret the output of R procedures; - critically collate results and conclusions; - present the main results and conclusions in the form of concise summaries; - present results of analyses in the form of written reports; - critically assess published applications of statistical analysis; - work independently on practical data analysis problems.

Contenuti

Proposed Module by Prof.Fabrizio Carinci

fabrizio.carinci@unibo.it [mailto:fabrizio.carinci@unibo.it]

Contents of the proposed statistical module

A specific statistical module is intended to support the student towards a fundamental understanding of advanced statistical techniques used to model complex sets of relations, with specific applications to health research, planning and policy.

The module will present solutions to current challenges involving the use of large scale routine databases available at national and international level. Practical cases of data analysis will be presented using relevant statistical software.

In particular the student should be able:

• to apply advanced techniques of model building using multivariate models e.g. generalized estimating equations (GEE) and derived forms of Cox proportional hazards
• to apply the principles and conduct quantitative meta-analysis of binary and continuous outcomes
• to calculate and interpret risk-adjusted health care quality indicators used in regional, national and international reports
• to apply advanced analytical techniques e.g. regression trees, cross validation and the analysis of distributed data in a range of applications

Lectures will cover the following subjects:

• Advanced Regression Models and Meta-analysis
  • ◦Multivariate models for complex data: GEE logistic regression, time dependent covariates using the Cox Proportional hazard model.
  • ◦Meta-analysis of epidemiological studies: principles, methodology and the quantitative analysis of binary and continuous data.
  • ◦R Labs: model building strategies and meta-analysis.

• Regional, national and international health data sources
  • ◦International data sources, projects and classifications (Health status and quality of life, Surveys, Reports and Health Databases - WHO, European Commission, OECD). Standardized health care data sources in the Italian National Health System.
  • ◦Theory and applications of health systems performance assessment.
  • ◦R Labs: analysis of health care quality indicators

• Risk stratification and standardization
  • ◦Classification and regression trees: supervised/unsupervised and the generalized RECPAM model. Cross validation and bootstrapping methods.
  • ◦Standardization methods in AHRQ and OECD indicators.
  • ◦Distributed data analysis: the BIRO approach and the OECD hospital performance benchmarking method.
  • ◦R Labs: model building strategies in epidemiology.

Module proposed by Prof. Luke B Connelly

Contents of the proposed statistical module

A statistical module that introduces students to econometric formulations, their estimation and interpretation to provide students with the ability to undertake analyses of non-experimental (or “observational”) data using modern econometric methods. Specific policy-relevant applications of these techniques to a wide range of topics in economics (e.g. education, labour markets, determinants of crime, health and health care).

In particular, at the end of this module, the student should be able to:

• identify and apply modern econometric techniques to analyse complex datasets including panel (i.e., longitudinal) datasets;
• analyse continuous and limited dependent variables, including unordered and ordered categorical variables;
• characterise and identify specification and “identification” problems in applied econometric models, and to compute and interpret the coefficients and marginal effects (MEs) produced by those models
• understand and employ commonly-used econometric model selection and specification criteria; and
• to explain, in a way that is intelligible to a general audience, the results of policy-relevant applications of econometrics.

• Applied Econometrics: Review
  • Review economic approaches to the analysis of non-experimental (observational) data: linear regression models in economics; typical econometric specifications; derivation and interpretation of marginal effects (MEs); common sources of bias; goodness of fit; the “identification” problem.

    Laboratory applications of linear regression to supplied datasets using Stata.

• Econometric models with instrumental variables (IVs) limited dependent variables
  • Econometric formulations of logit and probit models and models for ordered and unordered dependent variables (ordered probit and logit; multinomial logit and probit); heteroscedasticity; specifications and data issues (including misspecification and detection of misspecification problems); model selection criteria and diagnostic testing.

    Laboratory implementation of IV and other estimation methods; estimation of limited dependent variable models and testing, using Stata.

• Econometric panel (longitudinal) data models
  • Analysis of data with cross-section (x) and time-series (t) elements; unobserved heterogeneity; fixed- and random-effects panel data regression; extensions (e.g., count-data econometric models with panel datasets); diagnostic testing and model comparison.

Testi/Bibliografia

Prof Carinci

Main References

• Kleinbaum DG, M Klein, Survival analysis. A self learning Text. 3^rd Edition. Springer 2012.

Selected sections

• Kleinbaum DG, Logistic Regression. A self learning text. 3^rd Edition, Springer, 2010.
• OECD, Health at a glance 2015, Paris 2015. Available at: http://www.oecd.org/health/health-systems/health-at-a-glance-19991312.htm

Papers (provisional list)

• Carinci F et al., Towards actionable international comparisons of health system performance: expert revision of the OECD framework and quality indicators, International Journal for Quality in Health Care, Apr; 27(2):137-46, 2015. Available at: http://intqhc.oxfordjournals.org/content/27/2/137.long .
• Nicolucci A, Carinci F, Ciampi A, Stratifying Patients at Risk of Diabetes Complications: An integrated look at clinical, socio-economic and care-related factors, Diabetes Care, 21 (9):1439-1444, 1998.
• Carinci F, Nicolucci A, Ciampi A, Labbrozzi D, Bettinardi O, Zotti AM, Tognoni G on behalf of the GISSI Investigators, Role of Interactions between Psychological and Clinical Factors in determining 6-Month Mortality among patients with Acute Myocardial Infarction. Application of recursive partitioning techniques to the GISSI-2 Data-Base, European Heart Journal, 18:835-845, 1997.
• Carinci F et al. Lower extremity amputation rates in people with diabetes as an indicator of health systems performance. A critical appraisal of the data collection 2000-2011 by the Organization for Economic Cooperation and Development (OECD), Acta Diabetologica, 2016 Oct;53(5):825-32. Available at: http://link.springer.com/article/10.1007%2Fs00592-016-0879-4 .

Prof Connelly:

Main References

Wooldridge JM (2016) Introductory Econometrics, South-Western Cengage Learning: Mason (OH).

Selected Sections

Cameron ACC and Trivedi P (2009) Microeconometrics Using Stata, Stata Press: New York.

Further reading (papers and reports)

To be advised.

 

 

Metodi didattici

Lectures with open discussion + Computer labs + Selected readings

Modalità di verifica e valutazione dell'apprendimento

The exam will evaluate the achievements of the student through a direct assessment of the following learning outcomes:

• ability to identify the most suitable methods required to carry out the analysis and interpretation of health-related data
• ability to conduct a critical appraisal of the methodology adopted in a research project or scientific publication

The exam will consist of an in depth oral assessment concerning the ability to understand a problem and conduct an advanced statistical analysis in any of the fields of interest presented in the course.

The student should be able to respond to questions regarding the choice of methods in different situations, with specific application in health research (and other empirical problems), including aspects related to data acquisition.

 

Strumenti a supporto della didattica

Textbook, further readings, lectures, datasets, laboratory exercises, class discussion.

Orario di ricevimento

Consulta il sito web di Luke Brian Connelly

Consulta il sito web di Fabrizio Carinci