Methods for Predictor Analysis of Repeated Measurements: Application to Psychiatric Data

Summary

Objectives: In schizophrenia research, little attention yet has been directed on methods for analyzing data from studies with repeated measurements over time. Motivation for this research stems from a project within the German Research Network on Schizophrenia, in which an algorithm is developed to guide prodrome-based early intervention strategies in stable first episode patients.

Methods: We present two different approaches for the analysis of correlated response data, the Generalized Estimating Equations (GEE) method and the Artificial Neural Network (ANN) approach. We illustrate the methods using the data of the A.N.I. study, which is one of the largest German multicenter treatment studies in regard to the long-term treatment of schizophrenia conducted between 1983 and 1989.

Results: The results of statistical model selection prior to GEE analysis and various data presentation methods for ANNs are presented. The primary goal of our evaluation is to investigate if the defined prodromes are valid predictors for relapse. Additionally, it is shown that both methods are applicable on a realistic data set.

Conclusions: It is concluded that both methods are suitable for predictor analysis especially since all variable time points of the patients are included instead of only selected, so that it can be assumed that results are not biased. With the GEE method a test of association for each predictor can be performed whereas with ANNs a general proposition can be made for pro-dromes depending on the type of data presentation. Using the A.N.I. data the prodrome ‘trouble sleeping’ seems to be the most informative predictor. Finally, the important differences of the two methods are discussed.

Keywords

Generalized Estimating Equations (GEE) - Neural Networks (Computer) - predictor analysis - prodromes - schizophrenia

^* The first two authors contributed equally to this work

^** The German A.N.I. Study Group consists of: Pietzcker A, Freie Universität Berlin, Clinic for Psychiatry, Germany

^** Linden M, Klinik Seehof, Teltow, Germany Müller P, University of Göttingen, Clinic for Psychiatry, Germany Müller-Spahn F, University of Basel, Clinic for Psychiatry, Switzerland

^** Tegeler J, Park-Krankenhaus, Leipzig-Dösen, Germany

• References

• 1 Donner A, Klar N. Design and Analysis of Cluster Randomization Trials in Health Research. Arnold Publishers Limited; London: 2000
• 2 Sherman M, Le Cessie S. A comparison between bootstrap methods and generalized estimating equations for correlated outcomes in generalized linear models. Comm in Statistics 1997; 26: 901-25.
• 3 Diggle P, Heagerty P, Liang KY, Zeger SL. Longitudinal Data. Oxford Statistical Science Service Oxford University Press Inc; USA: 2002
• 4 McCulloch W, Pitts W. A logical calculus of the ideas immanent in nervous activity. Bull Math Biophys 1943; 5: 115-33.
• 5 Papadourakis G, Vourkas M, Micheloyannis S, Jervis B. Use of artificial neural networks for clinical diagnosis. Math Comput Simul 1996; 40: 623-35.
• 6 Kippenhan JS, Barker WW, Nagel J, Grady C, Duara R. Neural-network classification of normal and Alzheimer’s disease subjects using high-resolution and low-resolution PET cameras. J Nucl Med 1994; 35 (01) 7-15.
• 7 Schwarz G, Vach W, Schuhmacher M. On the misuses of artificial neural networks for prognostic and diagnostic classification in oncology. Statist Med 2000; 19: 541-61.
• 8 Birchwood M, Spencer E, McGovern D. Schizophrenia: early warning signs. Ad Psych Treat 2000; 6: 93-101.
• 9 Herz M, Melville C. Relapse in schizophrenia. Am J Psych 1980; 137: 801-12.
• 10 Jorgensen P. Early signs of psychotic relapse in schizophrenia. British Journal of Psychiatry 1998; 172: 327-30.
• 11 Gaebel W, Jänner M, Frommann N, Pietzcker A, Köpcke W, Linden M, Müller P, Müller-Spahn F, Tegler J. First vs. multiple episode schizophrenia: two-year outcome of intermittent and maintenance medication strategies. Schizo Res 2002; 53: 145-59.
• 12 Liang KY, Zeger SL. Regression analysis for correlated data. Ann Review of Public Health 1992; 14: 43-68.
• 13 Akaike J. A new look at the statistical model identification. IEEE Trans Automat Control AC 1974; 19: 716-21.
• 14 STATISTICA Neural Networks, Release 4.0 F, Copyright ^© 2000 Trajan Software Ltd, 1996-2000. Portions Copyright ^© StatSoft Inc., 1998-2000.
• 15 Mancl LA, DeRouen TA. Small sample properties of generalized estimating equation methods relevant to clinical studies of periodontal disease. Technical report, Biometry Core Technical Report No. 9402. Seattle WA: Regional Clinical Dental Research Center of the University of Washington; 1994
• 16 Ziegler A, Kastner C, Blettner M. The Generalized Estimating Equations: An Annotated Bibliography. Biometrical Journal 1998; 40 (02) 115-39.
• 17 Seuchter SA, Eisenacher M, Heinecke A, Pietzcker A, Gäbel W, Linden M, Müller P, Müller-Spahn F, Tegeler J, Köpcke W. Prodromgestützte Prädiktoranalyse in der Schizophrenieforschung. Informatik, Biometrie und Epidemiologie in Medizin und Biologie 2001; 32 2-3. 293-4.
• 18 White H. Maximum Likelihood Estimation of misspecified models. Econometrica 1982; 50: 1-25.
• 19 SAS Institute, Statistical Analysis System, release 8.2, Copyright ©, SAS Inst. Inc., 1999-2002.
• 20 Heckerling PS, Gerber BS, Tape TG, Wigton RS. Entering the Black Box of Neuronal Networks. Methods Inf Med 2003; 42: 287-96.