Abstract
In the artificial neural network field, no universal algorithm of modeling ensures obtaining the best possible model for a given task. Researchers frequently regard artificial neural networks with suspicion caused by the lack of repeatability of single experiments. We propose a systematic approach that may increase the probability of finding the optimal network architecture. In the experiments, the average effectiveness in groups of networks rather than single networks should be compared. Such an approach facilitates the analysis of the results caused by changes in the network parameters, while the influence of chance effects becomes negligible. As an example of this protocol, we present optimization of a neural network applied for prediction of persistent facial pain in patients operated for chronic rhinosinusitis. In the stepwise approach, the percentage of correct predictions was gradually increased from 54% to 75% for the external validation set.
Author contributions: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.
Research funding: None declared.
Employment or leadership: None declared.
Honorarium: None declared.
Competing interests: The funding organization(s) played no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the report for publication.
References
1. Szaleniec M. Prediction of enzyme activity with neural network models based on electronic and geometrical features of substrates. Pharmacol Rep 2012;64:761–81.10.1016/S1734-1140(12)70873-3Search in Google Scholar
2. Tadeusiewicz R, Chaki R, Chaki N. Exploring neural networks with C#. Boca Raton: CRC Press, Taylor & Francis Group, 2014.Search in Google Scholar
3. Baxt WG. Application of artificial neural networks to clinical medicine. Lancet 1995;346:1135.10.1016/S0140-6736(95)91804-3Search in Google Scholar
4. Szaleniec J, Skladzien J, Tadeusiewicz R, Oles K, Konior M, Przeklasa R. How can an otolaryngologist benefit from artificial neural networks? Otolaryngol Pol 2012;66:241–8.10.1016/j.otpol.2012.06.015Search in Google Scholar
5. Szaleniec J, Wiatr M, Szaleniec M, Skladzien J, Tomik J, Oles K, et al. Artificial neural network modelling of the results of tympanoplasty in chronic suppurative otitis media patients. Comput Biol Med 2013;43:16–22.10.1016/j.compbiomed.2012.10.003Search in Google Scholar
6. Tadeusiewicz R. Neural networks as a tool for modeling of biological systems. Bio-Algorithm Med-Syst 2015;11:135.10.1515/bams-2015-0021Search in Google Scholar
7. Waligorski P, Szaleniec M. Prediction of white cabbage (Brassica oleracea var. capitata) self-incompatibility based on neural network and discriminant analysis of complex electrophoretic patterns. Comput Biol Chem 2010;34:115–21.10.1016/j.compbiolchem.2010.03.002Search in Google Scholar
8. Fokkens WJ, Lund VJ, Mullol J, Bachert C, Alobid I, Baroody F, et al. European position paper on rhinosinusitis and nasal polyps 2012. Rhinol Suppl 2012;3 p preceding table of contents, 1–298.Search in Google Scholar
9. Daudia AT, Jones NS. Facial migraine in a rhinological setting. Clin Otolaryngol Allied Sci 2002;27:521–5.10.1046/j.1365-2273.2002.00628.xSearch in Google Scholar
10. Jones NS, Cooney TR. Facial pain and sinonasal surgery. Rhinology 2003;41:193–200.Search in Google Scholar
11. Tarabichi M. Characteristics of sinus-related pain. Otolaryngol Head Neck Surg 2000;122:842–7.10.1016/S0194-5998(00)70011-8Search in Google Scholar
12. Soler ZM, Mace J, Smith TL. Symptom-based presentation of chronic rhinosinusitis and symptom-specific outcomes after endoscopic sinus surgery. Am J Rhinol 2008;22:297–301.10.2500/ajr.2008.22.3172Search in Google Scholar PubMed
13. Deal RT, Kountakis SE. Significance of nasal polyps in chronic rhinosinusitis: symptoms and surgical outcomes. Laryngoscope 2004;114:1932–5.10.1097/01.mlg.0000147922.12228.1fSearch in Google Scholar PubMed
14. Bugten V, Nordgard S, Romundstad P, Steinsvag S. Chronic rhinosinusitis and nasal polyposis; indicia of heterogeneity. Rhinology 2008;46:40–4.Search in Google Scholar
15. Banerji A, Piccirillo JF, Thawley SE, Levitt RG, Schechtman KB, Kramper MA, et al. Chronic rhinosinusitis patients with polyps or polypoid mucosa have a greater burden of illness. Am J Rhinol 2007;21:19–26.10.2500/ajr.2007.21.2979Search in Google Scholar PubMed
16. de Loos DA, Hopkins C, Fokkens WJ. Symptoms in chronic rhinosinusitis with and without nasal polyps. Laryngoscope 2013;123:57–63.10.1002/lary.23671Search in Google Scholar PubMed
17. Poetker DM, Mendolia-Loffredo S, Smith TL. Outcomes of endoscopic sinus surgery for chronic rhinosinusitis associated with sinonasal polyposis. Am J Rhinol 2007;21:84–8.10.2500/ajr.2007.21.2978Search in Google Scholar PubMed
18. DelGaudio JM, Wise SK, Wise JC. Association of radiological evidence of frontal sinus disease with the presence of frontal pain. Am J Rhinol 2005;19:167–73.10.1177/194589240501900209Search in Google Scholar
19. Krzeski A, Gromek I. Zapalenia zatok przynosowych. Gdańsk: Via Medica, 2008.Search in Google Scholar
20. Marks SC, Shamsa F. Evaluation of prognostic factors in endoscopic sinus surgery. Am J Rhinol 1997;11:187–91.10.2500/105065897781751947Search in Google Scholar PubMed
21. Busaba NY, Sin HJ, Salman SD. Impact of gender on clinical presentation of chronic rhinosinusitis with and without polyposis. J Laryngol Otol 2008;122:1180–4.10.1017/S0022215107001302Search in Google Scholar PubMed
22. DiBaise JK, Olusola BF, Huerter JV, Quigley EM. Role of GERD in chronic resistant sinusitis: a prospective, open label, pilot trial. Am J Gastroenterol 2002;97:843–50.10.1111/j.1572-0241.2002.05598.xSearch in Google Scholar PubMed
23. StatSoft Inc. STATISTICA (data analysis software system), version 7.1. www.statsoft.com, 2005.Search in Google Scholar
24. Rogers D, Hopfinger AJ. Application of genetic function approximation to quantitative structure-activity relationships and quantitative structure-property relationships. J Chem Inf Comput Sci 1994;34:854–66.10.1021/ci00020a020Search in Google Scholar
25. So S-S, Karplus M. Three-dimensional quantitative structure-activity relationships from molecular similarity matrices and genetic neural networks: 1. Method and validations. J Med Chem 1997;40:4347–59.10.1021/jm970487vSearch in Google Scholar PubMed
26. So S-S, Karplus M. Genetic neural networks for quantitative structure-activity relationships: improvements and application of benzodiazepine affinity for benzodiazepine/GABAA receptors. J Med Chem 1996;39:5246–56.10.1021/jm960536oSearch in Google Scholar PubMed
27. So S-S, Karplus M. Evolutionary optimization in quantitative structure-activity relationship: an application of genetic neural networks. J Med Chem 1996;39:1521–30.10.1021/jm9507035Search in Google Scholar PubMed
28. Andrea TA, Kalayeh H. Applications of neural networks in quantitative structure-activity relationships of dihydrofolate reductase inhibitors. J Med Chem 1991;34:2824–36.10.1021/jm00113a022Search in Google Scholar PubMed
29. Burden FR. Robust QSAR models using Bayesian regularized neural networks. J Med Chem 1999;42:3183–7.10.1021/jm980697nSearch in Google Scholar PubMed
30. Gonzalez MP, Caballero J, Tundidor-Camba A, Helguera AM, Fernandez M. Modeling of farnesyltransferase inhibition by some thiol and non-thiol peptidomimetic inhibitors using genetic neural networks and RDF approaches. Bioorg Med Chem 2006;14:200–13.10.1016/j.bmc.2005.08.009Search in Google Scholar PubMed
©2016 by De Gruyter
