Arthur de Sá Ferreira1, Antonio Guilherme Pacheco2. 1. Postgraduation Program in Rehabilitation Science, Augusto Motta University Center, Rio de Janeiro, Brazil; E-mail: arthur_sf@ig.com.br. 2. Program of Scientific Computing, Fiocruz, Rio de Janeiro, Brazil.
Abstract
OBJECTIVE: The aim of this work is to develop and implement the SimTCM, an advanced computational model that incorporates relevant aspects from traditional Chinese medicine (TCM) theory as well as advanced statistical and epidemiological techniques for simulation and analysis of human patients. METHODS: SimTCM presents five major attributes for simulation: representation of true and false profiles for any single pattern; variable count of manifestations for each manifestation profile; empirical distributions of patterns and manifestations in a disease-specific population; incorporation of uncertainty in clinical data; and the combination of the four examinations. The proposed model is strengthened by following international standards for reporting diagnostic accuracy studies, and incorporates these standards in its treatment of study population, sample size calculation, data collection of manifestation profiles, exclusion criteria and missing data handling, reference standards, randomization and blinding, and test reproducibility. RESULTS: Simulations using data from patients diagnosed with hypertension and post-stroke sensory-motor impairments yielded no significant differences between expected and simulated frequencies of patterns (P=0.22 or higher). Time for convergence of simulations varied from 9.90 s (9.80, 10.27) to 28.31 s (26.33, 29.52). The ratio iteration profile necessary for convergence varied between 1:1 and 5:1. CONCLUSION: This model is directly connected to forthcoming models in a large project to design and implement the SuiteTCM: ProntTCM, SciTCM, DiagTCM, StudentTCM, ResearchTCM, HerbsTCM, AcuTCM, and DataTCM. It is expected that the continuity of the SuiteTCM project enhances the evidence-based practice of Chinese medicine. The software is freely available for download at: http://suitetcm.unisuam.edu.br.
OBJECTIVE: The aim of this work is to develop and implement the SimTCM, an advanced computational model that incorporates relevant aspects from traditional Chinese medicine (TCM) theory as well as advanced statistical and epidemiological techniques for simulation and analysis of humanpatients. METHODS: SimTCM presents five major attributes for simulation: representation of true and false profiles for any single pattern; variable count of manifestations for each manifestation profile; empirical distributions of patterns and manifestations in a disease-specific population; incorporation of uncertainty in clinical data; and the combination of the four examinations. The proposed model is strengthened by following international standards for reporting diagnostic accuracy studies, and incorporates these standards in its treatment of study population, sample size calculation, data collection of manifestation profiles, exclusion criteria and missing data handling, reference standards, randomization and blinding, and test reproducibility. RESULTS: Simulations using data from patients diagnosed with hypertension and post-stroke sensory-motor impairments yielded no significant differences between expected and simulated frequencies of patterns (P=0.22 or higher). Time for convergence of simulations varied from 9.90 s (9.80, 10.27) to 28.31 s (26.33, 29.52). The ratio iteration profile necessary for convergence varied between 1:1 and 5:1. CONCLUSION: This model is directly connected to forthcoming models in a large project to design and implement the SuiteTCM: ProntTCM, SciTCM, DiagTCM, StudentTCM, ResearchTCM, HerbsTCM, AcuTCM, and DataTCM. It is expected that the continuity of the SuiteTCM project enhances the evidence-based practice of Chinese medicine. The software is freely available for download at: http://suitetcm.unisuam.edu.br.
Authors: Ingrid Jardim de Azeredo Souza Oliveira; Arthur de Sá Ferreira Journal: Evid Based Complement Alternat Med Date: 2015-04-07 Impact factor: 2.629