Abstract P2-07-13: OncoproMex®: An intelligent decision support system for Mexican breast cancer patients

Abstract

Abstract Background: Breast cancer in Mexico is the first cause of mortality due to malignant tumors among women. The five-year overall survival among Mexican breast cancer patients (MexBCP)treated at governmental facilities is about 75-80% as a result of an increased access to oncology treatments (WHO 20thmodel list essential medicines) in the Public Health Insurance called "Seguro Popular". Expert systems are computer programs that are derived from a branch of computer Science research called Artificial Intelligence (AI). We do not have a system based on artificial intelligence for MexBCP prognostic and predictive evaluation. The aim was develop an expert system that generates a model based on data mining techniques, which allowed predict the MexBCP survival Patients and Methods: This study was carried out by the methodology currently used in the processes of Knowledge Discovery from Databases (KDD), supported by the WEKA free distribution tool for the modeling of data mining techniques. The breast cancer data of 4,773 were provided by INCAN cohort of 4300 patients diagnosed from 2006 to 2013 with a median follow-up of 40.5 months and by INCMNSZ cohort of 473 patients from 2011 to may 2018 with a median follow-up of 39 months. The clinical and pathologic variables were: age, TNMc stage, hormonal status (pre or perimenopause or postmenopause), ER, PR, HER2, Ki67, nuclear grade. Date of histological diagnosis, date of recurrence or last medical consultation, date of death, date of death, specific cancer were used for Survival analysis. Results: The knowledge base for the expert system was based on the rules generated by the different data mining techniques. The rules used were generated by the Prism classification algorithm, which classify with a 97% percentage of instances correctly and a Kappa statistic of 0.9208. These rules obtained characteristics in each of the attributes, as well as the percentage of certainty of each of those rules. In addition to determining the average life of the group of patients that was classified in each of the generated rules. Finally, the basic elements that formed part of the architecture of the expert system carried out were the knowledge base, the inference engine, the database and the interface with the user. An on-line expert system was created, which allows users to interact and thus allow decision-making based on the results presented. Conclusion: As far as we know this is the first expert system that allows calculate prognosis according to clinical-pathological variables. It is of great relevance know the survival of a Mexican patient with breast cancer in the public health system with access to essential treatment. The applications of the system can be multiple in the usual clinical practice, education and in the taking of public policies for breast cancer in Mexico. We are currently working on a predictive model of oncological treatment benefit based also on an expert system. Citation Format: Armengol-Alonso A, Villalobos-Castaldi FM, Cabrera-Galeana P, Bargallo-Rocha E, Reynoso-Noverón N, Mohar A, Melo-Morin JP. OncoproMex®: An intelligent decision support system for Mexican breast cancer patients [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P2-07-13.