ADA Data Solutions Co-fouder - Cientista de Dados
Jun. 2022Recife, Pernambuco, BrazilProject 1: Support for Diagnosis of Leprosy and Skin Cancer
This project aims to develop an intelligent tool for the early detection of these diseases. The process includes medical consultations, photographic recording of the lesion, application of computational intelligence techniques, feature extraction, and image classification into one of seven classes. Utilizing the Dermatological Atlas, the project achieved remarkable performance with an accuracy of 89.0%, a Kappa index of 88.0%, sensitivity of 73.0%, specificity of 98.0%, and an AUC-ROC of 95.0%. Project 2: Breast Cancer Diagnosis via Imaging
This project aims to develop intelligent systems for non-invasive breast cancer diagnosis using mammography and thermography images. The system classifies early-stage breast lesions, achieving high accuracy. The current performance shows an accuracy of 99.0%, a Kappa index of 95.0%, sensitivity of 98.87%, and specificity of 91.14% for thermography. For mammography, the accuracy is 95.0%, with a Kappa index of 95.0%, sensitivity of 98.0%, and specificity of 82.24%. Project 3: Epidemiological Surveillance of Arboviruses
The surveillance system addresses cases of dengue, chikungunya fever, and Zika, utilizing data on arboviruses, breeding sites, temperature, and wind speed. Notable contributions include distribution maps and predictive models. The performance of case and breeding site prediction models is robust, with Pearson correlation coefficients above 0.9875 and Root Relative Squared Error ranging from 3.29% to 14.60%. Project 4: Epidemiological Surveillance of COVID-19
In this project, the models achieved high Pearson correlation coefficients (above 0.9990), with Root Relative Squared Error ranging from 2.20% to 11.42%. In the state of Pernambuco, the performance was particularly notable, with a Pearson correlation coefficient of 99.91% and a Root Relative Squared Error of 1.92%. The project resulted in six scientific publications.