Publicação
Profiling and relapse prediction of breast cancer patients
| Resumo: | Cancer is the world’s second leading cause of death, summing up an estimated 10 million deaths in 2020. Breast cancer is the fifth-highest death cause among all cancer types, being the first in incidence and prevalence. Several predictions and statistics suggest that worldwide incidences of breast cancer are rising, opening doors to developing decision support systems that may help clinicians improve their patients’ prognoses. This thesis focuses on profiling and characterising breast cancer patients from the Hospital Universitario Puerta de Hierro Majadahonda (HUPHM), estimating the survival probabilities regarding several features using the Kaplan-Meier estimator, assessing the survival outcome between patients with different treatments, prognoses, age distribution, and others. The whole cohort of patients was subject to a complete descriptive analysis reported to the associated clinicians. Throughout the dissertation, the raw dataset provided by the partner Hospital was preprocessed, transformed and stored in a database, allowing clinicians to access their data through a previously built dashboard. Moreover, a tool that assists physicians in predicting the occurrence of relapse was explored since an early diagnosis of the reap- pearance of cancer can be crucial to the patient’s prognosis. The results obtained through the implementation of the methodology proposed re- garding the relapse prediction conclude that the dataset is statistically irrelevant, proba- bly due to the vital information loss in the middle age group (the dataset provided only contained patients until 45 and more than 65 years old). Nevertheless, the descriptive and survival analysis performed in the cohort of patients available is highly relevant to the clinical practice since it is the first time these patients are being described and analysed. |
|---|---|
| Autores principais: | Vieira, Bruno André Rebelo |
| Assunto: | Breast Cancer Data Engineering Survival Analysis Kaplan-Meier Relapse Logistic Regression |
| Ano: | 2022 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade Nova de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório Institucional da UNL |
| Resumo: | Cancer is the world’s second leading cause of death, summing up an estimated 10 million deaths in 2020. Breast cancer is the fifth-highest death cause among all cancer types, being the first in incidence and prevalence. Several predictions and statistics suggest that worldwide incidences of breast cancer are rising, opening doors to developing decision support systems that may help clinicians improve their patients’ prognoses. This thesis focuses on profiling and characterising breast cancer patients from the Hospital Universitario Puerta de Hierro Majadahonda (HUPHM), estimating the survival probabilities regarding several features using the Kaplan-Meier estimator, assessing the survival outcome between patients with different treatments, prognoses, age distribution, and others. The whole cohort of patients was subject to a complete descriptive analysis reported to the associated clinicians. Throughout the dissertation, the raw dataset provided by the partner Hospital was preprocessed, transformed and stored in a database, allowing clinicians to access their data through a previously built dashboard. Moreover, a tool that assists physicians in predicting the occurrence of relapse was explored since an early diagnosis of the reap- pearance of cancer can be crucial to the patient’s prognosis. The results obtained through the implementation of the methodology proposed re- garding the relapse prediction conclude that the dataset is statistically irrelevant, proba- bly due to the vital information loss in the middle age group (the dataset provided only contained patients until 45 and more than 65 years old). Nevertheless, the descriptive and survival analysis performed in the cohort of patients available is highly relevant to the clinical practice since it is the first time these patients are being described and analysed. |
|---|