Cause-specific mortality prediction with machine learning on a longitudinal sample of 502,632 individuals

• Grant number: 17/09369-8
• Support Opportunities: Regular Research Grants
• Start date: February 01, 2018
• End date: July 31, 2020
• Field of knowledge: Health Sciences - Collective Health - Epidemiology
• Principal Investigator: Alexandre Dias Porto Chiavegatto Filho
• Grantee: Alexandre Dias Porto Chiavegatto Filho
• Host Institution: Faculdade de Saúde Pública (FSP). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Associated researchers: Marcel de Moraes Pedroso

Abstract

Recent advances in computational capacity and the availability of large datasets have stimulated the application of predictive methods from artificial intelligence, known as machine learning. In healthcare, mortality prediction can help improve decisions about preventive care, the necessity and intensity of medical care and to set priorities for hospitalization. We will analyze the results of the UK Biobank, a longitudinal study of 502,632 individuals followed since 2006-2010. The study collected data on a large number of characteristics such as demographics, socioeconomic factors, health risk factors, medical history, anthropometric measurements and blood and urine samples, totaling more than 600 variables for each participant. Deaths are continuously collected by linkage and, by the beginning of this project, it is estimated to have surpassed 25,000 deaths. The performance of 10 to 20 machine learning algorithms will be tested to predict death in 1, 2 and 5 years by all-cause and according to groups of underlying causes, using the variables collected at baseline. After the identification of the highest-performing algorithm, its generalization capacity will be tested on the Saúde, Bem-Estar e Envelhecimento study (Projeto Temático 14/50649-6). The study will identify if it is possible to predict with high accuracy who will die, and by which underlying cause, in the next years using a large number of baseline characteristics. Funding will be spent on the data transfer fee and on machines for big data and machine learning. The project will also establish a new partnership between the Laboratório de Big Data e Análise Preditiva em Saúde (LABDAPS) of FSP/USP and the Laboratório de Ciência de Dados Aplicada à Saúde of Fiocruz-RJ. (AU)

Scientific publications

(5)

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

HELLEN GEREMIAS DOS SANTOS; CARLA FERREIRA DO NASCIMENTO; RAFAEL IZBICKI; YEDA APARECIDA DE OLIVEIRA DUARTE; ALEXANDRE DIAS PORTO CHIAVEGATTO FILHO. Machine learning para análises preditivas em saúde: exemplo de aplicação para predizer óbito em idosos de São Paulo, Brasil. Cadernos de Saúde Pública, v. 35, n. 7, 2019-07-29. (17/09369-8)

PORTO CHIAVEGATTO FILHO, ALEXANDRE DIAS; DOS SANTOS, HELLEN GEREMIAS; DO NASCIMENTO, CARLA FERREIRA; MASSA, KAIO; KAWACHI, ICHIRO. Overachieving Municipalities in Public Health: A Machine-learning Approach. EPIDEMIOLOGY, v. 29, n. 6, p. 836-840, NOV 2018. (17/09369-8)

BATISTA, ANDRE F. M.; DINIZ, CARMEN S. G.; BONILHA, ELIANA A.; KAWACHI, ICHIRO; CHIAVEGATTO FILHO, ALEXANDRE D. P.. Neonatal mortality prediction with routinely collected data: a machine learning approach. BMC PEDIATRICS, v. 21, n. 1, JUL 21 2021. (17/09369-8)

DOS SANTOS, HELLEN GEREMIAS; ZAMPIERI, FERNANDO GODINHO; NORMILIO-SILVA, KARINA; DA SILVA, GISELA TUNES; PEDROSO DE LIMA, ANTONIO CARLOS; CAVALCANTI, ALEXANDRE BIASI; PORTO CHIAVEGATTO FILHO, ALEXANDRE DIAS. Machine learning to predict 30-day quality-adjusted survival in critically ill patients with cancer. JOURNAL OF CRITICAL CARE, v. 55, p. 73-78, FEB 2020. (17/09369-8, 09/17245-0)

BARROS, VIVIAN BOSCHESI; VIANNA SCHMIDT, FERNANDA FORTTI; CHIAVEGATTO FILHO, ALEXANDRE DIAS PORTO. Mortality, survival, and causes of death in mental disorders: comprehensive prospective analyses of the UK Biobank cohort. PSYCHOLOGICAL MEDICINE, v. N/A, p. 10-pg., 2022-05-06. (17/09369-8)