Machine Learning for Time Series in mHealth Applications - Semi-supervised Learning Algorithms

• Grant number: 25/13341-8
• Support Opportunities: Scholarships in Brazil - Doctorate (Direct)
• Start date: September 01, 2025
• End date: April 30, 2029
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Diego Furtado Silva
• Grantee: Giovani Decico Lucafó
• Host Institution: Instituto de Ciências Matemáticas e de Computação (ICMC). Universidade de São Paulo (USP). São Carlos , SP, Brazil
• Associated research grant: 22/03176-1 - Machine learning for time series obtained in mHealth applications, AP.PNGP.PI

Abstract

Monitoring physiological signs, vital signs, and other parameters that can be collected over time is essential in several tasks in Health, such as estimating heart rate and identifying abnormal heartbeats. However, obtaining annotated data, especially in the Health domain, can be costly. However, in some cases, obtaining labels for part of the data is possible. This scenario configures semi-supervised learning, a category of Machine Learning algorithms capable of using the information in a few annotated examples to obtain potentially better models than those induced without supervision or only with the few labeled data in the set. Semi-supervised learning can also rely on other assumptions about data annotations. This work will explore the potential of Machine Learning for time series in health applications with different annotation assumptions. Moreover, this research considers that these signals may represent different parameters, such as ECG and PPG. Therefore, they need to be handled as multimodal data. Therefore, this activity aims to adapt or create algorithms and neural architectures from semi-supervised learning for this application domain.