Predictive modeling of health status using motion analysis from mobile phones

It is unknown what physiological functions can be monitored at clinical quality with a normal smartphone, which is ubiquitous. There are standard measures like walk speed, pulmonary function and oxygen saturation variation for health status of cardiopulmonary patients. The dissertation is to summarize my studies of using sensor data from regular smartphones to accurately measure walking patterns, in order to monitoring health status for cardiopulmonary patients. Fifty five pulmonary patients were participated in the study. The sensor data for their walk test and free walk are collected and stored by a novel designed Android smartphone application. Different machine learning techniques are applied and compared to predict gait speed, pulmonary function and oxygen saturation. The result shows that walking patterns are highly correlated with health status. The trained models can predict health status accurately for each patient. Initial testing indicates the same high accuracy as with active monitors, for patients in hospitals during walk tests. The ultimate goal is that patients can simply carry their phones during everyday living, while models support automatic prediction of pulmonary function for health monitoring.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2017

Author dc:creator

• Cheng, Qian

Contributors dc:contributor

• Schatz, Bruce R.
• Han, Jiawei
• Smaragdis, Paris
• Konig, Christian

• Mobile health
• Smartphone application
• Medical information retrieval
• Chronic disease

Statement dc:rights

• Copyright 2017 Qian Cheng

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/97278

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/97278