Towards Predictive Medicine — On Remote Monitoring, Privacy and Scientific Bias

Daniel Aranki

EECS Department
University of California, Berkeley
Technical Report No. UCB/EECS-2017-145
August 11, 2017

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2017/EECS-2017-145.pdf

The current healthcare model in the United States of America (US) is reactive in nature. That is, individuals usually seek medical attention after symptoms manifest. In 2015, the total cost of healthcare in the US was $3.2 trillion (17.8% of US Gross Domestic Product), which amounts to $9,990 per capita. In the same year, the 30-days all-condition rate of unplanned rehospitalizations in patients in the Medicare fee-for-service program was around 17.9%; and between October 1, 2003 and December 31, 2003, 3.5% of patients in the same program died within 30 days of initial discharge.

Alternatively, a healthcare model that utilizes medical intervention based on personalized predictions of the patient's clinical status and possible deterioration could potentially decrease costs, unplanned rehospitalizations and mortality rates. This model also has the potential to improve the overall quality of care. We refer to this model as the predictive healthcare model.

In this dissertation, we examine three outstanding challenges towards fully realizing the predictive healthcare model as the prevalent care model. Namely, i) we investigate means to streamline the costly longitudinal epidemiological studies using remote mobile monitoring and introduce the Berkeley Telemonitoring project; ii) we investigate the privacy challenge that is particular to the remote monitoring model and introduce the Private Disclosure of Information (PDI) semantic privacy model; and iii) we investigate the problem of publication bias in empirical sciences (including biomedicine) that hinders the credibility of empirical scientific findings and introduce a statistical test that detects bias in a sample of scientific publications which utilize the Student t-test.

Advisor: Ruzena Bajcsy


BibTeX citation:

@phdthesis{Aranki:EECS-2017-145,
    Author = {Aranki, Daniel},
    Title = {Towards Predictive Medicine — On Remote Monitoring, Privacy and Scientific Bias},
    School = {EECS Department, University of California, Berkeley},
    Year = {2017},
    Month = {Aug},
    URL = {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2017/EECS-2017-145.html},
    Number = {UCB/EECS-2017-145},
    Abstract = {The current healthcare model in the United States of America (US) is reactive in nature. That is, individuals usually seek medical attention after symptoms manifest. In 2015, the total cost of healthcare in the US was $3.2 trillion (17.8% of US Gross Domestic Product), which amounts to $9,990 per capita. In the same year, the 30-days all-condition rate of unplanned rehospitalizations in patients in the Medicare fee-for-service program was around 17.9%; and between October 1, 2003 and December 31, 2003, 3.5% of patients in the same program died within 30 days of initial discharge.

Alternatively, a healthcare model that utilizes medical intervention based on personalized predictions of the patient's clinical status and possible deterioration could potentially decrease costs, unplanned rehospitalizations and mortality rates. This model also has the potential to improve the overall quality of care. We refer to this model as the predictive healthcare model.

In this dissertation, we examine three outstanding challenges towards fully realizing the predictive healthcare model as the prevalent care model. Namely, i) we investigate means to streamline the costly longitudinal epidemiological studies using remote mobile monitoring and introduce the Berkeley Telemonitoring project; ii) we investigate the privacy challenge that is particular to the remote monitoring model and introduce the Private Disclosure of Information (PDI) semantic privacy model; and iii) we investigate the problem of publication bias in empirical sciences (including biomedicine) that hinders the credibility of empirical scientific findings and introduce a statistical test that detects bias in a sample of scientific publications which utilize the Student t-test.}
}

EndNote citation:

%0 Thesis
%A Aranki, Daniel
%T Towards Predictive Medicine — On Remote Monitoring, Privacy and Scientific Bias
%I EECS Department, University of California, Berkeley
%D 2017
%8 August 11
%@ UCB/EECS-2017-145
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2017/EECS-2017-145.html
%F Aranki:EECS-2017-145