Presentation
Mobile Health Platform for Individual and Population-Level Surveillance
SessionDH10 - Poster
Event Type
Poster Presentation
TimeThursday, April 152:27pm - 2:28pm EDT
LocationDigital Health
DescriptionIntroduction
Public health surveillance is the collection, analysis and dissemination of data to improve population health. These data are the most important source of information to support decision making and interventions by public health agencies. One of the main sources of data are surveys, such as in-person questionnaires and interviews. However, traditional survey methods have significant limitations related to self-reported data such as social and recall bias, loss due to follow-up, delays between collection and reporting, and costs/logistics. These limitations increase the burden on the user and can lead to incorrect or inaccurate data being collected.
An alternative to data collection in surveys is the use of mobile, wearable, and Internet of Things (IoT) technology, such as smartphones, smartwatches, and wireless scales, as additional survey and assessment tools. Notably, smart technologies have sensors that provide zero-effort monitoring of vital signs, environmental variables, and behavioural metrics, such as: heart rate, temperature, movements in the house, among others. Sensor data are also continuously measured, providing richer and more representative information.
Currently, to reduce the spread of COVID-19, populations around the globe are being asked to self-isolate at home, making clinic measurements and in-person interviews extremely difficult. In addition, social distancing for large periods of time can have adverse effects in the mental and physical health of populations. Connected devices that can safely, continuously and effortlessly monitor the health of individuals can be of great help during the pandemic and highlight the pressing need of using personal devices for public health surveillance.
Apple Health is one of the most popular sources of health data from sensors, collecting information from devices such as smartphones, smartwatches and wireless scales, that are connected to Apple operating systems. However, little focus has been given to the use of all this personal smart technology and data to support public health, despite its popularity.
As an example of the benefits of the use of personal devices to data collection by public health agencies, the Canadian Health Measures Survey (CHMS) is a major survey comprised of: (i) an hour-long interview in the respondents’ home; (ii) a visit to a temporary clinic to collect physical measures; and (iii) use of a fitness tracker for a week. Most of the measures in this survey, such as body composition or heart rate, can be collected using smart technologies, and the aforementioned platform can reduce social and recall biases. Incorporating smart technologies in survey design will minimize time and financial burdens of clinicians and interviewers, while data can be reported in real-time. By leveraging the data already collected from personal devices for long periods of time, studies could minimize follow-up losses by providing automated data collection while ensuring the data are more representative than those obtained from the fitness tracker.
Objective
Currently, there is a gap in the field of public health surveillance: smart technologies can improve data collection accuracy and minimize limitations in traditional data collection methods; however, they are not currently being used in this context. Public health agencies are not able to access large volumes of diverse and real-time data, collected by personal devices, which would allow them to conduct more complex analyses and interventions in their mission of improving the health of populations.
The overall goal of this project is to create of a mobile platform that will collect and store health data from devices connected to Apple Health continuously and in real-time, with zero effort to users and respondents (unlike traditional surveys). The diverse and real-time data will be extremely relevant to public health officials in supporting their decision making and health policies.
Methods
An iPhone prototype has been developed, collecting weight, steps, heart rate, blood pressure and sleep data from Apple Health and associated connected devices such as iPhones, Apple Watches and wireless scales. These variables were chosen for the proof of concept as they are usually collected in Canadian public health surveys. For the final version, users will be able to add or remove variables, allowing flexibility for researchers to collect specific data types for each survey or study and improving overall user experience.
Data Validation To validate the platform, stress will be used as a use case: we will conduct a study to collect stress-related variables from working age individuals in real-life scenarios with the platform. These data will be used to train Machine/Deep Learning algorithms to predict the stress levels of each individual, demonstrating the effectiveness of the platform. Stress was chosen as it is a significant issue in Canada. Also, stress is collected in research through self-report questionnaires, despite associations between stress and vital/behavioural metrics that can be measured using smart devices (e.g., sleep, physical activity, heart rate). Therefore, continuous collection of objective measures on stress is an effective use case for the platform. We will be looking at the application of the Random Forest, Support Vector Machine, and Deep Learning algorithms.
Privacy Since this work proposes to collect personal health data, it is necessary to consider the privacy of these data. In Canada, the Protection and Electronic Documents Act (PIPEDA) regulates the collection, use and disclosure of personally identifiable information (PII) for private sector organizations involved in a commercial activity. This applies to all types of PII including health data. Therefore, this work will also be looking into maintaining user privacy and obtaining user consent for data collection using mobiles and wearables. We will explore how to anonymize and maintain user privacy for the collected data, while making sure that the user has a great experience with the platform through the use of consent forms.
Platform Evaluation To evaluate the platform, we will look into several attributes that are typically used to evaluate an informatics-based surveillance system. This will allow us to make sure the platform follows best practices and guidelines for human factors deployment: (i) simplicity of the system; (ii) fleixibility, meaning the system should be able to adapt with little effort; (iii) data quality, meaning the collected data must be valid; (iv) acceptability, meaning that users will be willing and eager to use the system; (v) timeliness, meaning the system must be fast; (vi) stability, related to the availability and reliability of the system; (vii) information quality, which refers to completeness and consistency of the collected data; (viii) system quality, referring to the usability, minimal error, response time and functionality of the system, among others; (ix) user experience: refers to the assurancy and empathy of the system.
Conclusion
Ultimately, with the proposed platform, individuals will have a complete and more accurate picture of their health through zero-effort methods, including quantitative and qualitative metrics. With more robust, diverse and accurate data to support decision making, public health agencies will have a new tool to help in their mission of improving population health and saving lives.
Public health surveillance is the collection, analysis and dissemination of data to improve population health. These data are the most important source of information to support decision making and interventions by public health agencies. One of the main sources of data are surveys, such as in-person questionnaires and interviews. However, traditional survey methods have significant limitations related to self-reported data such as social and recall bias, loss due to follow-up, delays between collection and reporting, and costs/logistics. These limitations increase the burden on the user and can lead to incorrect or inaccurate data being collected.
An alternative to data collection in surveys is the use of mobile, wearable, and Internet of Things (IoT) technology, such as smartphones, smartwatches, and wireless scales, as additional survey and assessment tools. Notably, smart technologies have sensors that provide zero-effort monitoring of vital signs, environmental variables, and behavioural metrics, such as: heart rate, temperature, movements in the house, among others. Sensor data are also continuously measured, providing richer and more representative information.
Currently, to reduce the spread of COVID-19, populations around the globe are being asked to self-isolate at home, making clinic measurements and in-person interviews extremely difficult. In addition, social distancing for large periods of time can have adverse effects in the mental and physical health of populations. Connected devices that can safely, continuously and effortlessly monitor the health of individuals can be of great help during the pandemic and highlight the pressing need of using personal devices for public health surveillance.
Apple Health is one of the most popular sources of health data from sensors, collecting information from devices such as smartphones, smartwatches and wireless scales, that are connected to Apple operating systems. However, little focus has been given to the use of all this personal smart technology and data to support public health, despite its popularity.
As an example of the benefits of the use of personal devices to data collection by public health agencies, the Canadian Health Measures Survey (CHMS) is a major survey comprised of: (i) an hour-long interview in the respondents’ home; (ii) a visit to a temporary clinic to collect physical measures; and (iii) use of a fitness tracker for a week. Most of the measures in this survey, such as body composition or heart rate, can be collected using smart technologies, and the aforementioned platform can reduce social and recall biases. Incorporating smart technologies in survey design will minimize time and financial burdens of clinicians and interviewers, while data can be reported in real-time. By leveraging the data already collected from personal devices for long periods of time, studies could minimize follow-up losses by providing automated data collection while ensuring the data are more representative than those obtained from the fitness tracker.
Objective
Currently, there is a gap in the field of public health surveillance: smart technologies can improve data collection accuracy and minimize limitations in traditional data collection methods; however, they are not currently being used in this context. Public health agencies are not able to access large volumes of diverse and real-time data, collected by personal devices, which would allow them to conduct more complex analyses and interventions in their mission of improving the health of populations.
The overall goal of this project is to create of a mobile platform that will collect and store health data from devices connected to Apple Health continuously and in real-time, with zero effort to users and respondents (unlike traditional surveys). The diverse and real-time data will be extremely relevant to public health officials in supporting their decision making and health policies.
Methods
An iPhone prototype has been developed, collecting weight, steps, heart rate, blood pressure and sleep data from Apple Health and associated connected devices such as iPhones, Apple Watches and wireless scales. These variables were chosen for the proof of concept as they are usually collected in Canadian public health surveys. For the final version, users will be able to add or remove variables, allowing flexibility for researchers to collect specific data types for each survey or study and improving overall user experience.
Data Validation To validate the platform, stress will be used as a use case: we will conduct a study to collect stress-related variables from working age individuals in real-life scenarios with the platform. These data will be used to train Machine/Deep Learning algorithms to predict the stress levels of each individual, demonstrating the effectiveness of the platform. Stress was chosen as it is a significant issue in Canada. Also, stress is collected in research through self-report questionnaires, despite associations between stress and vital/behavioural metrics that can be measured using smart devices (e.g., sleep, physical activity, heart rate). Therefore, continuous collection of objective measures on stress is an effective use case for the platform. We will be looking at the application of the Random Forest, Support Vector Machine, and Deep Learning algorithms.
Privacy Since this work proposes to collect personal health data, it is necessary to consider the privacy of these data. In Canada, the Protection and Electronic Documents Act (PIPEDA) regulates the collection, use and disclosure of personally identifiable information (PII) for private sector organizations involved in a commercial activity. This applies to all types of PII including health data. Therefore, this work will also be looking into maintaining user privacy and obtaining user consent for data collection using mobiles and wearables. We will explore how to anonymize and maintain user privacy for the collected data, while making sure that the user has a great experience with the platform through the use of consent forms.
Platform Evaluation To evaluate the platform, we will look into several attributes that are typically used to evaluate an informatics-based surveillance system. This will allow us to make sure the platform follows best practices and guidelines for human factors deployment: (i) simplicity of the system; (ii) fleixibility, meaning the system should be able to adapt with little effort; (iii) data quality, meaning the collected data must be valid; (iv) acceptability, meaning that users will be willing and eager to use the system; (v) timeliness, meaning the system must be fast; (vi) stability, related to the availability and reliability of the system; (vii) information quality, which refers to completeness and consistency of the collected data; (viii) system quality, referring to the usability, minimal error, response time and functionality of the system, among others; (ix) user experience: refers to the assurancy and empathy of the system.
Conclusion
Ultimately, with the proposed platform, individuals will have a complete and more accurate picture of their health through zero-effort methods, including quantitative and qualitative metrics. With more robust, diverse and accurate data to support decision making, public health agencies will have a new tool to help in their mission of improving population health and saving lives.
Authors
