The aim of the Sensible project is to give a good model of integrating helathcare sensors into a global architecture, which enables the analyzis of the data with a big data system. BI tools are used to find correlations in the data of a single or multiple (anonymized) users.
For more information please contact the project lead:
Dr. Bertalan Forstner
Bertalan.Forstner[AT]aut.bme.hu
Introduction
We have seen the emerging popularity of a phenomenon called „quantified self”. Followers of this movement regard every aspect of their life as input data, which they record and store in order to improve daily functioning. The history of self-tracking using wearable sensors in combination with wearable computing and wireless communication already exists for many years, and also appeared, in the form of sousveillance back in the 1970s [1]. Today, healthcare sensors and different kinds of sport trackers become cheaper and affordable, and even smart devices have sensors capable of performing health related measurements.
The average user collecting self-tracking data is not medical expert, it is difficult for her to interpret her medical results or similar self-monitoring data in depth. She is not aware of the importance of the individual values or the meaning of deviance from normal intervals, nor can she combine different measured values to infer her health status. What such users can do is paying for the doctor’s time or look up some uncontrolled source on the Internet to learn the meaning of these data.
Motivated by increasing healthcare costs, using medical grade sensors is also regarded as a way of cost-effectively observing the required biological signals of a patient. [2] This phenomenon transforms the healthcare industry in a form where remote experts decide, for example, on the necessity of a surgical intervention for a given patient, based on sensor data collected for days. Similar to knowledge engineering, it is possible to run learning algorithms on voluntary provided sensor data of thousands of users to infer hidden correlations. Automated processes can even warn the user if some suspicious results make it legitimate to visit a general practitioner or a specialist. [3] The experts can harness the availability of historical data during analysis.
A shortcoming of the current state-of-the-art systems for the described challenge is that they are closed proprietary solutions. Sensor data from one system cannot be used with the system of another player on the market, as the data or the provided service are holding market value. There are couple of manufacturers providing application programming interface for their sensors or trackers, however, most of them cannot be integrated into third party software. One excuse for that is the sensibility of personal or medical data, as their privacy cannot be guaranteed if they are offered for third parties via uncontrolled interfaces.
Combining the SensorHUB framework with those medical sensors, we are concentrating on a method which enables the collection of various kinds of health data from different sensor sources, and then utilizing the framework to infer the health status or find correlations and predictions.
1. Fig. Using smartphone as a display for simple wired sensors
As explained earlier, a smart phone can be used as a gateway and controller device for the measurements. As a result, there is no need to learn the usage of multiple applications, nor is it necessary for the user to authenticate itself with different systems. Information about an ongoing measurement can be shown on the mobile device of the user, together with the final result and analysis at the end of the process. Users can utilize their own sensors or trackers for this process, but it is also possible to share sensors among many users. The data analysis and storage is done on a dedicated server. Due to the massive amount of information steps to insure scalability had to be done, therefore, at the server side SensorHUB was used.
Special care has to be taken to the security of the personal data. Our approach also requires a complex authentication system which encrypts medical data, and authenticates the measurement device and measurement process at the same time.
The Sensible project
With the considerations above, we designed and implemented our framework on the top of SensorHUB, and called it Sensible. We selected a set of sensor types to integrate to the system, both wired and wireless. Wireless sensors can harness the connectivity of the smartphone device of the users. In case of the wired sensors, there should be an intermediary agent that can receive the signals from those sensors, and load the data into the SensorHUB. Due to its flexibility, price and appropriate performance, we selected Raspberry Pi devices for this task, running our software and the drivers of those sensors.
2. Fig. Sensor units currently connected to the Raspberry Pi agent.
A challenge here is to authenticate the patient who is initiating the measurement, and then authorize the intermediary agent to start the process and upload the data into SensorHUB on behalf of the given user. The following figure represents the simplest way, when the mobile device connects to the same subnet as the measurement agents. In that case, the mobile device has to send the application server the ID of the intermediary device he is about to use. In order to make this process simple, intermediary agents can share their ID via NFC technology. That way, NFC-enabled smartphones can authorize these devices with „touching” their phone to the casings of the Raspberry Pi.
3. Fig. Connecting the smartphone with the agents on the same subnet
In case the parties are on different subnets (for example, the smartphone is connected to a mobile network), we maintain a registry service where the measurement units can register their ID with the actual IP address, and there is no direct connection between the units.
Wireless sensors with public API connect directly to the mobile device, which is then responsible to upload the data of the authenticated user to the SensorHUB.
In case of the first version of Sensible, we are looking for correlations between different uploaded measurement parameters and illnesses identified by doctors. Our algorithm is designed after the model of Pearson Correlation Coefficient [4][5]. It describes a linear connection between two value pairs with a number between -1 and 1. Values close to zero mean statistically not significant results, while those far from zero imply strong linear correlation.
We believe that in the near future the sensors built on advanced technology will play an important role in efficient healthcare services and also in early recognition of illnesses. Our research, the Sensible, is part of the efforts made to achieve this goal.
[1] "Invasion of the body hackers". Financial Times. 2011-06-10.
[2] A. Pantelopoulos, N.G. Bourbakis, “A survey on wearable sensor-based systems for health monitoring and prognosis,” IEEE Trans. Systems, Man, and Cybernetics, Part C: Applications and Reviews, vol. 40, no. 1, pp. 1-12, 2010
[3] Clifton, L.; Clifton, D.A.; Pimentel, M.A.F.; Watkinson, P.J.; Tarassenko, L. "Predictive Monitoring of Mobile Patients by Combining Clinical Observations With Data From Wearable Sensors", Biomedical and Health Informatics, IEEE Journal of, On page(s): 722 - 730 Volume: 18, Issue: 3, May 2014
[4] Lawrence I-Kuei Lin, A Concordance Correlation Coefficient to Evaluate Reproducibility, 45th ed.: International Biometric Society, 1989.
[5] Ziad S. Saad, A new method for improving functional-to-structural MRI alignment using local Pearson correlation. Houston, Texas, United States, 2008.
The aim of the Sensible project is to give a good model of integrating helathcare sensors into a global architecture, which enables the analyzis of the data with a big data system. BI tools are used to find correlations in the data of a single or multiple (anonymized) users.
For more information please contact the project lead:
Dr. Bertalan Forstner
Bertalan.Forstner[AT]aut.bme.hu
Introduction
We have seen the emerging popularity of a phenomenon called „quantified self”. Followers of this movement regard every aspect of their life as input data, which they record and store in order to improve daily functioning. The history of self-tracking using wearable sensors in combination with wearable computing and wireless communication already exists for many years, and also appeared, in the form of sousveillance back in the 1970s [1]. Today, healthcare sensors and different kinds of sport trackers become cheaper and affordable, and even smart devices have sensors capable of performing health related measurements.
The average user collecting self-tracking data is not medical expert, it is difficult for her to interpret her medical results or similar self-monitoring data in depth. She is not aware of the importance of the individual values or the meaning of deviance from normal intervals, nor can she combine different measured values to infer her health status. What such users can do is paying for the doctor’s time or look up some uncontrolled source on the Internet to learn the meaning of these data.
Motivated by increasing healthcare costs, using medical grade sensors is also regarded as a way of cost-effectively observing the required biological signals of a patient. [2] This phenomenon transforms the healthcare industry in a form where remote experts decide, for example, on the necessity of a surgical intervention for a given patient, based on sensor data collected for days. Similar to knowledge engineering, it is possible to run learning algorithms on voluntary provided sensor data of thousands of users to infer hidden correlations. Automated processes can even warn the user if some suspicious results make it legitimate to visit a general practitioner or a specialist. [3] The experts can harness the availability of historical data during analysis.
A shortcoming of the current state-of-the-art systems for the described challenge is that they are closed proprietary solutions. Sensor data from one system cannot be used with the system of another player on the market, as the data or the provided service are holding market value. There are couple of manufacturers providing application programming interface for their sensors or trackers, however, most of them cannot be integrated into third party software. One excuse for that is the sensibility of personal or medical data, as their privacy cannot be guaranteed if they are offered for third parties via uncontrolled interfaces.
Combining the SensorHUB framework with those medical sensors, we are concentrating on a method which enables the collection of various kinds of health data from different sensor sources, and then utilizing the framework to infer the health status or find correlations and predictions.
1. Fig. Using smartphone as a display for simple wired sensors
As explained earlier, a smart phone can be used as a gateway and controller device for the measurements. As a result, there is no need to learn the usage of multiple applications, nor is it necessary for the user to authenticate itself with different systems. Information about an ongoing measurement can be shown on the mobile device of the user, together with the final result and analysis at the end of the process. Users can utilize their own sensors or trackers for this process, but it is also possible to share sensors among many users. The data analysis and storage is done on a dedicated server. Due to the massive amount of information steps to insure scalability had to be done, therefore, at the server side SensorHUB was used.
Special care has to be taken to the security of the personal data. Our approach also requires a complex authentication system which encrypts medical data, and authenticates the measurement device and measurement process at the same time.
The Sensible project
With the considerations above, we designed and implemented our framework on the top of SensorHUB, and called it Sensible. We selected a set of sensor types to integrate to the system, both wired and wireless. Wireless sensors can harness the connectivity of the smartphone device of the users. In case of the wired sensors, there should be an intermediary agent that can receive the signals from those sensors, and load the data into the SensorHUB. Due to its flexibility, price and appropriate performance, we selected Raspberry Pi devices for this task, running our software and the drivers of those sensors.
2. Fig. Sensor units currently connected to the Raspberry Pi agent.
A challenge here is to authenticate the patient who is initiating the measurement, and then authorize the intermediary agent to start the process and upload the data into SensorHUB on behalf of the given user. The following figure represents the simplest way, when the mobile device connects to the same subnet as the measurement agents. In that case, the mobile device has to send the application server the ID of the intermediary device he is about to use. In order to make this process simple, intermediary agents can share their ID via NFC technology. That way, NFC-enabled smartphones can authorize these devices with „touching” their phone to the casings of the Raspberry Pi.
3. Fig. Connecting the smartphone with the agents on the same subnet
In case the parties are on different subnets (for example, the smartphone is connected to a mobile network), we maintain a registry service where the measurement units can register their ID with the actual IP address, and there is no direct connection between the units.
Wireless sensors with public API connect directly to the mobile device, which is then responsible to upload the data of the authenticated user to the SensorHUB.
In case of the first version of Sensible, we are looking for correlations between different uploaded measurement parameters and illnesses identified by doctors. Our algorithm is designed after the model of Pearson Correlation Coefficient [4][5]. It describes a linear connection between two value pairs with a number between -1 and 1. Values close to zero mean statistically not significant results, while those far from zero imply strong linear correlation.
We believe that in the near future the sensors built on advanced technology will play an important role in efficient healthcare services and also in early recognition of illnesses. Our research, the Sensible, is part of the efforts made to achieve this goal.
[1] "Invasion of the body hackers". Financial Times. 2011-06-10.
[2] A. Pantelopoulos, N.G. Bourbakis, “A survey on wearable sensor-based systems for health monitoring and prognosis,” IEEE Trans. Systems, Man, and Cybernetics, Part C: Applications and Reviews, vol. 40, no. 1, pp. 1-12, 2010
[3] Clifton, L.; Clifton, D.A.; Pimentel, M.A.F.; Watkinson, P.J.; Tarassenko, L. "Predictive Monitoring of Mobile Patients by Combining Clinical Observations With Data From Wearable Sensors", Biomedical and Health Informatics, IEEE Journal of, On page(s): 722 - 730 Volume: 18, Issue: 3, May 2014
[4] Lawrence I-Kuei Lin, A Concordance Correlation Coefficient to Evaluate Reproducibility, 45th ed.: International Biometric Society, 1989.
[5] Ziad S. Saad, A new method for improving functional-to-structural MRI alignment using local Pearson correlation. Houston, Texas, United States, 2008.
The aim of the Sensible project is to give a good model of integrating helathcare sensors into a global architecture, which enables the analyzis of the data with a big data system. BI tools are used to find correlations in the data of a single or multiple (anonymized) users.
For more information please contact the project lead:
Dr. Bertalan Forstner
Bertalan.Forstner[AT]aut.bme.hu
Introduction
We have seen the emerging popularity of a phenomenon called „quantified self”. Followers of this movement regard every aspect of their life as input data, which they record and store in order to improve daily functioning. The history of self-tracking using wearable sensors in combination with wearable computing and wireless communication already exists for many years, and also appeared, in the form of sousveillance back in the 1970s [1]. Today, healthcare sensors and different kinds of sport trackers become cheaper and affordable, and even smart devices have sensors capable of performing health related measurements.
The average user collecting self-tracking data is not medical expert, it is difficult for her to interpret her medical results or similar self-monitoring data in depth. She is not aware of the importance of the individual values or the meaning of deviance from normal intervals, nor can she combine different measured values to infer her health status. What such users can do is paying for the doctor’s time or look up some uncontrolled source on the Internet to learn the meaning of these data.
Motivated by increasing healthcare costs, using medical grade sensors is also regarded as a way of cost-effectively observing the required biological signals of a patient. [2] This phenomenon transforms the healthcare industry in a form where remote experts decide, for example, on the necessity of a surgical intervention for a given patient, based on sensor data collected for days. Similar to knowledge engineering, it is possible to run learning algorithms on voluntary provided sensor data of thousands of users to infer hidden correlations. Automated processes can even warn the user if some suspicious results make it legitimate to visit a general practitioner or a specialist. [3] The experts can harness the availability of historical data during analysis.
A shortcoming of the current state-of-the-art systems for the described challenge is that they are closed proprietary solutions. Sensor data from one system cannot be used with the system of another player on the market, as the data or the provided service are holding market value. There are couple of manufacturers providing application programming interface for their sensors or trackers, however, most of them cannot be integrated into third party software. One excuse for that is the sensibility of personal or medical data, as their privacy cannot be guaranteed if they are offered for third parties via uncontrolled interfaces.
Combining the SensorHUB framework with those medical sensors, we are concentrating on a method which enables the collection of various kinds of health data from different sensor sources, and then utilizing the framework to infer the health status or find correlations and predictions.
1. Fig. Using smartphone as a display for simple wired sensors
As explained earlier, a smart phone can be used as a gateway and controller device for the measurements. As a result, there is no need to learn the usage of multiple applications, nor is it necessary for the user to authenticate itself with different systems. Information about an ongoing measurement can be shown on the mobile device of the user, together with the final result and analysis at the end of the process. Users can utilize their own sensors or trackers for this process, but it is also possible to share sensors among many users. The data analysis and storage is done on a dedicated server. Due to the massive amount of information steps to insure scalability had to be done, therefore, at the server side SensorHUB was used.
Special care has to be taken to the security of the personal data. Our approach also requires a complex authentication system which encrypts medical data, and authenticates the measurement device and measurement process at the same time.
The Sensible project
With the considerations above, we designed and implemented our framework on the top of SensorHUB, and called it Sensible. We selected a set of sensor types to integrate to the system, both wired and wireless. Wireless sensors can harness the connectivity of the smartphone device of the users. In case of the wired sensors, there should be an intermediary agent that can receive the signals from those sensors, and load the data into the SensorHUB. Due to its flexibility, price and appropriate performance, we selected Raspberry Pi devices for this task, running our software and the drivers of those sensors.
2. Fig. Sensor units currently connected to the Raspberry Pi agent.
A challenge here is to authenticate the patient who is initiating the measurement, and then authorize the intermediary agent to start the process and upload the data into SensorHUB on behalf of the given user. The following figure represents the simplest way, when the mobile device connects to the same subnet as the measurement agents. In that case, the mobile device has to send the application server the ID of the intermediary device he is about to use. In order to make this process simple, intermediary agents can share their ID via NFC technology. That way, NFC-enabled smartphones can authorize these devices with „touching” their phone to the casings of the Raspberry Pi.
3. Fig. Connecting the smartphone with the agents on the same subnet
In case the parties are on different subnets (for example, the smartphone is connected to a mobile network), we maintain a registry service where the measurement units can register their ID with the actual IP address, and there is no direct connection between the units.
Wireless sensors with public API connect directly to the mobile device, which is then responsible to upload the data of the authenticated user to the SensorHUB.
In case of the first version of Sensible, we are looking for correlations between different uploaded measurement parameters and illnesses identified by doctors. Our algorithm is designed after the model of Pearson Correlation Coefficient [4][5]. It describes a linear connection between two value pairs with a number between -1 and 1. Values close to zero mean statistically not significant results, while those far from zero imply strong linear correlation.
We believe that in the near future the sensors built on advanced technology will play an important role in efficient healthcare services and also in early recognition of illnesses. Our research, the Sensible, is part of the efforts made to achieve this goal.
[1] "Invasion of the body hackers". Financial Times. 2011-06-10.
[2] A. Pantelopoulos, N.G. Bourbakis, “A survey on wearable sensor-based systems for health monitoring and prognosis,” IEEE Trans. Systems, Man, and Cybernetics, Part C: Applications and Reviews, vol. 40, no. 1, pp. 1-12, 2010
[3] Clifton, L.; Clifton, D.A.; Pimentel, M.A.F.; Watkinson, P.J.; Tarassenko, L. "Predictive Monitoring of Mobile Patients by Combining Clinical Observations With Data From Wearable Sensors", Biomedical and Health Informatics, IEEE Journal of, On page(s): 722 - 730 Volume: 18, Issue: 3, May 2014
[4] Lawrence I-Kuei Lin, A Concordance Correlation Coefficient to Evaluate Reproducibility, 45th ed.: International Biometric Society, 1989.
[5] Ziad S. Saad, A new method for improving functional-to-structural MRI alignment using local Pearson correlation. Houston, Texas, United States, 2008.