HeartRateStress: Can Heart Rate features detect subjective stress?
GitHub Repo Final ReportStress is defined as a feeling of mental or emotional strain resulting from different circumstances. This strain can lead to several diseases including but not limited to cardiovascular disease, diabetes, asthma and other chronic illness. New fitness trackers claim they go beyond counting steps and measuring sleep. Currently, these devices can also identify stress. The goal is to help people identify the factors which trigger stress, and try to steer clear of them. The primary objective of this project is to predict subjective stress based on the heart-rate features.
LG Smart-watch Urbane: This is a commercially available smart-watch which has a heart-rate sensor and can provide an accurate measure of heart-rate. It also has a tri-axial accelerometer, a proximity sensor and a gyroscope. The device provides heart-rate values in the intervals of one second. It sends the data from the smart-watch to the back-end server through Wi-Fi.
Bio-stamp sensor: The BioStamp Sensor, a body-worn sensor so flexible and soft that it naturally conforms to the contours of the human body. BioStamp Sensors are discreet, allowing study subjects to move from lab to home, through exercise and sleep, remaining unaffected by the tightly-coupled Sensors capturing a wealth of data. It has 3 axis accelerometer that will measure acceleration forces. It also has a heart-rate sensor that will calculate the heart-rate based on the raw ECG sginal.
The objective of this project is to determine whether heart rate features would be able to detect subjective stress. This was done by inducing stress in a controlled environment and different features were identified and analyzed. Participants' response to various forms of stressors were recorded and the model was validated.
Data Analysis:We had to update our protocol to add some actvities which would make the subject be non stressed Eg: Eating, Conversation. We worked more with the Bio-stamp Java code in order to tune the values obtained. The fluctuations did reduce, but it is still under progress. We retrieved some of data logged on to the server and started pre-processing. We decided on some of the features such as mean, standard devaiation, interquartile range etc. Started on implementing a classifier in Weka
Smartwatch application: We developed an Android wear application which can detect the heart rate values and log them on to a back-end server. We tested this at different intervals and on different subjects. We built a smartphone application which could send the data values obtained from the bio-stamp onto the server. We made the application show only the heart rate values and did not include any of the other parameters. We noticed some fluctuations in the readings and are still tweaking those values. We finished the data collection module and tested its correctness and accuracy. We finished the first draft of final paper. We prepared the midterm presentation.
LG Smarthwatch: We decided to have a backup, that is use another sensor in order to check heart rate and decided that LG W150 smartwatch was a good fit. We tried experimenting with the LG W150 smart watch, but it was facing some software issues. Some of the applications were constantly crashing and that's why we need to get our Smartwatch replaced. We recieved a new smart watch and started trying to measure heart rate in the smart watch.
Bio-stamp:We met Dr. Bass and Mr. Jong also joined in via Skype. He answered some of our questions and also agreed to send us a more documented code to help us understand things further. He agreed to help us in completing our task. While working with the Bio-stamp sensor, we ran into a small issue. While testing the device, one of the electrical wires snapped and the sensor isn't working anymore. We have reached out to Ha-uk, who has agreed to meet us and fix the sensor on Monday. In the mean time we are working on smartwatch and trying to log the heart rate values on to the backend server
Project kick-off: We met with Dr. Bass and his team at the Chicago campus. We learned more about the Bio-Stamp sensor and also spoke to Mr. Ha-uk who fabricated the sensor. We also got some of the Java Code which was being used by the team to obtain heart rate values on an Android Phone.We spoke about the various parameters which can be detected by the sensor and also about how it measures those parameters.
Code Debugging:The code provided to us was difficult to understand. We spent a lot of time trying to understand what exactly was happening through each module. We met with Dr. Bass where we noted down the parts of code which did not make sense. We setup an appointment with Mr Jong, who wrote this Java code so that he could explain it better
Pradyoth is a graduate student in Computer Engineering at Nortwestern University. Involved in application development and data analysis
Santhosh is a graduate student in Information Technology at Northwestern University. Involved in android mobile application development & database integration.
Tianju Ma is a master student of Northwestern University, majoring in Computer Engineering. Worked on theoretical analysis of data processing, the design of data collection module, preparation of presentation, as well as paper polishment.
Northwestern EECS, McCormick Engineering
HABits Lab, Prof. Nabil Alshurafa
Dr. Lauren Wakschlag
Dr. Judith Moskowitz
Dr. Michael Bass
Dr. John Rogers
Ha Uk Chung
Shibo Zhang
