Parkinson's is a devastating disease that leaves sufferers unable to control their movements and with distinct tremors that severely affect their day to day lives. Currently, Parkinson's is not perfectly understood, and its cause, cure and appropriate treatments remain a mystery and dependant on individual circumstances. One issue doctors face is access to tangible data that documents the progression of the disease in their patients. The symptoms are often documented based on patient experience, with them judging how different they feel. This can cause [subjectivity] issues, particularly with patients who, on top of Parkinson's, also have illnesses such as dementia, and therefore do not have a clear idea how how their disease is progressing. By giving doctors a Hardware/Software solution, they would be able to capture real time data from their patients and would have the information necessary to give accurate diagnosis and treatment plans. Furthermore, building a database of monitoring data would allow us to track how effective different treatment options are, and what changes can be made to the current approach to treating Parkinson's. This is where parkBAND comes in.
The parkBAND is a platform that combines the Myo Armband with a web server and web app in order to acquire raw data from patients and relay it to a doctor. This data is processed such that it can help a doctor follow the patient's progression of the disease, and therefore would allow for faster, targeted treatment plans.
The Myo Armband has been hacked in order to obtain raw data from accelerometers, gyroscopes and EMG electrodes, to represent the movement of a person's arm in real time. We use this data to analyse the resting tremor experienced by Parkinson's patients in their dominant arm. The patient uses the armband and a app to take a 30 second, daily readings at home. The data from this reading is sent to the cloud, where it accumulates over a time period of a week. After a week, basic analysis is performed. 30 second readings produce over 6000 data samples per sensor. Due to the nature of a tremor and the orientation of the Myo, only certain data is further analysed, as it proved to be relevant. This data is actively compared to the first reading taken with the patient, as this reading represents the start of their monitoring and therefore allows for a good baseline for judging the progression of the disease. At the end of the week, the doctor and the patient are able to view the analytic results through a web app, which discloses whether the tremor has improved or not over the last week. The web app also provides insight into possible other signs of progression, through the analysis of patient symptom surveys. The patient continues to collect data indefinitely, and another analysis is performed once a month. This analysis looks at the overall magnitude of change in the tremor values (increased or decreased) over the last month, and since the start of the treatment/monitoring. This gives scope as to the patient's progress speed and overall progress. It re-evaluates the patient's disease stage, which is a number from 1 to 5 characterizing the symptoms and severity of Parkinson's experienced. Furthermore, the app goes on to suggest possible treatment changes given the updated patient condition.
A local app, loaded on a user's computer or tablet, uses a C++ and Java in order to acquire raw sensor data over the 30 second capture period. It then finds the absolute value of the signals, and their respective means, which are then sent to a PHP which is connected to Microsoft's Azure. A MySQL database establishes connection between Azure and a webapp, allowing for the webapp to access the data captured by the Myo. Since we are most concerned with the changes seen in the data collected, we calculate the percent difference between all data points collected and the first, initial reading, which is captured the day when the patient starts treatment / monitoring (baseline). These percent differences are then analysed against thresholds, and along with patient survey data, are used to determine the stage and possible treatment implementations and changes.
We faced many challenges including issues with setting up a server, requiring us to switch from AWS, to Google Cloud and finally to an PHP on Azure. We are very close to establishing proper connection with Google Cloud, which would allow us to make use of their server and APIs. Other challenges included extracting raw data from the Myo, since the Firmware is not open sourced. We utilised existing Myo applications and integrated them into our own Java program in order to extract the data we needed.
We are proud to have made a functional product that demonstrates our general idea; no where is it as clean and polished or efficient as we would like it to be, but it gave us great insight into how wearables can be used in healthcare, particularly with the elderly population, and furthermore at home, with a simple, easy to user patient app. We all learned valuable skills and experienced new platforms and softwares. We all came from different backgrounds, and most of us had worked with something that the others hadn't. This allowed us to each take a part of the project as our own, and provided everyone else on the team with a chance to see how that technology could be used in the real world.
parkBAND can be expanded to a much broader medical field, and can become a daily worn system. In terms of Parkinson's, the platform can be built to incorporate monitoring of freezing of gait and tremors in the legs that can cause dangerous falls. Other sensors could monitor parameters such as heart rate, which can expose other symptoms such as anxiety and depression. The platform can also be extended to diseases such as diabetes, allowing for the users to monitor insulin injection forces and diabetic tremors. The platform can also incorporate more sensors to detect heart rate, blood pressure and blood glucose, which can identify insulin injection efficiency. In general, the Myo can be used as a way to send alerts, all while collecting data that can be extremely beneficial to the healthcare industry. parkBAND started as an initiative to help a vulnerable population receive the help they need, however, it can easily evolve to a industry-wide platform, allowing for a dataplay that could change how we treat the world's most common medical problems.
Log in or sign up for Devpost to join the conversation.