Our goal is for Seisir to become a revolutionary wearable technology to combat epilepsy. The current gold standard for epilepsy monitoring are electroencephalogram (EEG) recordings, which are an outpatient procedure capturing only a short time window of time ranging from hours to days (ambulatory EEGs up to 72 hrs) or, for longer periods, requires an inpatient stay at high health care costs, an intense use of resources, and a significant burden to the family. Seisir could consistently capture data for days, weeks, months or even years without interruption. Seisir can monitor individuals routinely during sleep, when seizures are more likely to occur in many.
Sudden unexpected death in epilepsy (SUDEP) is the most prevalent cause of epilepsy related deaths, most often causing respiratory arrest during sleep. The American Epilepsy Society estimates that as many as 42,000 deaths are caused by seizures each year. SUDEP accounts for 8-17% of the total deaths in people with epilepsy according to a study by New York Methodist Hospital. A Harvard-MIT study indicates a strong correlation between skin conductance and generalized tonic-clonic seizures, which indicate a higher risk of SUDEP. Traditional seizure monitoring technologies fail to provide any effective means to assist in the prevention of epilepsy related deaths.
Our goal is to build a wearable seizure detection system that is comfortable enough for daily use and able to capture data over extended timeframes. This would enable Seisir to capture more clinically relevant data, assist in the evaluation of drug efficacy and to aid in the prevention of epilepsy related deaths.
What it does
Seisir is designed to detect life-threatening seizures, and then trigger text notifications to a caretaker for help. Although still in development, our current version of Seisir pulls data from a Microsoft Band (MB), through the app, and then uploads the information to a secure server. Once an individual's data is on the server, Seisir takes advantage of Microsoft Azure (MA) to analyze data in real-time using an advanced machine-learning algorithm. In cases where the program detects a high likelihood of seizure activity, the app contacts a caretaker or medical staff via text. Seisir has a strong potential to save lives by notifying others in emergency cases.
How we built it
Our current program utilizes a combination of technologies starting with the MB. The Android application is connected via Bluetooth with the MB to obtain various sensor data. This data is then passed through the app to MA on the cloud. Using MA's machine learning platform, our program identifies seizure activity from the sensor data originally recorded by the MB. The cloud system responds to the android app in moments when seizure activity is identified. Upon receiving a seizure alert from MA, the android app notifies emergency contacts via a traditional text message.
We intend to improve the accuracy of our machine-learning platform by creating a research database wherein healthcare providers can manually mark seizure activity as well, minimizing any false detection rate.
Since winning at Hack Harvard for the best use of Microsoft’s APIs, our team has expanded to potentially include a diverse array of students and faculty from Harvard, Yale. Seisir team members collectively possess a unique combination of specific technical and medical expertise with a strong interest in neuroscience. Our medical team is led by Jurriaan Peters, MD, an epileptologist with board certifications in clinical neurophysiology and in adult neurology with a special qualification in child neurology and clinical neurophysiology. He specializes in developing epilepsy related technologies. We hope to coordinate the distributed efforts of our team with on site clinical trials at Boston Children’s Hospital in the future with the potential of starting a digital laboratory.
How do we score?
Average Judge Score
10 judges have scored this entry. The highest possible score is 5 .2.9
We divide the average judge score with the maximum score possible and add 1 to create a multiplier.1.6
We apply the judge multiplier to the total number of crowd votes.1.6